diff --git "a/Generation/meta/val/llava_format/full_single_multi_contact_obj_traj_qa.json" "b/Generation/meta/val/llava_format/full_single_multi_contact_obj_traj_qa.json" new file mode 100644--- /dev/null +++ "b/Generation/meta/val/llava_format/full_single_multi_contact_obj_traj_qa.json" @@ -0,0 +1,42002 @@ +[ + { + "id": "RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_104422070044#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the block on the other blocks, pick up the block on the plane, transfer the block from the plane to the other blocks, place the block on the other blocks.'\nYou start from [0.57, 0.36]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.36], [0.53, 0.31], [0.5, 0.36], [0.51, 0.41], [0.51, 0.3], [0.53, 0.24], [0.56, 0.28], [0.57, 0.32], [0.56, 0.29], [0.59, 0.26]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_104422070044#648.jpg", + "gt": "[[363, 131], [340, 111], [322, 131], [328, 148], [325, 107], [341, 88], [361, 101], [362, 114], [360, 104], [375, 94]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "35690_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the black and yellow bottle to the left, move the water bottle to the right, then move the black and yellow bottle closer to the water bottle', and now you are performing subtask 'push the black and yellow bottle next to the water bottle.'\nStarting at [0.55, 0.46], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.46], [0.49, 0.51], [0.44, 0.63], [0.45, 0.65], [0.54, 0.67], [0.53, 0.64], [0.55, 0.63], [0.52, 0.62], [0.58, 0.64], [0.57, 0.48]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/35690_exterior_image_1_left#456.jpg", + "gt": "[[176, 82], [157, 92], [141, 113], [144, 117], [174, 120], [169, 116], [175, 114], [168, 112], [187, 115], [181, 86]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "38414_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the plush toy and put it on the plate, then move the tap to the right', and now you are performing subtask 'twist the faucet counterclockwise.'\nStarting at [0.5, 0.34], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.34], [0.53, 0.37], [0.55, 0.38], [0.54, 0.43], [0.56, 0.57], [0.55, 0.63], [0.54, 0.63], [0.48, 0.58], [0.42, 0.54], [0.44, 0.56]]}" + } + ], + "images": "38414_exterior_image_1_left#125", + "gt": "[[160.5, 62.0], [168.5, 67.0], [175.5, 68.0], [171.5, 78.0], [178.5, 102.0], [176.5, 114.0], [171.5, 114.0], [154.5, 105.0], [135.5, 97.0], [139.5, 100.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0010_user_0007_scene_0003_cfg_0005_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the handset of the telephone on the corresponding phone cradle', and now you are performing subtask 'pick up the handset from the table, transfer the handset from the table to the phone cradle, place the handset on the phone cradle.'\nBegin at [0.47, 0.4] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.4], [0.57, 0.46], [0.61, 0.51], [0.58, 0.75], [0.56, 0.81], [0.56, 0.83], [0.56, 0.78], [0.41, 0.66], [0.38, 0.61], [0.39, 0.58]]}" + } + ], + "images": "RH20T_cfg5_task_0010_user_0007_scene_0003_cfg_0005_037522062165#0", + "gt": "[[298, 143], [366, 165], [388, 184], [374, 271], [360, 290], [359, 300], [360, 280], [262, 237], [246, 219], [248, 210]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "56634_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the bottom drawer of the right cabinet', and now you are performing subtask 'pull the handle of the bottom drawer of the right cabinet to the position of closing the drawer.'\nYou start from [0.57, 0.57]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.57], [0.57, 0.57], [0.58, 0.57], [0.58, 0.57], [0.59, 0.56], [0.59, 0.55], [0.58, 0.54], [0.59, 0.53], [0.59, 0.54], [0.59, 0.53]]}" + } + ], + "images": "56634_exterior_image_2_left#122", + "gt": "[[183, 103], [183, 102], [186, 103], [186, 102], [189, 100], [189, 99], [187, 98], [188, 96], [190, 97], [190, 96]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "29259_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the glass lid from the pot', and now you are performing subtask 'remove the glass lid from the pot.'\nStarting at [0.68, 0.25], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.25], [0.6, 0.42], [0.58, 0.49], [0.58, 0.56], [0.57, 0.49], [0.51, 0.56], [0.49, 0.64], [0.49, 0.68], [0.5, 0.73], [0.49, 0.68]]}" + } + ], + "images": "29259_exterior_image_2_left#0", + "gt": "[[216, 45], [193, 75], [187, 88], [186, 100], [183, 89], [162, 100], [156, 115], [156, 122], [161, 131], [158, 122]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0037_user_0014_scene_0008_cfg_0006_104122064161#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the block on the stacked blocks, pick up the block on the table, transfer the block from the table to the stacked blocks, place the block on the stacked blocks.'\nYou begin at [0.51, 0.09]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.09], [0.55, 0.17], [0.57, 0.48], [0.55, 0.59], [0.55, 0.23], [0.48, 0.0], [0.53, 0.02], [0.55, 0.14], [0.53, 0.26], [0.56, 0.17]]}" + } + ], + "images": "RH20T_cfg6_task_0037_user_0014_scene_0008_cfg_0006_104122064161#575", + "gt": "[[327, 31], [355, 60], [364, 173], [355, 214], [349, 82], [308, 1], [342, 7], [350, 49], [339, 93], [356, 61]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "86813_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'get the pink hanger from the hanger pole and put it on the table', and now you are performing subtask 'get the pink hanger from the hanger pole and put it on the table.'\nBegin at [0.53, 0.6] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.6], [0.44, 0.41], [0.45, 0.29], [0.46, 0.32], [0.49, 0.29], [0.48, 0.28], [0.46, 0.31], [0.6, 0.36], [0.68, 0.49], [0.62, 0.46]]}" + } + ], + "images": "86813_exterior_image_2_left#0", + "gt": "[[169, 108], [142, 74], [145, 52], [148, 58], [157, 53], [154, 50], [146, 55], [191, 65], [218, 88], [198, 83]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "48472_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the green plush toy in the drawer', and now you are performing subtask 'put the green plush toy in the drawer.'\nFrom start point [0.66, 0.13], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.66, 0.13], [0.6, 0.09], [0.43, 0.08], [0.47, 0.11], [0.39, 0.22], [0.38, 0.39], [0.42, 0.23], [0.58, 0.47], [0.62, 0.47], [0.59, 0.36]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/48472_exterior_image_1_left#0.jpg", + "gt": "[[212, 23], [193, 16], [139, 15], [150, 19], [125, 39], [120, 70], [134, 42], [186, 84], [197, 85], [189, 64]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "57878_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the red-white book and the purple object in the black box', and now you are performing subtask 'pick up the purple object on the desk.'\nYour initial coordinate is [0.35, 0.29]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.29], [0.51, 0.26], [0.42, 0.45], [0.35, 0.53], [0.35, 0.46], [0.41, 0.29], [0.45, 0.41], [0.42, 0.37], [0.41, 0.47], [0.47, 0.51]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/57878_exterior_image_2_left#0.jpg", + "gt": "[[112, 53], [163, 46], [134, 81], [112, 96], [113, 83], [131, 52], [143, 74], [133, 66], [131, 85], [150, 92]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "30544_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the marker out of the bowl and put it on the table', and now you are performing subtask 'take the marker out of the bowl and put it on the table.'\nWith start location [0.59, 0.3], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.3], [0.52, 0.42], [0.49, 0.56], [0.5, 0.62], [0.48, 0.66], [0.46, 0.73], [0.47, 0.72], [0.48, 0.64], [0.46, 0.51], [0.47, 0.57]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/30544_exterior_image_2_left#0.jpg", + "gt": "[[188, 54], [165, 76], [158, 100], [160, 111], [153, 118], [146, 132], [150, 130], [153, 116], [148, 92], [151, 103]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0052_user_0007_scene_0010_cfg_0005_105422061350#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grab a basketball, release it and shoot it into the basket', and now you are performing subtask 'grab a basketball, release it and shoot it into the basket.'\nYou begin at [0.43, 0.13]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.13], [0.36, 0.09], [0.33, 0.28], [0.3, 0.32], [0.31, 0.4], [0.33, 0.18], [0.39, 0.01], [0.58, 0.0], [0.5, 0.02], [0.47, 0.25]]}" + } + ], + "images": "RH20T_cfg5_task_0052_user_0007_scene_0010_cfg_0005_105422061350#0", + "gt": "[[276, 47], [231, 33], [213, 100], [195, 114], [196, 143], [214, 64], [249, 5], [371, 0], [320, 6], [301, 90]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "52362_exterior_image_1_left#TrajectoryPrediction_Task#10", + "task": "TrajectoryPrediction_Task", + "p_index": 10, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'arrange the two small pillows against the right armrest of the sofa', and now you are performing subtask 'pick up the small pillow on the left side of the sofa.'\nStart at [0.51, 0.56]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.56], [0.52, 0.58], [0.55, 0.63], [0.55, 0.64], [0.53, 0.64], [0.51, 0.68], [0.5, 0.61], [0.51, 0.57], [0.5, 0.58], [0.54, 0.62]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/52362_exterior_image_1_left#506.jpg", + "gt": "[[164, 100], [166, 105], [176, 113], [176, 115], [171, 116], [164, 123], [161, 109], [162, 103], [160, 104], [174, 112]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "78997_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the purple object from the counter and put it on top of the shelf', and now you are performing subtask 'pick up the purple object from the counter and put it on top of the shelf.'\nStart at [0.34, 0.52]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.34, 0.52], [0.39, 0.73], [0.37, 0.81], [0.33, 0.74], [0.33, 0.39], [0.36, 0.25], [0.42, 0.3], [0.39, 0.27], [0.33, 0.4], [0.32, 0.54]]}" + } + ], + "images": "78997_exterior_image_1_left#0", + "gt": "[[109.5, 93.5], [124.5, 130.5], [117.5, 145.5], [105.5, 132.5], [104.5, 69.5], [116.5, 45.5], [133.5, 53.5], [125.5, 48.5], [106.5, 71.5], [101.5, 97.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0091_user_0014_scene_0007_cfg_0004_038522062288#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move an object from one box to another', and now you are performing subtask 'move an object from one box to another.'\nStarting at [0.37, 0.33], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.33], [0.39, 0.3], [0.35, 0.14], [0.39, 0.16], [0.41, 0.27], [0.39, 0.14], [0.43, 0.03], [0.45, 0.05], [0.42, 0.1], [0.44, 0.26]]}" + } + ], + "images": "RH20T_cfg4_task_0091_user_0014_scene_0007_cfg_0004_038522062288#0", + "gt": "[[235, 120], [250, 109], [225, 50], [247, 56], [265, 97], [249, 50], [274, 11], [287, 19], [266, 35], [279, 95]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0066_user_0014_scene_0007_cfg_0007_104122061330#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'place the building blocks on the pile of building blocks on the building board, pick up the building blocks on the table, transfer the building blocks from the desktop to the pile of building blocks on the building board, place the building blocks on the pile of building blocks on the building board.'\nYour initial coordinate is [0.5, 0.43]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.43], [0.54, 0.41], [0.58, 0.43], [0.63, 0.48], [0.63, 0.54], [0.62, 0.5], [0.53, 0.47], [0.45, 0.49], [0.44, 0.52], [0.48, 0.44]]}" + } + ], + "images": "RH20T_cfg7_task_0066_user_0014_scene_0007_cfg_0007_104122061330#675", + "gt": "[[319, 154], [344, 146], [370, 153], [403, 173], [405, 193], [396, 181], [338, 168], [291, 175], [283, 186], [305, 160]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "81823_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the cup to the right', and now you are performing subtask 'move the cup to the right.'\nBegin at [0.49, 0.34] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.34], [0.48, 0.4], [0.52, 0.47], [0.53, 0.55], [0.53, 0.43], [0.58, 0.36], [0.63, 0.39], [0.66, 0.51], [0.63, 0.4], [0.59, 0.36]]}" + } + ], + "images": "81823_exterior_image_1_left#0", + "gt": "[[155.5, 60.5], [152.5, 71.5], [166.5, 85.5], [168.5, 98.5], [168.5, 76.5], [185.5, 65.5], [202.5, 69.5], [212.5, 92.5], [200.5, 71.5], [187.5, 64.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122061602#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'place the block on the block board, pick up the block on the flat surface, transfer the block from the flat surface to the other blocks, place the block on the other blocks.'\nThe trajectory starts at [0.65, 0.43]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.43], [0.69, 0.4], [0.75, 0.42], [0.77, 0.47], [0.77, 0.42], [0.73, 0.44], [0.72, 0.52], [0.71, 0.51], [0.72, 0.47], [0.72, 0.47]]}" + } + ], + "images": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122061602#459", + "gt": "[[414, 156], [442, 144], [478, 152], [494, 170], [490, 150], [469, 159], [462, 187], [453, 184], [460, 171], [460, 171]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "52362_exterior_image_1_left#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'arrange the two small pillows against the right armrest of the sofa', and now you are performing subtask 'pick up the small pillow on the left side of the sofa.'\nYour starting position is [0.54, 0.61]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.61], [0.54, 0.62], [0.54, 0.63], [0.51, 0.67], [0.5, 0.63], [0.5, 0.59], [0.51, 0.56], [0.52, 0.55], [0.52, 0.56], [0.51, 0.56]]}" + } + ], + "images": "52362_exterior_image_1_left#422", + "gt": "[[174, 110], [172, 112], [173, 114], [163, 121], [160, 113], [160, 106], [163, 100], [166, 99], [168, 101], [164, 100]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0081_user_0015_scene_0008_cfg_0004_036422060909#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the leg of the glasses with one hand', and now you are performing subtask 'fold the right leg of the glasses on the table.'\nStarting at [0.57, 0.27], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.27], [0.53, 0.27], [0.57, 0.27], [0.55, 0.39], [0.56, 0.28], [0.48, 0.31], [0.54, 0.27], [0.5, 0.29], [0.55, 0.23], [0.53, 0.19]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0081_user_0015_scene_0008_cfg_0004_036422060909#177.jpg", + "gt": "[[363, 96], [342, 97], [363, 98], [349, 141], [357, 99], [308, 110], [344, 98], [320, 103], [349, 83], [340, 68]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0061_user_0020_scene_0001_cfg_0007_104122061018#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the lid of the gift box on the table, then take everything out of the gift box and place it all on the table', and now you are performing subtask 'place the cube on the tabletop, pick up the object from the gift box, transfer the object from the gift box to the tabletop, place the object on the tabletop.'\nThe trajectory starts at [0.4, 0.3]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.3], [0.4, 0.25], [0.42, 0.23], [0.43, 0.23], [0.49, 0.36], [0.47, 0.42], [0.47, 0.29], [0.45, 0.21], [0.46, 0.25], [0.45, 0.27]]}" + } + ], + "images": "RH20T_cfg7_task_0061_user_0020_scene_0001_cfg_0007_104122061018#366", + "gt": "[[255, 107], [259, 91], [267, 82], [273, 84], [313, 128], [304, 151], [304, 105], [289, 74], [296, 90], [288, 98]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0047_user_0020_scene_0004_cfg_0007_104122061602#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the cup off the cup rack', and now you are performing subtask 'take the cup off the cup rack.'\nWith start location [0.7, 0.32], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.32], [0.72, 0.35], [0.71, 0.29], [0.74, 0.3], [0.77, 0.36], [0.74, 0.26], [0.66, 0.23], [0.64, 0.27], [0.67, 0.21], [0.71, 0.21]]}" + } + ], + "images": "RH20T_cfg7_task_0047_user_0020_scene_0004_cfg_0007_104122061602#0", + "gt": "[[448, 116], [459, 125], [454, 105], [474, 108], [493, 129], [472, 94], [423, 82], [412, 98], [426, 77], [454, 75]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "33431_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the lid from the blue pot', and now you are performing subtask 'remove the lid from the blue pot.'\nBegin at [0.83, 0.52] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.83, 0.52], [0.76, 0.59], [0.66, 0.56], [0.67, 0.64], [0.65, 0.7], [0.58, 0.73], [0.57, 0.81], [0.57, 0.74], [0.66, 0.69], [0.65, 0.59]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/33431_exterior_image_1_left#0.jpg", + "gt": "[[265.0, 93.0], [243.0, 107.0], [210.0, 100.0], [215.0, 116.0], [207.0, 126.0], [186.0, 132.0], [182.0, 145.0], [183.0, 134.0], [211.0, 124.0], [207.0, 106.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0083_user_0015_scene_0004_cfg_0004_104122063550#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'swat with a flyswatter', and now you are performing subtask 'pick up the flyswatter from the surface, place the flyswatter on the surface.'\nYou begin at [0.57, 0.56]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.56], [0.56, 0.57], [0.58, 0.54], [0.57, 0.51], [0.6, 0.47], [0.61, 0.47], [0.6, 0.47], [0.54, 0.54], [0.56, 0.55], [0.55, 0.56]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0083_user_0015_scene_0004_cfg_0004_104122063550#116.jpg", + "gt": "[[367, 203], [361, 206], [369, 196], [368, 183], [381, 170], [391, 171], [383, 170], [345, 193], [357, 199], [354, 202]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "67484_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the faucet to the right, push the faucet handle upwards', and now you are performing subtask 'move the faucet to the right, push the faucet handle upwards.'\nWith start location [0.4, 0.47], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.47], [0.48, 0.4], [0.72, 0.51], [0.83, 0.49], [0.66, 0.47], [0.66, 0.42], [0.71, 0.39], [0.72, 0.27], [0.61, 0.36], [0.61, 0.36]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/67484_exterior_image_1_left#0.jpg", + "gt": "[[129.5, 85.5], [153.5, 72.5], [229.5, 91.5], [264.5, 87.5], [211.5, 84.5], [211.5, 75.5], [228.5, 69.5], [230.5, 48.5], [194.5, 64.5], [194.5, 64.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "78514_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick the gray object and move it to the left on the table', and now you are performing subtask 'pick the gray object and move it to the left on the table.'\nFrom start point [0.74, 0.01], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.74, 0.01], [0.68, 0.17], [0.65, 0.39], [0.63, 0.29], [0.6, 0.31], [0.59, 0.39], [0.62, 0.45], [0.6, 0.41], [0.59, 0.53], [0.65, 0.37]]}" + } + ], + "images": "78514_exterior_image_2_left#0", + "gt": "[[238, 1], [218, 31], [207, 70], [203, 53], [193, 55], [190, 70], [198, 81], [193, 74], [190, 96], [209, 67]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "90565_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'slide the tap to the left', and now you are performing subtask 'slide the tap to the left.'\nThe trajectory starts at [0.54, 0.02]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.02], [0.59, 0.18], [0.6, 0.37], [0.55, 0.36], [0.52, 0.25], [0.53, 0.21], [0.51, 0.21], [0.47, 0.16], [0.37, 0.15], [0.39, 0.17]]}" + } + ], + "images": "90565_exterior_image_1_left#0", + "gt": "[[174, 4], [190, 33], [192, 67], [175, 64], [167, 45], [169, 37], [164, 38], [151, 29], [119, 27], [126, 30]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0201_user_0015_scene_0007_cfg_0001_819112071037#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'the medium pipette tip is inserted into the holder for medium pipette tips', and now you are performing subtask 'place the tip of the medium pipette into the holder for medium pipette tips, place the tip of the medium pipette into the holder for medium pipette tips.'\nFrom start point [0.57, 0.24], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.24], [0.51, 0.21], [0.5, 0.21], [0.51, 0.23], [0.5, 0.21], [0.51, 0.21], [0.53, 0.24], [0.54, 0.22], [0.54, 0.25], [0.55, 0.23]]}" + } + ], + "images": "RH20T_cfg1_task_0201_user_0015_scene_0007_cfg_0001_819112071037#198", + "gt": "[[366, 88], [328, 74], [323, 77], [325, 83], [322, 74], [326, 74], [340, 85], [345, 81], [348, 89], [351, 84]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "56862_exterior_image_1_left#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the bottle of water, the black bag, the dark blue bag, the top plastic and the remaining snack pack in the box respectively', and now you are performing subtask 'place the transparent bag in the box.'\nYou begin at [0.58, 0.54]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.54], [0.54, 0.54], [0.47, 0.56], [0.47, 0.61], [0.52, 0.46], [0.56, 0.56], [0.55, 0.65], [0.55, 0.66], [0.53, 0.59], [0.48, 0.57]]}" + } + ], + "images": "56862_exterior_image_1_left#639", + "gt": "[[186, 98], [172, 97], [152, 101], [150, 109], [166, 82], [178, 100], [177, 117], [175, 118], [170, 107], [155, 103]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0025_user_0010_scene_0006_cfg_0005_104422070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold a piece of paper that has been folded once (one end is pressed down by a heavy object)', and now you are performing subtask 'unfold a piece of paper that has been folded once (one end is pressed down by a heavy object).'\nFrom start point [0.58, 0.37], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.37], [0.6, 0.39], [0.63, 0.4], [0.65, 0.46], [0.63, 0.5], [0.63, 0.48], [0.64, 0.51], [0.6, 0.6], [0.6, 0.5], [0.58, 0.37]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0025_user_0010_scene_0006_cfg_0005_104422070044#0.jpg", + "gt": "[[374, 132], [382, 142], [404, 145], [413, 166], [406, 181], [404, 174], [409, 184], [383, 216], [384, 181], [374, 134]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0094_user_0014_scene_0003_cfg_0004_104122062823#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'push the lid on the top of the box, push the lid on the bottom of the box.'\nYou start from [0.65, 0.0]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.0], [0.64, 0.03], [0.64, 0.04], [0.66, 0.0], [0.68, 0.0], [0.67, 0.0], [0.67, 0.02], [0.67, 0.0], [0.63, 0.0], [0.62, 0.04]]}" + } + ], + "images": "RH20T_cfg4_task_0094_user_0014_scene_0003_cfg_0004_104122062823#311", + "gt": "[[413, 0], [408, 9], [410, 16], [421, 0], [436, 0], [429, 0], [426, 6], [430, 0], [403, 0], [398, 13]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "89810_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the blue cloth from the back of the chair', and now you are performing subtask 'pick up the blue cloth from the back of the chair.'\nFrom start point [0.31, 0.64], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.31, 0.64], [0.42, 0.6], [0.45, 0.68], [0.48, 0.7], [0.46, 0.74], [0.47, 0.77], [0.49, 0.72], [0.47, 0.69], [0.43, 0.71], [0.44, 0.74]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/89810_exterior_image_2_left#0.jpg", + "gt": "[[99, 116], [135, 108], [144, 122], [154, 126], [147, 133], [150, 139], [158, 130], [149, 124], [139, 128], [140, 133]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0094_user_0007_scene_0009_cfg_0005_105422061350#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'press the cardboard box on the side, press the cardboard box on the side.'\nBegin at [0.36, 0.18] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.18], [0.37, 0.17], [0.38, 0.22], [0.38, 0.2], [0.36, 0.17], [0.39, 0.2], [0.41, 0.17], [0.46, 0.23], [0.46, 0.25], [0.45, 0.21]]}" + } + ], + "images": "RH20T_cfg5_task_0094_user_0007_scene_0009_cfg_0005_105422061350#320", + "gt": "[[229, 65], [234, 62], [242, 79], [243, 73], [228, 62], [251, 72], [263, 63], [293, 83], [295, 91], [285, 76]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "62491_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the switch on the socket, open the box', and now you are performing subtask 'pull the lid on the box.'\nYour initial coordinate is [0.56, 0.38]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.56, 0.38], [0.49, 0.38], [0.47, 0.47], [0.41, 0.36], [0.39, 0.35], [0.4, 0.4], [0.35, 0.28], [0.28, 0.29], [0.34, 0.34], [0.44, 0.3]]}" + } + ], + "images": "62491_exterior_image_2_left#98", + "gt": "[[180, 68], [158, 68], [150, 85], [131, 64], [126, 63], [129, 72], [112, 51], [88, 52], [109, 62], [142, 54]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0064_user_0015_scene_0010_cfg_0001_035622060973#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the roll of paper from the desk and place it on the shelf', and now you are performing subtask 'pick up the roll of paper from the desk and place it on the shelf.'\nThe trajectory starts at [0.38, 0.36]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.36], [0.33, 0.43], [0.4, 0.6], [0.38, 0.59], [0.51, 0.54], [0.48, 0.59], [0.59, 0.59], [0.49, 0.62], [0.36, 0.44], [0.37, 0.43]]}" + } + ], + "images": "RH20T_cfg1_task_0064_user_0015_scene_0010_cfg_0001_035622060973#0", + "gt": "[[246, 130], [211, 154], [259, 216], [243, 214], [324, 193], [305, 214], [379, 214], [314, 223], [228, 159], [238, 156]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "39884_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the top left corner of the towel towards the center', and now you are performing subtask 'fold the top left corner of the towel towards the center.'\nWith start location [0.92, 0.02], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.92, 0.02], [0.78, 0.07], [0.62, 0.19], [0.58, 0.34], [0.58, 0.51], [0.59, 0.44], [0.63, 0.42], [0.65, 0.42], [0.66, 0.5], [0.68, 0.41]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/39884_exterior_image_2_left#0.jpg", + "gt": "[[293, 3], [249, 13], [200, 35], [185, 61], [187, 92], [190, 80], [202, 75], [209, 76], [211, 90], [216, 74]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "70873_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the bottle to the left, use the blue towel to wipe the table', and now you are performing subtask 'wipe the table with the yellow towel.'\nYour initial coordinate is [0.52, 0.56]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.56], [0.51, 0.45], [0.6, 0.53], [0.48, 0.53], [0.52, 0.49], [0.49, 0.54], [0.6, 0.57], [0.55, 0.58], [0.57, 0.41], [0.57, 0.41]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/70873_exterior_image_1_left#294.jpg", + "gt": "[[167.0, 100.0], [164.0, 81.0], [191.0, 96.0], [153.0, 95.0], [166.0, 89.0], [158.0, 98.0], [193.0, 103.0], [175.0, 104.0], [184.0, 73.0], [184.0, 73.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "51525_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the microwave door, move the pink jar on the counter', and now you are performing subtask 'pick up the pink jar on the microwave.'\nThe trajectory starts at [0.52, 0.38]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.38], [0.57, 0.36], [0.58, 0.37], [0.59, 0.36], [0.47, 0.21], [0.46, 0.16], [0.47, 0.13], [0.47, 0.14], [0.48, 0.14], [0.49, 0.12]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/51525_exterior_image_2_left#61.jpg", + "gt": "[[166, 68], [182, 64], [185, 66], [188, 65], [150, 37], [147, 28], [150, 23], [151, 25], [154, 25], [156, 22]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0061_user_0015_scene_0001_cfg_0001_038522062547#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the things out of the gift box and put them on the table', and now you are performing subtask 'place the block on the table, pick up the orange ball from the gift box, transfer the gift box from the table to the table, place the orange ball on the table.'\nYou start from [0.64, 0.51]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.64, 0.51], [0.62, 0.43], [0.59, 0.2], [0.6, 0.14], [0.55, 0.16], [0.55, 0.19], [0.6, 0.17], [0.7, 0.45], [0.69, 0.48], [0.7, 0.46]]}" + } + ], + "images": "RH20T_cfg1_task_0061_user_0015_scene_0001_cfg_0001_038522062547#613", + "gt": "[[412, 182], [394, 156], [376, 71], [381, 51], [353, 59], [351, 68], [383, 62], [446, 161], [443, 174], [447, 164]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0094_user_0007_scene_0009_cfg_0005_036422060215#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'press the cardboard box on the side.'\nWith start location [0.45, 0.37], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.37], [0.45, 0.37], [0.44, 0.36], [0.42, 0.29], [0.41, 0.26], [0.4, 0.3], [0.4, 0.27], [0.4, 0.27], [0.44, 0.32], [0.45, 0.33]]}" + } + ], + "images": "RH20T_cfg5_task_0094_user_0007_scene_0009_cfg_0005_036422060215#430", + "gt": "[[287, 132], [286, 133], [282, 128], [271, 105], [260, 92], [253, 107], [256, 97], [258, 96], [279, 116], [286, 120]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0023_user_0014_scene_0002_cfg_0001_819112070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the power strip by pressing the button', and now you are performing subtask 'turn on the power strip by pressing the button.'\nYou start from [0.62, 0.46]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.46], [0.6, 0.49], [0.61, 0.49], [0.6, 0.52], [0.6, 0.59], [0.59, 0.61], [0.58, 0.64], [0.58, 0.66], [0.59, 0.65], [0.6, 0.59]]}" + } + ], + "images": "RH20T_cfg1_task_0023_user_0014_scene_0002_cfg_0001_819112070044#0", + "gt": "[[400, 167], [386, 178], [389, 178], [384, 186], [384, 213], [375, 220], [373, 231], [374, 237], [378, 234], [384, 211]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "30962_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the lid on the black pot and put it on the blue pot, then take the object in the pot and put it on the counter, finally take the lid on the blue pot and put it back on the black pot', and now you are performing subtask 'place the lid on the blue pot.'\nYou begin at [0.38, 0.25]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.25], [0.45, 0.32], [0.52, 0.32], [0.56, 0.33], [0.57, 0.38], [0.62, 0.43], [0.62, 0.47], [0.63, 0.41], [0.66, 0.37], [0.65, 0.44]]}" + } + ], + "images": "30962_exterior_image_2_left#0", + "gt": "[[123, 45], [143, 58], [166, 57], [179, 60], [182, 69], [197, 78], [199, 85], [202, 73], [211, 66], [208, 79]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "30221_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer', and now you are performing subtask 'open the drawer.'\nYou start from [0.4, 0.44]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.44], [0.45, 0.45], [0.45, 0.55], [0.53, 0.65], [0.52, 0.76], [0.57, 0.74], [0.48, 0.9], [0.46, 0.85], [0.5, 0.75], [0.5, 0.75]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/30221_exterior_image_1_left#0.jpg", + "gt": "[[127.5, 79.5], [142.5, 81.5], [145.5, 98.5], [170.5, 116.5], [165.5, 136.5], [181.5, 133.5], [152.5, 161.5], [148.5, 152.5], [158.5, 134.5], [158.5, 134.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "60678_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the marker from the table and put it in the cup', and now you are performing subtask 'take the marker from the table and put it in the cup.'\nWith start location [0.62, 0.58], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.58], [0.54, 0.59], [0.52, 0.61], [0.51, 0.67], [0.54, 0.77], [0.54, 0.75], [0.53, 0.68], [0.52, 0.58], [0.54, 0.65], [0.55, 0.63]]}" + } + ], + "images": "60678_exterior_image_1_left#0", + "gt": "[[197.5, 104.5], [173.5, 106.5], [167.5, 109.5], [163.5, 120.5], [172.5, 138.5], [173.5, 135.5], [168.5, 121.5], [167.5, 104.5], [171.5, 116.5], [174.5, 113.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "40683_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'wipe the table with the green towel', and now you are performing subtask 'wipe the table with the green towel.'\nFrom start point [0.28, 0.05], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.28, 0.05], [0.48, 0.49], [0.59, 0.58], [0.61, 0.69], [0.57, 0.61], [0.62, 0.69], [0.59, 0.56], [0.65, 0.7], [0.57, 0.57], [0.59, 0.43]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/40683_exterior_image_1_left#0.jpg", + "gt": "[[89, 9], [154, 88], [189, 104], [195, 124], [184, 109], [199, 124], [188, 101], [208, 126], [181, 102], [189, 78]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "17853_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the spoon out of the mug and put it on the table', and now you are performing subtask 'take the spoon out of the mug and put it on the table.'\nThe trajectory starts at [0.53, 0.0]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.0], [0.47, 0.04], [0.44, 0.15], [0.43, 0.24], [0.46, 0.0], [0.49, 0.0], [0.51, 0.06], [0.52, 0.17], [0.48, 0.32], [0.48, 0.15]]}" + } + ], + "images": "17853_exterior_image_2_left#0", + "gt": "[[170, 0], [151, 7], [141, 27], [139, 44], [146, 0], [156, 0], [163, 11], [165, 31], [153, 58], [155, 27]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "75812_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the clothes in the bag', and now you are performing subtask 'pick up the clothes on the table.'\nYou start from [0.37, 0.78]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.78], [0.59, 0.71], [0.7, 0.83], [0.69, 0.87], [0.57, 0.64], [0.52, 0.65], [0.51, 0.74], [0.59, 0.62], [0.53, 0.63], [0.49, 0.75]]}" + } + ], + "images": "75812_exterior_image_1_left#0", + "gt": "[[117.5, 141.0], [189.5, 127.0], [225.5, 150.0], [221.5, 156.0], [182.5, 116.0], [167.5, 117.0], [163.5, 133.0], [187.5, 111.0], [170.5, 114.0], [156.5, 135.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_035622060973#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the first block from the desktop, transfer the first block from the desktop to the second position above the first block, place the first block on top of the second position.'\nThe trajectory starts at [0.39, 0.36]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.36], [0.34, 0.35], [0.34, 0.36], [0.31, 0.36], [0.29, 0.39], [0.29, 0.44], [0.3, 0.44], [0.31, 0.41], [0.32, 0.47], [0.35, 0.49]]}" + } + ], + "images": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_035622060973#0", + "gt": "[[247, 129], [220, 125], [220, 131], [200, 130], [187, 142], [185, 157], [195, 159], [200, 149], [208, 169], [223, 176]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0004_user_0006_scene_0003_cfg_0002_104422070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the small block on the left and move it to the right', and now you are performing subtask 'pick up the small block on the left and move it to the right.'\nStarting at [0.48, 0.52], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.52], [0.59, 0.51], [0.62, 0.54], [0.63, 0.56], [0.61, 0.66], [0.61, 0.73], [0.62, 0.69], [0.56, 0.65], [0.55, 0.69], [0.53, 0.61]]}" + } + ], + "images": "RH20T_cfg2_task_0004_user_0006_scene_0003_cfg_0002_104422070044#0", + "gt": "[[309, 188], [376, 184], [397, 196], [403, 203], [388, 238], [391, 263], [395, 249], [359, 233], [353, 249], [341, 220]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0033_user_0015_scene_0008_cfg_0001_035622060973#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the block', and now you are performing subtask 'push the square in front of the cardboard to the adjacent square in the middle of the cardboard, pick up the square on the cardboard, transfer the square from the cardboard to the right side of the middle square on the cardboard.'\nThe trajectory starts at [0.38, 0.36]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.36], [0.0, 1.0], [0.33, 0.43], [0.29, 0.51], [0.32, 0.58], [0.38, 0.53], [0.37, 0.5], [0.39, 0.52], [0.37, 0.47], [0.38, 0.45]]}" + } + ], + "images": "RH20T_cfg1_task_0033_user_0015_scene_0008_cfg_0001_035622060973#0", + "gt": "[[246, 130], [0, 360], [209, 156], [187, 183], [205, 210], [243, 190], [236, 181], [247, 187], [236, 170], [244, 161]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "61861_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the cloth and wipe the pot then place it on the table', and now you are performing subtask 'pick up the cloth and wipe the pot then place it on the table.'\nYou start from [0.38, 0.01]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.01], [0.43, 0.45], [0.43, 0.52], [0.38, 0.46], [0.4, 0.73], [0.37, 0.67], [0.36, 0.49], [0.41, 0.46], [0.42, 0.51], [0.46, 0.42]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/61861_exterior_image_2_left#0.jpg", + "gt": "[[122, 1], [139, 81], [136, 93], [121, 82], [127, 132], [117, 120], [115, 88], [132, 83], [135, 91], [146, 76]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0081_user_0010_scene_0001_cfg_0001_035622060973#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the leg of the glasses (with one hand)', and now you are performing subtask 'pull the leg of the glasses outward, pull the other leg of the glasses outward.'\nStart at [0.41, 0.55]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.55], [0.42, 0.48], [0.41, 0.49], [0.42, 0.49], [0.42, 0.51], [0.41, 0.59], [0.42, 0.58], [0.42, 0.62], [0.42, 0.58], [0.42, 0.59]]}" + } + ], + "images": "RH20T_cfg1_task_0081_user_0010_scene_0001_cfg_0001_035622060973#291", + "gt": "[[260, 198], [267, 174], [264, 177], [268, 176], [269, 183], [265, 212], [269, 210], [268, 224], [270, 208], [267, 212]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0081_user_0007_scene_0004_cfg_0004_104422070011#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the leg of the glasses (with one hand)', and now you are performing subtask 'pick up the glasses on the patterned blanket, transfer the glasses from the patterned blanket to the vertical position.'\nYour initial coordinate is [0.44, 0.53]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.53], [0.44, 0.53], [0.44, 0.57], [0.43, 0.45], [0.46, 0.48], [0.48, 0.34], [0.52, 0.45], [0.54, 0.51], [0.53, 0.52], [0.51, 0.49]]}" + } + ], + "images": "RH20T_cfg4_task_0081_user_0007_scene_0004_cfg_0004_104422070011#154", + "gt": "[[284, 190], [279, 191], [279, 205], [277, 163], [297, 172], [305, 122], [334, 161], [344, 184], [341, 188], [325, 178]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "10884_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker into the cup', and now you are performing subtask 'put the marker into the cup.'\nThe trajectory starts at [0.28, 0.27]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.28, 0.27], [0.34, 0.32], [0.4, 0.47], [0.37, 0.67], [0.38, 0.77], [0.45, 0.85], [0.49, 0.66], [0.52, 0.82], [0.57, 0.89], [0.53, 0.79]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/10884_exterior_image_1_left#0.jpg", + "gt": "[[91, 48], [109, 58], [128, 84], [117, 120], [120, 138], [143, 153], [157, 119], [167, 147], [181, 161], [170, 142]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "88693_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the white lid from the food container and put it on the table', and now you are performing subtask 'remove the white lid from the food container and put it on the table.'\nBegin at [0.25, 0.25] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.25, 0.25], [0.47, 0.31], [0.48, 0.59], [0.47, 0.42], [0.52, 0.47], [0.49, 0.5], [0.5, 0.39], [0.59, 0.39], [0.6, 0.73], [0.58, 0.44]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/88693_exterior_image_2_left#0.jpg", + "gt": "[[79, 45], [149, 55], [155, 107], [150, 75], [166, 85], [156, 90], [161, 70], [189, 70], [191, 132], [187, 80]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "35815_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the white cloth from the black frame', and now you are performing subtask 'remove the white cloth from the black frame.'\nWith start location [0.41, 0.02], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.02], [0.38, 0.01], [0.39, 0.06], [0.49, 0.15], [0.48, 0.23], [0.45, 0.26], [0.42, 0.19], [0.38, 0.17], [0.35, 0.3], [0.29, 0.34]]}" + } + ], + "images": "35815_exterior_image_2_left#0", + "gt": "[[131, 3], [123, 2], [126, 11], [156, 27], [155, 41], [144, 46], [135, 34], [121, 30], [111, 54], [94, 62]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "48125_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the kettle', and now you are performing subtask 'close the kettle.'\nYour initial coordinate is [0.25, 0.44]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.25, 0.44], [0.23, 0.42], [0.23, 0.37], [0.25, 0.32], [0.28, 0.31], [0.39, 0.34], [0.39, 0.42], [0.37, 0.37], [0.32, 0.39], [0.29, 0.37]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/48125_exterior_image_2_left#0.jpg", + "gt": "[[79, 79], [73, 75], [73, 67], [80, 58], [89, 55], [124, 61], [126, 76], [119, 67], [102, 70], [92, 66]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "5850_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the grey cup in the orange cup', and now you are performing subtask 'put the grey cup in the orange cup.'\nWith start location [0.97, 0.07], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.98, 0.07], [0.85, 0.29], [0.78, 0.23], [0.71, 0.26], [0.7, 0.21], [0.77, 0.18], [0.79, 0.19], [0.74, 0.2], [0.71, 0.33], [0.73, 0.28]]}" + } + ], + "images": "5850_exterior_image_2_left#0", + "gt": "[[312, 12], [272, 53], [250, 42], [226, 46], [225, 37], [246, 32], [253, 35], [236, 36], [226, 60], [235, 50]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_819112070044#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the third button on the desk.'\nBegin at [0.48, 0.53] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.53], [0.47, 0.52], [0.48, 0.53], [0.45, 0.57], [0.45, 0.59], [0.45, 0.54], [0.51, 0.47], [0.54, 0.49], [0.59, 0.48], [0.62, 0.44]]}" + } + ], + "images": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_819112070044#314", + "gt": "[[309, 191], [300, 188], [305, 192], [287, 205], [290, 214], [289, 196], [328, 168], [345, 175], [376, 172], [394, 159]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0111_user_0007_scene_0001_cfg_0004_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the object with the corresponding shape into the corresponding hole', and now you are performing subtask 'put the object with the corresponding shape into the corresponding hole.'\nYour starting position is [0.45, 0.0]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.0], [0.56, 0.09], [0.62, 0.24], [0.65, 0.19], [0.66, 0.31], [0.66, 0.2], [0.6, 0.14], [0.67, 0.16], [0.6, 0.08], [0.63, 0.14]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0111_user_0007_scene_0001_cfg_0004_104122062823#0.jpg", + "gt": "[[289, 0], [358, 31], [396, 87], [417, 70], [423, 113], [422, 71], [387, 51], [430, 58], [382, 27], [405, 50]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "17955_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the green can from the cabinet and put it on the counter', and now you are performing subtask 'remove the green can from the cabinet and put it on the counter.'\nYour starting position is [0.7, 0.41]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.41], [0.59, 0.11], [0.53, 0.07], [0.48, 0.17], [0.48, 0.1], [0.55, 0.04], [0.59, 0.17], [0.59, 0.59], [0.55, 0.75], [0.59, 0.69]]}" + } + ], + "images": "17955_exterior_image_2_left#0", + "gt": "[[225, 73], [189, 20], [170, 12], [154, 31], [155, 18], [176, 8], [189, 31], [190, 107], [176, 135], [188, 124]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "23179_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the black marker in the pot', and now you are performing subtask 'put the black marker in the pot.'\nThe trajectory starts at [0.55, 0.02]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.02], [0.51, 0.12], [0.49, 0.21], [0.51, 0.49], [0.52, 0.4], [0.55, 0.33], [0.63, 0.31], [0.62, 0.39], [0.61, 0.29], [0.61, 0.29]]}" + } + ], + "images": "23179_exterior_image_2_left#0", + "gt": "[[177, 3], [162, 22], [156, 38], [163, 89], [167, 72], [176, 60], [203, 56], [198, 70], [196, 52], [196, 52]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0004_user_0010_scene_0001_cfg_0005_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the small block on the left and move it to the right', and now you are performing subtask 'pick up the small block on the left and move it to the right.'\nStarting at [0.48, 0.14], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.14], [0.49, 0.22], [0.48, 0.47], [0.44, 0.56], [0.44, 0.74], [0.46, 0.63], [0.52, 0.49], [0.51, 0.31], [0.54, 0.27], [0.55, 0.29]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0004_user_0010_scene_0001_cfg_0005_104122061850#0.jpg", + "gt": "[[305, 50], [314, 79], [305, 170], [280, 203], [284, 265], [294, 227], [333, 177], [327, 110], [345, 98], [355, 106]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "60757_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the black marker into the white cup', and now you are performing subtask 'put the black marker into the white cup.'\nBegin at [0.82, 0.32] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.82, 0.32], [0.66, 0.46], [0.67, 0.58], [0.64, 0.62], [0.62, 0.59], [0.63, 0.57], [0.62, 0.42], [0.6, 0.39], [0.6, 0.45], [0.57, 0.46]]}" + } + ], + "images": "60757_exterior_image_1_left#0", + "gt": "[[262, 57], [212, 83], [213, 105], [206, 112], [198, 106], [203, 102], [199, 76], [191, 70], [191, 81], [184, 83]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "76853_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the towel and put it on the black rack', and now you are performing subtask 'pick up the towel from the black rack.'\nYou begin at [0.98, 0.33]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.98, 0.33], [0.77, 0.79], [0.85, 0.48], [0.84, 0.34], [0.87, 0.14], [0.87, 0.12], [0.79, 0.21], [0.79, 0.3], [0.8, 0.33], [0.79, 0.38]]}" + } + ], + "images": "76853_exterior_image_2_left#0", + "gt": "[[314, 59], [246, 143], [272, 87], [269, 62], [279, 25], [277, 22], [254, 38], [252, 54], [256, 60], [252, 69]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "70682_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the yellow plush toy in the silver pot', and now you are performing subtask 'put the yellow plush toy in the silver pot.'\nYour starting position is [0.79, 0.01]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.79, 0.01], [0.53, 0.53], [0.54, 0.56], [0.54, 0.51], [0.55, 0.48], [0.65, 0.51], [0.62, 0.46], [0.6, 0.49], [0.59, 0.41], [0.59, 0.41]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/70682_exterior_image_2_left#0.jpg", + "gt": "[[254, 2], [170, 95], [172, 101], [173, 91], [177, 87], [209, 92], [199, 83], [193, 89], [188, 73], [188, 73]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0001_user_0010_scene_0004_cfg_0004_104122062823#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the button from top to bottom', and now you are performing subtask 'press the button on the surface from top to bottom.'\nBegin at [0.57, 0.15] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.15], [0.55, 0.22], [0.53, 0.16], [0.55, 0.09], [0.53, 0.14], [0.54, 0.16], [0.54, 0.12], [0.55, 0.18], [0.55, 0.1], [0.54, 0.08]]}" + } + ], + "images": "RH20T_cfg4_task_0001_user_0010_scene_0004_cfg_0004_104122062823#202", + "gt": "[[362, 55], [351, 80], [342, 57], [349, 33], [342, 52], [346, 58], [347, 45], [351, 64], [354, 37], [343, 29]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0058_user_0014_scene_0008_cfg_0004_038522062288#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn off the power strips', and now you are performing subtask 'press the third button of the power socket to turn it off, press the fourth button of the power socket to turn it off.'\nBegin at [0.45, 0.2] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.2], [0.45, 0.19], [0.45, 0.19], [0.45, 0.19], [0.46, 0.21], [0.46, 0.21], [0.45, 0.19], [0.44, 0.19], [0.43, 0.19], [0.43, 0.18]]}" + } + ], + "images": "RH20T_cfg4_task_0058_user_0014_scene_0008_cfg_0004_038522062288#182", + "gt": "[[285, 72], [285, 70], [288, 69], [288, 70], [293, 74], [293, 75], [290, 70], [279, 67], [274, 67], [273, 65]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0061_user_0007_scene_0008_cfg_0005_037522062165#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the first block out of the box and place it on the table, then take the second block and place it on top of the first block', and now you are performing subtask 'place the first block on the table, pick up the second block in the box, transfer the second block from the box to above the first block, place the second block on the first block.'\nFrom start point [0.53, 0.72], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.72], [0.52, 0.67], [0.44, 0.6], [0.4, 0.58], [0.38, 0.63], [0.48, 0.61], [0.51, 0.64], [0.52, 0.69], [0.51, 0.61], [0.43, 0.5]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0061_user_0007_scene_0008_cfg_0005_037522062165#286.jpg", + "gt": "[[336, 260], [332, 242], [281, 216], [254, 210], [242, 226], [310, 218], [329, 229], [331, 250], [326, 221], [274, 180]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "49729_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the glass lid on the black pot', and now you are performing subtask 'place the glass lid on the black pot.'\nYou begin at [0.97, 0.31]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.98, 0.31], [0.83, 0.47], [0.69, 0.86], [0.7, 0.72], [0.76, 0.6], [0.79, 0.59], [0.8, 0.63], [0.78, 0.64], [0.79, 0.59], [0.79, 0.59]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/49729_exterior_image_2_left#0.jpg", + "gt": "[[312, 56], [266, 85], [222, 154], [224, 129], [243, 108], [254, 106], [255, 113], [250, 115], [254, 106], [254, 106]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "62551_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the faucet, open the faucet, close the faucet', and now you are performing subtask 'pull the faucet up.'\nYou begin at [0.22, 0.47]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.22, 0.47], [0.24, 0.48], [0.24, 0.49], [0.25, 0.52], [0.25, 0.48], [0.25, 0.49], [0.26, 0.47], [0.25, 0.43], [0.25, 0.42], [0.26, 0.42]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/62551_exterior_image_1_left#280.jpg", + "gt": "[[71, 85], [78, 87], [77, 88], [80, 93], [80, 87], [79, 88], [82, 84], [81, 78], [81, 76], [82, 76]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_039422060546#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the first button on the desk, press the second button on the desk.'\nWith start location [0.41, 0.34], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.34], [0.38, 0.41], [0.35, 0.68], [0.33, 0.74], [0.34, 0.79], [0.32, 0.84], [0.33, 0.86], [0.32, 0.83], [0.33, 0.85], [0.37, 0.69]]}" + } + ], + "images": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_039422060546#0", + "gt": "[[260, 123], [240, 147], [222, 244], [211, 266], [219, 286], [205, 303], [213, 310], [208, 299], [213, 306], [235, 249]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0038_user_0007_scene_0009_cfg_0005_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the red cup from the cube to the table', and now you are performing subtask 'move the red cup from the cube to the table.'\nYou begin at [0.45, 0.3]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.3], [0.43, 0.34], [0.51, 0.52], [0.5, 0.54], [0.5, 0.54], [0.49, 0.56], [0.48, 0.58], [0.48, 0.52], [0.45, 0.45], [0.43, 0.46]]}" + } + ], + "images": "RH20T_cfg5_task_0038_user_0007_scene_0009_cfg_0005_036422060215#0", + "gt": "[[286, 108], [278, 124], [327, 187], [323, 196], [319, 193], [312, 200], [310, 210], [306, 188], [286, 163], [278, 164]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "50306_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the tissue box on top then place it back on the other box', and now you are performing subtask 'pick up the tissue box on top then place it back on the other box.'\nStarting at [0.38, 0.56], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.56], [0.46, 0.53], [0.42, 0.58], [0.47, 0.61], [0.45, 0.59], [0.44, 0.63], [0.49, 0.56], [0.45, 0.56], [0.44, 0.64], [0.45, 0.51]]}" + } + ], + "images": "50306_exterior_image_2_left#0", + "gt": "[[122, 100], [148, 95], [135, 105], [149, 109], [143, 106], [141, 113], [156, 100], [144, 100], [141, 115], [145, 92]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0037_user_0014_scene_0010_cfg_0001_819112070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up a block from the table, transfer a block from the table to the top of a block.'\nStarting at [0.62, 0.46], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.46], [0.6, 0.48], [0.57, 0.53], [0.56, 0.54], [0.55, 0.55], [0.54, 0.57], [0.54, 0.59], [0.54, 0.6], [0.53, 0.52], [0.52, 0.49]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0037_user_0014_scene_0010_cfg_0001_819112070044#0.jpg", + "gt": "[[400, 167], [386, 173], [367, 192], [360, 193], [350, 199], [348, 207], [344, 211], [345, 217], [342, 187], [335, 178]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "7475_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the mug backwards then forwards', and now you are performing subtask 'slide the mug forward on the table.'\nStart at [0.5, 0.34]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.34], [0.56, 0.28], [0.55, 0.32], [0.56, 0.3], [0.53, 0.32], [0.5, 0.31], [0.53, 0.33], [0.5, 0.37], [0.61, 0.43], [0.57, 0.46]]}" + } + ], + "images": "7475_exterior_image_1_left#153", + "gt": "[[160, 61], [179, 51], [176, 57], [179, 54], [171, 58], [160, 56], [169, 60], [160, 66], [194, 78], [182, 83]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "91226_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the green marker inside the yellow mug', and now you are performing subtask 'put the green marker inside the yellow mug.'\nYou start from [0.47, 0.37]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.37], [0.43, 0.42], [0.5, 0.47], [0.43, 0.47], [0.43, 0.66], [0.44, 0.68], [0.45, 0.64], [0.44, 0.54], [0.38, 0.63], [0.36, 0.62]]}" + } + ], + "images": "91226_exterior_image_2_left#0", + "gt": "[[149, 67], [139, 75], [159, 84], [138, 85], [138, 119], [142, 123], [145, 115], [140, 97], [120, 114], [116, 111]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "78996_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick the cob on the pan and put it on the table', and now you are performing subtask 'pick the cob on the pan and put it on the table.'\nYour initial coordinate is [0.27, 0.01]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.27, 0.01], [0.46, 0.32], [0.49, 0.42], [0.49, 0.36], [0.52, 0.39], [0.53, 0.56], [0.51, 0.4], [0.47, 0.36], [0.51, 0.38], [0.48, 0.36]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/78996_exterior_image_1_left#0.jpg", + "gt": "[[85, 1], [148, 57], [156, 75], [158, 64], [168, 71], [171, 101], [164, 72], [151, 64], [164, 68], [155, 64]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0004_user_0006_scene_0003_cfg_0002_105422061350#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the small block on the left and move it to the right', and now you are performing subtask 'pick up the small block on the left and move it to the right.'\nYou begin at [0.58, 0.42]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.43], [0.59, 0.27], [0.62, 0.19], [0.64, 0.17], [0.68, 0.25], [0.65, 0.34], [0.66, 0.29], [0.66, 0.32], [0.62, 0.44], [0.65, 0.36]]}" + } + ], + "images": "RH20T_cfg2_task_0004_user_0006_scene_0003_cfg_0002_105422061350#0", + "gt": "[[369, 153], [378, 97], [394, 67], [408, 63], [436, 90], [413, 122], [424, 106], [425, 117], [398, 158], [416, 131]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0033_user_0020_scene_0001_cfg_0007_104122061330#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pushing another small block to the designated position while grasping one small block', and now you are performing subtask 'place the front square in the position aligned with the second square, push the front square to move the second square to the position on the black dot on the cardboard.'\nThe trajectory starts at [0.63, 0.29]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.29], [0.61, 0.38], [0.52, 0.58], [0.53, 0.54], [0.54, 0.5], [0.53, 0.38], [0.53, 0.36], [0.51, 0.33], [0.5, 0.32], [0.49, 0.33]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0033_user_0020_scene_0001_cfg_0007_104122061330#279.jpg", + "gt": "[[403, 106], [392, 135], [331, 210], [341, 193], [343, 180], [340, 137], [338, 130], [326, 119], [318, 115], [314, 118]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "88528_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the glue stick inside the mug cup', and now you are performing subtask 'put the glue stick inside the mug cup.'\nThe trajectory starts at [0.4, 0.01]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.01], [0.45, 0.24], [0.46, 0.42], [0.5, 0.54], [0.45, 0.62], [0.53, 0.49], [0.52, 0.46], [0.53, 0.48], [0.5, 0.55], [0.52, 0.47]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/88528_exterior_image_2_left#0.jpg", + "gt": "[[127, 1], [143, 44], [146, 75], [159, 97], [144, 112], [169, 89], [168, 83], [171, 86], [159, 99], [165, 85]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "17163_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick the marker and put it in the cup', and now you are performing subtask 'pick the marker and put it in the cup.'\nFrom start point [0.38, 0.85], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.85], [0.44, 0.78], [0.47, 0.78], [0.53, 0.91], [0.48, 0.69], [0.41, 0.68], [0.44, 0.66], [0.47, 0.69], [0.44, 0.72], [0.44, 0.72]]}" + } + ], + "images": "17163_exterior_image_2_left#0", + "gt": "[[123, 153], [141, 141], [149, 141], [170, 163], [155, 124], [130, 122], [140, 118], [152, 125], [141, 129], [141, 129]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122060811#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a sliding window', and now you are performing subtask 'open a sliding window.'\nStart at [0.6, 0.35]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.35], [0.58, 0.38], [0.58, 0.39], [0.58, 0.36], [0.57, 0.39], [0.58, 0.38], [0.6, 0.4], [0.6, 0.37], [0.6, 0.41], [0.6, 0.36]]}" + } + ], + "images": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122060811#0", + "gt": "[[385, 127], [369, 136], [373, 140], [369, 130], [366, 142], [372, 136], [387, 145], [386, 134], [384, 148], [385, 129]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0057_user_0007_scene_0002_cfg_0005_036422060215#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press a row of four power strip buttons to turn on the power strips', and now you are performing subtask 'press the third button on the power socket, press the fourth button on the power socket.'\nBegin at [0.45, 0.51] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.51], [0.47, 0.49], [0.47, 0.51], [0.47, 0.51], [0.47, 0.51], [0.47, 0.5], [0.46, 0.46], [0.47, 0.45], [0.48, 0.41], [0.49, 0.41]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0057_user_0007_scene_0002_cfg_0005_036422060215#594.jpg", + "gt": "[[289, 182], [298, 178], [300, 184], [298, 185], [300, 184], [300, 179], [295, 166], [301, 161], [309, 149], [316, 147]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "69300_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the glue stick inside the open drawer', and now you are performing subtask 'put the glue stick inside the open drawer.'\nYour initial coordinate is [0.68, 0.24]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.24], [0.58, 0.22], [0.46, 0.3], [0.38, 0.47], [0.45, 0.41], [0.52, 0.48], [0.49, 0.48], [0.51, 0.49], [0.48, 0.63], [0.54, 0.48]]}" + } + ], + "images": "69300_exterior_image_2_left#0", + "gt": "[[219, 44], [185, 39], [147, 54], [123, 84], [143, 74], [167, 86], [157, 86], [164, 89], [155, 114], [174, 87]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "60211_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the lid on the pot', and now you are performing subtask 'place the lid on the pot.'\nWith start location [0.98, 0.38], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.98, 0.38], [0.77, 0.54], [0.68, 0.71], [0.68, 0.81], [0.76, 0.71], [0.78, 0.49], [0.76, 0.52], [0.79, 0.56], [0.77, 0.57], [0.77, 0.53]]}" + } + ], + "images": "60211_exterior_image_2_left#0", + "gt": "[[314, 69], [245, 97], [219, 127], [218, 145], [243, 127], [251, 89], [242, 93], [254, 100], [246, 102], [245, 96]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0045_user_0010_scene_0004_cfg_0005_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the brush on the pen rack', and now you are performing subtask 'place the brush on the pen rack.'\nThe trajectory starts at [0.47, 0.4]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.4], [0.39, 0.61], [0.37, 0.75], [0.4, 0.34], [0.42, 0.32], [0.38, 0.28], [0.38, 0.32], [0.33, 0.36], [0.44, 0.42], [0.45, 0.39]]}" + } + ], + "images": "RH20T_cfg5_task_0045_user_0010_scene_0004_cfg_0005_037522062165#0", + "gt": "[[298, 144], [251, 220], [237, 271], [256, 124], [269, 114], [242, 100], [241, 116], [210, 128], [282, 151], [290, 142]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "91740_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push the plastic wrap off the container, then pick up the container and move it slightly up onto the table', and now you are performing subtask 'place the container on the plastic wrap.'\nYour initial coordinate is [0.67, 0.62]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.67, 0.62], [0.55, 0.68], [0.52, 0.65], [0.59, 0.64], [0.68, 0.6], [0.72, 0.66], [0.73, 0.62], [0.65, 0.69], [0.59, 0.69], [0.59, 0.67]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/91740_exterior_image_2_left#290.jpg", + "gt": "[[214, 112], [175, 122], [168, 117], [189, 116], [216, 108], [229, 118], [235, 111], [207, 125], [190, 125], [188, 120]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "66111_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the small bottle backward and then place the big bottle beside the small bottle', and now you are performing subtask 'pick up the big bottle on the table.'\nStart at [0.42, 0.54]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.54], [0.41, 0.53], [0.42, 0.54], [0.35, 0.57], [0.32, 0.6], [0.28, 0.74], [0.28, 0.73], [0.3, 0.65], [0.33, 0.64], [0.33, 0.62]]}" + } + ], + "images": "66111_exterior_image_1_left#265", + "gt": "[[135.0, 97.5], [131.0, 95.5], [134.0, 97.5], [111.0, 102.5], [102.0, 108.5], [91.0, 133.5], [90.0, 130.5], [96.0, 117.5], [105.0, 115.5], [107.0, 112.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_104122061018#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the third block on the table, transfer the third block from the table to above the first block, place the third block on top of the first block, pick up the fourth block on the table.'\nYour initial coordinate is [0.55, 0.17]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.17], [0.59, 0.25], [0.59, 0.37], [0.56, 0.43], [0.63, 0.2], [0.58, 0.06], [0.55, 0.11], [0.55, 0.06], [0.56, 0.07], [0.55, 0.14]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_104122061018#613.jpg", + "gt": "[[350, 63], [379, 89], [378, 132], [360, 154], [403, 72], [372, 21], [349, 38], [349, 23], [361, 24], [350, 50]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0029_user_0020_scene_0010_cfg_0007_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp the handle and open the drawer', and now you are performing subtask 'grasp the handle and open the drawer.'\nYour initial coordinate is [0.53, 0.32]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.32], [0.56, 0.37], [0.54, 0.36], [0.54, 0.37], [0.55, 0.36], [0.57, 0.34], [0.61, 0.37], [0.6, 0.55], [0.58, 0.31], [0.57, 0.29]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0029_user_0020_scene_0010_cfg_0007_104122064161#0.jpg", + "gt": "[[337, 115], [357, 133], [347, 130], [348, 134], [351, 128], [363, 123], [389, 133], [385, 199], [373, 111], [368, 105]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0082_user_0007_scene_0002_cfg_0005_104422070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'scoop water with a large spoon from one bowl to another', and now you are performing subtask 'scoop water with a large spoon from one bowl to another.'\nBegin at [0.58, 0.37] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.37], [0.61, 0.37], [0.62, 0.49], [0.63, 0.27], [0.61, 0.25], [0.6, 0.3], [0.63, 0.32], [0.63, 0.43], [0.6, 0.34], [0.6, 0.34]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0082_user_0007_scene_0002_cfg_0005_104422070044#0.jpg", + "gt": "[[374, 132], [391, 133], [398, 177], [403, 96], [392, 91], [384, 108], [403, 115], [403, 153], [384, 124], [384, 124]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "72950_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the yellow pepper from the container', and now you are performing subtask 'remove the yellow pepper from the container.'\nYou start from [0.51, 0.12]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.12], [0.62, 0.24], [0.69, 0.49], [0.65, 0.58], [0.66, 0.61], [0.67, 0.58], [0.63, 0.4], [0.73, 0.65], [0.71, 0.53], [0.64, 0.43]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/72950_exterior_image_1_left#0.jpg", + "gt": "[[162.5, 21.5], [198.5, 42.5], [219.5, 87.5], [207.5, 104.5], [211.5, 109.5], [213.5, 104.5], [201.5, 71.5], [233.5, 116.5], [226.5, 95.5], [205.5, 77.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "54208_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the pen inside the cup', and now you are performing subtask 'place the pen inside the cup.'\nFrom start point [0.46, 0.45], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.45], [0.45, 0.37], [0.38, 0.24], [0.35, 0.27], [0.36, 0.31], [0.34, 0.33], [0.34, 0.39], [0.41, 0.3], [0.45, 0.3], [0.45, 0.3]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/54208_exterior_image_2_left#136.jpg", + "gt": "[[148, 81], [143, 67], [122, 43], [112, 48], [114, 56], [108, 60], [109, 71], [130, 54], [143, 54], [143, 54]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "39944_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the mug to the left, then to the right, then back to the left', and now you are performing subtask 'push the cup to the right.'\nStart at [0.51, 0.17]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.17], [0.52, 0.15], [0.54, 0.19], [0.5, 0.24], [0.5, 0.29], [0.49, 0.28], [0.48, 0.31], [0.51, 0.39], [0.51, 0.32], [0.52, 0.28]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/39944_exterior_image_1_left#190.jpg", + "gt": "[[164, 30], [166, 27], [173, 34], [159, 44], [159, 52], [156, 51], [154, 56], [162, 70], [162, 57], [167, 50]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "28714_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the spoon to stir the contents of the bowl', and now you are performing subtask 'use the spoon to stir the contents of the bowl.'\nThe trajectory starts at [0.34, 0.46]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.34, 0.46], [0.47, 0.51], [0.51, 0.72], [0.51, 0.6], [0.53, 0.59], [0.52, 0.64], [0.54, 0.61], [0.52, 0.58], [0.52, 0.63], [0.46, 0.52]]}" + } + ], + "images": "28714_exterior_image_2_left#0", + "gt": "[[108, 83], [150, 92], [164, 129], [162, 108], [171, 107], [166, 116], [173, 110], [165, 105], [165, 114], [146, 94]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "57005_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the marker from beneath the table to above the table', and now you are performing subtask 'move the marker from beneath the table to above the table.'\nYour initial coordinate is [0.1, 0.24]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.1, 0.24], [0.17, 0.33], [0.2, 0.43], [0.19, 0.52], [0.2, 0.49], [0.22, 0.53], [0.31, 0.71], [0.33, 0.59], [0.36, 0.61], [0.33, 0.54]]}" + } + ], + "images": "57005_exterior_image_2_left#0", + "gt": "[[31, 43], [56, 60], [63, 77], [62, 94], [63, 89], [72, 95], [100, 127], [107, 106], [116, 110], [107, 98]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0005_user_0008_scene_0001_cfg_0001_819112071037#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'approach and touch the side of the small block', and now you are performing subtask 'approach and touch the side of the small block.'\nYou start from [0.51, 0.0]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.0], [0.5, 0.0], [0.47, 0.22], [0.44, 0.28], [0.21, 0.42], [0.18, 0.39], [0.16, 0.41], [0.2, 0.36], [0.31, 0.27], [0.35, 0.19]]}" + } + ], + "images": "RH20T_cfg1_task_0005_user_0008_scene_0001_cfg_0001_819112071037#0", + "gt": "[[326, 0], [320, 1], [302, 78], [283, 101], [137, 150], [118, 142], [104, 148], [131, 128], [197, 97], [224, 67]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "21213_exterior_image_2_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the two paper bags out of the wooden crate', and now you are performing subtask 'place the second paper bag on the table.'\nYou begin at [0.51, 0.62]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.62], [0.52, 0.69], [0.52, 0.55], [0.51, 0.53], [0.48, 0.54], [0.47, 0.57], [0.48, 0.64], [0.5, 0.56], [0.51, 0.59], [0.46, 0.7]]}" + } + ], + "images": "21213_exterior_image_2_left#378", + "gt": "[[164, 112], [167, 125], [167, 99], [162, 96], [154, 97], [151, 102], [153, 115], [161, 100], [162, 107], [147, 126]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "72353_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the towel', and now you are performing subtask 'move the towel slightly backward.'\nYour starting position is [0.66, 0.65]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.66, 0.65], [0.66, 0.68], [0.61, 0.78], [0.52, 0.7], [0.57, 0.67], [0.61, 0.61], [0.61, 0.59], [0.59, 0.58], [0.6, 0.55], [0.59, 0.56]]}" + } + ], + "images": "72353_exterior_image_1_left#299", + "gt": "[[210, 117], [210, 123], [196, 140], [166, 126], [181, 120], [194, 109], [194, 106], [188, 104], [191, 99], [189, 101]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "58422_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the marker on the table', and now you are performing subtask 'place the marker on the table.'\nStarting at [0.47, 0.04], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.04], [0.48, 0.24], [0.52, 0.42], [0.51, 0.45], [0.53, 0.33], [0.57, 0.35], [0.58, 0.4], [0.57, 0.57], [0.55, 0.38], [0.48, 0.31]]}" + } + ], + "images": "58422_exterior_image_1_left#0", + "gt": "[[151, 8], [154, 43], [168, 75], [164, 81], [169, 60], [181, 63], [185, 72], [183, 102], [175, 68], [155, 56]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "29599_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the towel on the table, pull the towel to the right side of the table', and now you are performing subtask 'pull the towel to the right side of the table.'\nStarting at [0.72, 0.74], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.72, 0.74], [0.59, 0.72], [0.55, 0.8], [0.54, 0.64], [0.48, 0.65], [0.5, 0.56], [0.45, 0.53], [0.41, 0.42], [0.4, 0.48], [0.49, 0.57]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/29599_exterior_image_1_left#183.jpg", + "gt": "[[231, 134], [190, 129], [176, 144], [174, 115], [153, 117], [161, 100], [144, 96], [131, 76], [129, 86], [158, 103]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "50798_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the black spoon in the cutlery stand', and now you are performing subtask 'put the black spoon in the cutlery stand.'\nYour initial coordinate is [0.35, 0.38]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.38], [0.38, 0.33], [0.51, 0.54], [0.5, 0.34], [0.51, 0.36], [0.57, 0.24], [0.7, 0.17], [0.73, 0.21], [0.74, 0.28], [0.67, 0.14]]}" + } + ], + "images": "50798_exterior_image_1_left#0", + "gt": "[[111, 68], [120, 59], [164, 97], [159, 61], [164, 64], [182, 44], [223, 31], [235, 37], [236, 50], [215, 25]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0094_user_0007_scene_0006_cfg_0005_104122063678#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'press the part of the box, press the part of the box.'\nBegin at [0.59, 0.55] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.55], [0.58, 0.56], [0.56, 0.57], [0.55, 0.51], [0.54, 0.5], [0.52, 0.51], [0.5, 0.55], [0.52, 0.52], [0.54, 0.52], [0.57, 0.55]]}" + } + ], + "images": "RH20T_cfg5_task_0094_user_0007_scene_0006_cfg_0005_104122063678#290", + "gt": "[[375, 198], [369, 203], [359, 204], [353, 185], [344, 181], [332, 185], [321, 197], [335, 188], [345, 186], [362, 197]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0034_user_0007_scene_0005_cfg_0005_105422061350#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares in a row', and now you are performing subtask 'place the second square on the table next to the first square, pick up the third square from the table.'\nYou start from [0.37, 0.36]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.36], [0.37, 0.33], [0.39, 0.32], [0.4, 0.28], [0.47, 0.19], [0.52, 0.15], [0.51, 0.21], [0.5, 0.19], [0.43, 0.19], [0.44, 0.22]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0034_user_0007_scene_0005_cfg_0005_105422061350#571.jpg", + "gt": "[[236, 129], [237, 118], [247, 114], [257, 100], [303, 68], [333, 55], [326, 74], [322, 68], [275, 69], [280, 78]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_038522063145#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the third block on top of the second block, pick up the fourth block from the surface, transfer the fourth block from the desktop to the second position above the third block, place the fourth block on top of the third block.'\nYou begin at [0.72, 0.14]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.72, 0.14], [0.71, 0.07], [0.73, 0.04], [0.68, 0.27], [0.69, 0.12], [0.68, 0.02], [0.72, 0.03], [0.73, 0.1], [0.74, 0.04], [0.74, 0.04]]}" + } + ], + "images": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_038522063145#524", + "gt": "[[459, 52], [454, 24], [467, 15], [432, 96], [443, 43], [433, 7], [458, 10], [467, 36], [474, 16], [474, 16]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0053_user_0014_scene_0003_cfg_0001_039422060546#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use a clamp', and now you are performing subtask 'use a clamp.'\nFrom start point [0.41, 0.34], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.34], [0.45, 0.32], [0.47, 0.17], [0.52, 0.07], [0.49, 0.17], [0.51, 0.22], [0.49, 0.26], [0.48, 0.22], [0.5, 0.28], [0.54, 0.1]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0053_user_0014_scene_0003_cfg_0001_039422060546#0.jpg", + "gt": "[[265, 123], [286, 117], [300, 61], [336, 26], [316, 61], [324, 81], [316, 92], [309, 80], [320, 99], [348, 36]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "15117_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick the black lid and put it on the bottle', and now you are performing subtask 'pick the black lid and put it on the bottle.'\nWith start location [0.98, 0.01], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.98, 0.01], [0.75, 0.44], [0.67, 0.49], [0.65, 0.32], [0.61, 0.29], [0.55, 0.3], [0.59, 0.25], [0.57, 0.34], [0.58, 0.34], [0.59, 0.29]]}" + } + ], + "images": "15117_exterior_image_1_left#0", + "gt": "[[313.5, 1.5], [239.5, 79.5], [213.5, 87.5], [208.5, 58.5], [195.5, 51.5], [174.5, 54.5], [187.5, 45.5], [181.5, 60.5], [186.5, 61.5], [188.5, 51.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "84593_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the marker from the cup and put it on the table', and now you are performing subtask 'take the marker from the cup and put it on the table.'\nFrom start point [0.93, 0.0], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.93, 0.0], [0.76, 0.12], [0.76, 0.21], [0.66, 0.33], [0.7, 0.42], [0.66, 0.21], [0.57, 0.29], [0.55, 0.42], [0.64, 0.57], [0.64, 0.57]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/84593_exterior_image_1_left#0.jpg", + "gt": "[[297.5, 0.0], [243.5, 22.0], [244.5, 37.0], [212.5, 60.0], [225.5, 76.0], [210.5, 38.0], [182.5, 53.0], [175.5, 76.0], [203.5, 103.0], [203.5, 103.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "54770_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the charger to the left then move part of the cable to the right', and now you are performing subtask 'push a part of the cable to the right.'\nYou start from [0.51, 0.42]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.42], [0.47, 0.43], [0.48, 0.48], [0.53, 0.43], [0.59, 0.48], [0.57, 0.42], [0.58, 0.39], [0.71, 0.48], [0.66, 0.44], [0.67, 0.35]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/54770_exterior_image_1_left#115.jpg", + "gt": "[[162, 75], [151, 77], [154, 87], [171, 78], [189, 86], [183, 75], [187, 70], [227, 86], [212, 80], [215, 63]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "18085_exterior_image_1_left#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'folded shirt', and now you are performing subtask 'fold the shirt on the table.'\nYour starting position is [0.51, 0.66]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.66], [0.53, 0.74], [0.51, 0.83], [0.47, 0.59], [0.42, 0.54], [0.32, 0.58], [0.32, 0.73], [0.33, 0.69], [0.31, 0.63], [0.36, 0.56]]}" + } + ], + "images": "18085_exterior_image_1_left#383", + "gt": "[[163, 119], [170, 133], [163, 150], [149, 107], [135, 98], [103, 104], [101, 131], [105, 124], [100, 113], [114, 101]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0012_user_0010_scene_0005_cfg_0004_104122062823#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push the soccer ball into the goal', and now you are performing subtask 'push the soccer ball toward the goal.'\nThe trajectory starts at [0.34, 0.31]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.34, 0.31], [0.36, 0.33], [0.37, 0.37], [0.35, 0.36], [0.34, 0.37], [0.35, 0.38], [0.35, 0.38], [0.35, 0.35], [0.37, 0.35], [0.39, 0.39]]}" + } + ], + "images": "RH20T_cfg4_task_0012_user_0010_scene_0005_cfg_0004_104122062823#214", + "gt": "[[220, 111], [230, 118], [237, 133], [221, 128], [216, 134], [223, 138], [225, 135], [221, 126], [237, 127], [248, 141]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "86836_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the towel', and now you are performing subtask 'fold the towel on the table.'\nYou begin at [0.4, 0.94]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.94], [0.39, 0.99], [0.35, 1.0], [0.38, 0.77], [0.39, 0.61], [0.39, 0.65], [0.42, 0.67], [0.42, 0.7], [0.43, 0.65], [0.42, 0.62]]}" + } + ], + "images": "86836_exterior_image_1_left#383", + "gt": "[[127, 170], [125, 179], [112, 180], [123, 138], [125, 110], [124, 117], [134, 121], [134, 126], [138, 117], [135, 111]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0081_user_0010_scene_0001_cfg_0001_039422060546#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the leg of the glasses (with one hand)', and now you are performing subtask 'pull the leg of the glasses outward, pull the other leg of the glasses outward.'\nStart at [0.51, 0.24]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.24], [0.45, 0.26], [0.48, 0.36], [0.5, 0.35], [0.5, 0.29], [0.45, 0.27], [0.45, 0.24], [0.55, 0.26], [0.54, 0.31], [0.52, 0.3]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0081_user_0010_scene_0001_cfg_0001_039422060546#228.jpg", + "gt": "[[328, 86], [288, 94], [310, 129], [323, 125], [319, 104], [291, 97], [291, 88], [351, 94], [346, 111], [333, 109]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0081_user_0010_scene_0001_cfg_0001_035622060973#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the leg of the glasses (with one hand)', and now you are performing subtask 'pull the leg of the glasses outward.'\nThe trajectory starts at [0.42, 0.49]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.49], [0.43, 0.44], [0.43, 0.46], [0.42, 0.48], [0.42, 0.51], [0.42, 0.54], [0.42, 0.53], [0.4, 0.54], [0.4, 0.56], [0.41, 0.55]]}" + } + ], + "images": "RH20T_cfg1_task_0081_user_0010_scene_0001_cfg_0001_035622060973#203", + "gt": "[[268, 176], [276, 157], [275, 165], [271, 172], [271, 182], [267, 193], [269, 191], [259, 196], [255, 202], [260, 198]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "26549_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the brown coffee cup from the book and put it inside the brown coffee cup', and now you are performing subtask 'remove the brown coffee cup from the book and put it inside the brown coffee cup.'\nBegin at [0.41, 0.36] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.36], [0.48, 0.39], [0.52, 0.43], [0.52, 0.46], [0.57, 0.61], [0.56, 0.68], [0.57, 0.53], [0.56, 0.41], [0.57, 0.48], [0.57, 0.42]]}" + } + ], + "images": "26549_exterior_image_2_left#0", + "gt": "[[132, 65], [154, 70], [166, 78], [165, 83], [183, 110], [180, 123], [184, 95], [180, 74], [183, 87], [184, 76]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0065_user_0014_scene_0002_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with small Lego blocks for the nth time with n blocks', and now you are performing subtask 'pick up the pink block on the desktop, transfer the pink block from the desktop to the building area, place the pink block in the building area.'\nYour initial coordinate is [0.63, 0.37]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.61, 0.38], [0.6, 0.46], [0.62, 0.53], [0.64, 0.6], [0.65, 0.59], [0.67, 0.66], [0.64, 0.58], [0.61, 0.56], [0.6, 0.58]]}" + } + ], + "images": "RH20T_cfg1_task_0065_user_0014_scene_0002_cfg_0001_038522062547#0", + "gt": "[[405, 132], [388, 138], [387, 165], [394, 190], [410, 216], [417, 213], [427, 239], [407, 209], [391, 203], [382, 210]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0037_user_0017_scene_0006_cfg_0004_104122062823#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the fourth block on top of the third block on the right side of the table, pick up the fifth block from the left side of the table, place the fifth block on top of the fourth block on the right side of the table, transfer the fifth block from the left side of the table to the right side of the table.'\nYou start from [0.48, 0.16]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.16], [0.43, 0.09], [0.4, 0.11], [0.41, 0.17], [0.46, 0.09], [0.48, 0.09], [0.48, 0.12], [0.52, 0.1], [0.51, 0.06], [0.52, 0.08]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0037_user_0017_scene_0006_cfg_0004_104122062823#400.jpg", + "gt": "[[307, 56], [274, 32], [254, 41], [265, 61], [292, 33], [308, 32], [308, 45], [332, 37], [325, 22], [333, 29]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0026_user_0016_scene_0001_cfg_0004_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the gripper to push and close the drawer', and now you are performing subtask 'use the gripper to push and close the drawer.'\nWith start location [0.45, 0.0], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.0], [0.36, 0.0], [0.32, 0.02], [0.29, 0.01], [0.29, 0.02], [0.27, 0.04], [0.31, 0.03], [0.33, 0.0], [0.42, 0.0], [0.42, 0.0]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0026_user_0016_scene_0001_cfg_0004_104122062823#0.jpg", + "gt": "[[290, 0], [228, 0], [202, 7], [188, 5], [183, 6], [173, 15], [196, 12], [212, 1], [266, 0], [266, 0]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "47788_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the tap', and now you are performing subtask 'press the handle of the tap to close the tap.'\nStart at [0.26, 0.24]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.26, 0.24], [0.25, 0.24], [0.24, 0.25], [0.24, 0.27], [0.26, 0.28], [0.27, 0.31], [0.28, 0.33], [0.3, 0.26], [0.31, 0.25], [0.36, 0.27]]}" + } + ], + "images": "47788_exterior_image_1_left#156", + "gt": "[[84, 44], [79, 43], [76, 45], [76, 48], [83, 50], [85, 55], [90, 59], [96, 46], [99, 45], [116, 49]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "45897_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push the box into the cabinet', and now you are performing subtask 'push the box into the cabinet.'\nThe trajectory starts at [0.73, 0.67]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.73, 0.67], [0.68, 0.68], [0.67, 0.72], [0.65, 0.74], [0.64, 0.79], [0.61, 0.78], [0.52, 0.78], [0.49, 0.76], [0.55, 0.72], [0.66, 0.73]]}" + } + ], + "images": "45897_exterior_image_1_left#0", + "gt": "[[234, 120], [216, 123], [215, 130], [209, 133], [205, 143], [195, 140], [167, 140], [157, 136], [177, 130], [211, 131]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "79943_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the glass lid on the pot', and now you are performing subtask 'put the glass lid on the pot.'\nFrom start point [0.14, 0.13], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.14, 0.13], [0.39, 0.43], [0.44, 0.51], [0.43, 0.56], [0.42, 0.43], [0.45, 0.43], [0.51, 0.51], [0.51, 0.56], [0.5, 0.57], [0.49, 0.5]]}" + } + ], + "images": "79943_exterior_image_1_left#0", + "gt": "[[44.5, 23.5], [125.5, 76.5], [139.5, 92.5], [137.5, 100.5], [135.5, 77.5], [144.5, 76.5], [164.5, 91.5], [164.5, 101.5], [158.5, 102.5], [156.5, 90.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "78177_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the door', and now you are performing subtask 'open the door.'\nFrom start point [0.5, 0.51], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.51], [0.51, 0.44], [0.64, 0.36], [0.6, 0.42], [0.6, 0.47], [0.63, 0.47], [0.6, 0.47], [0.58, 0.41], [0.58, 0.36], [0.61, 0.33]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/78177_exterior_image_2_left#0.jpg", + "gt": "[[161, 92], [164, 79], [205, 65], [191, 75], [193, 85], [202, 85], [192, 84], [186, 74], [186, 64], [194, 60]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0073_user_0001_scene_0003_cfg_0001_035622060973#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the things in the drawer', and now you are performing subtask 'push the drawer handle to close the drawer.'\nThe trajectory starts at [0.43, 0.33]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.33], [0.44, 0.35], [0.46, 0.35], [0.46, 0.36], [0.47, 0.35], [0.46, 0.36], [0.44, 0.32], [0.42, 0.24], [0.41, 0.26], [0.38, 0.3]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0073_user_0001_scene_0003_cfg_0001_035622060973#316.jpg", + "gt": "[[278, 119], [282, 125], [294, 127], [293, 131], [299, 126], [297, 131], [283, 117], [266, 87], [265, 92], [243, 108]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0110_user_0007_scene_0009_cfg_0004_036422060909#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the pencil out from the pencil sharpener', and now you are performing subtask 'take the pencil out from the pencil sharpener.'\nYour initial coordinate is [0.53, 0.22]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.22], [0.51, 0.2], [0.48, 0.24], [0.46, 0.32], [0.47, 0.36], [0.45, 0.42], [0.53, 0.43], [0.57, 0.39], [0.54, 0.29], [0.52, 0.28]]}" + } + ], + "images": "RH20T_cfg4_task_0110_user_0007_scene_0009_cfg_0004_036422060909#0", + "gt": "[[341, 79], [324, 71], [307, 86], [294, 117], [302, 129], [287, 150], [341, 156], [362, 142], [345, 106], [330, 102]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "44786_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the top drawer of the white file cabinet', and now you are performing subtask 'close the top drawer of the white file cabinet.'\nYour initial coordinate is [0.84, 0.82]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.84, 0.82], [0.8, 0.78], [0.72, 0.67], [0.63, 0.6], [0.6, 0.57], [0.59, 0.55], [0.57, 0.56], [0.53, 0.55], [0.53, 0.56], [0.53, 0.56]]}" + } + ], + "images": "44786_exterior_image_1_left#0", + "gt": "[[268, 147], [256, 141], [230, 120], [203, 108], [192, 102], [190, 99], [181, 100], [171, 99], [170, 100], [171, 100]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg3_task_0057_user_0010_scene_0007_cfg_0003_104122062823#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn on the power strips', and now you are performing subtask 'press the button of the power socket in the third position, press the button of the power socket in the fourth position.'\nYou begin at [0.3, 0.28]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.3, 0.28], [0.31, 0.26], [0.32, 0.25], [0.32, 0.24], [0.32, 0.25], [0.31, 0.26], [0.3, 0.22], [0.3, 0.19], [0.34, 0.13], [0.35, 0.15]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg3_task_0057_user_0010_scene_0007_cfg_0003_104122062823#219.jpg", + "gt": "[[194, 100], [197, 95], [208, 91], [204, 86], [205, 91], [201, 92], [192, 79], [191, 67], [216, 46], [224, 54]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_036422060215#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the fourth block on top of the third block, pick up the fifth block on the table, transfer the fifth block from the table to the position of the fourth block, place the fifth block on top of the fourth block.'\nYour starting position is [0.49, 0.26]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.26], [0.51, 0.25], [0.43, 0.41], [0.42, 0.47], [0.42, 0.51], [0.46, 0.36], [0.51, 0.29], [0.52, 0.21], [0.51, 0.18], [0.53, 0.16]]}" + } + ], + "images": "RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_036422060215#859", + "gt": "[[316, 93], [324, 89], [278, 149], [266, 171], [266, 184], [295, 131], [326, 104], [334, 76], [325, 65], [342, 57]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "69546_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the cloth on the table', and now you are performing subtask 'put the cloth on the table.'\nWith start location [0.66, 0.47], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.66, 0.47], [0.58, 0.54], [0.69, 0.52], [0.77, 0.61], [0.81, 0.68], [0.82, 0.74], [0.79, 0.73], [0.76, 0.68], [0.73, 0.61], [0.72, 0.53]]}" + } + ], + "images": "69546_exterior_image_1_left#0", + "gt": "[[212.0, 85.5], [187.0, 96.5], [222.0, 94.5], [246.0, 109.5], [259.0, 122.5], [261.0, 133.5], [252.0, 130.5], [242.0, 121.5], [234.0, 110.5], [229.0, 95.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0006_user_0010_scene_0008_cfg_0004_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push the small block from the right side of the table to the left side', and now you are performing subtask 'push the small block from the right side of the table to the left side, push the small block from the right side of the table to the left side.'\nFrom start point [0.38, 0.0], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.47, 0.0], [0.63, 0.09], [0.62, 0.11], [0.73, 0.2], [0.8, 0.3], [0.78, 0.34], [0.65, 0.26], [0.59, 0.26], [0.59, 0.26]]}" + } + ], + "images": "RH20T_cfg4_task_0006_user_0010_scene_0008_cfg_0004_104122062823#0", + "gt": "[[243, 0], [301, 1], [401, 32], [398, 40], [467, 71], [510, 107], [500, 124], [413, 94], [376, 94], [376, 94]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0038_user_0015_scene_0010_cfg_0001_819112070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the cup', and now you are performing subtask 'pick up the cup.'\nStarting at [0.62, 0.46], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.46], [0.59, 0.44], [0.61, 0.48], [0.59, 0.49], [0.58, 0.52], [0.57, 0.51], [0.59, 0.45], [0.65, 0.37], [0.65, 0.29], [0.64, 0.28]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0038_user_0015_scene_0010_cfg_0001_819112070044#0.jpg", + "gt": "[[400, 167], [377, 159], [388, 172], [376, 178], [369, 188], [367, 183], [380, 162], [414, 133], [418, 106], [408, 101]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_038522063145#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the first block on top of the second block, pick up the third block from the table, transfer the third block from the table to the position above the fourth block, place the third block on top of the second block.'\nStarting at [0.57, 0.22], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.22], [0.62, 0.21], [0.65, 0.26], [0.67, 0.3], [0.68, 0.37], [0.66, 0.39], [0.67, 0.41], [0.66, 0.23], [0.58, 0.16], [0.58, 0.18]]}" + } + ], + "images": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_038522063145#228", + "gt": "[[365, 80], [397, 76], [419, 92], [430, 109], [432, 134], [423, 140], [430, 149], [421, 84], [371, 56], [371, 66]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0107_user_0015_scene_0009_cfg_0004_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the garbage in the trash can', and now you are performing subtask 'pick up the garbage from the desktop, transfer the garbage from the desktop to the trash can, place the garbage in the trash can.'\nBegin at [0.45, 0.0] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.0], [0.41, 0.0], [0.48, 0.19], [0.5, 0.18], [0.5, 0.23], [0.48, 0.11], [0.42, 0.0], [0.35, 0.0], [0.4, 0.02], [0.37, 0.0]]}" + } + ], + "images": "RH20T_cfg4_task_0107_user_0015_scene_0009_cfg_0004_104122062823#0", + "gt": "[[289, 0], [265, 0], [306, 70], [319, 64], [318, 84], [310, 39], [271, 0], [221, 0], [257, 8], [235, 0]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0003_user_0016_scene_0003_cfg_0003_036422060909#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the three buttons in sequence', and now you are performing subtask 'press the gray base red button, press the yellow base green button, press the orange base red button.'\nYou start from [0.62, 0.29]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.29], [0.62, 0.3], [0.62, 0.29], [0.62, 0.28], [0.53, 0.32], [0.52, 0.33], [0.52, 0.37], [0.54, 0.33], [0.52, 0.34], [0.49, 0.39]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg3_task_0003_user_0016_scene_0003_cfg_0003_036422060909#256.jpg", + "gt": "[[396, 104], [399, 107], [399, 103], [394, 101], [340, 116], [334, 120], [336, 132], [346, 120], [335, 122], [315, 141]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0065_user_0007_scene_0010_cfg_0002_105422061350#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with small Lego blocks for the nth time with n blocks', and now you are performing subtask 'pick up the black Lego block from the table, transfer the black Lego block from the table to above the building area, place the black Lego block in the building area, press the black Lego block in the building area, pick up the white Lego block from the table.'\nStarting at [0.3, 0.31], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.3, 0.31], [0.29, 0.33], [0.38, 0.3], [0.4, 0.32], [0.4, 0.36], [0.4, 0.33], [0.39, 0.32], [0.38, 0.35], [0.35, 0.28], [0.32, 0.24]]}" + } + ], + "images": "RH20T_cfg2_task_0065_user_0007_scene_0010_cfg_0002_105422061350#492", + "gt": "[[189, 112], [183, 118], [242, 108], [257, 115], [257, 129], [257, 119], [248, 117], [242, 126], [226, 100], [202, 85]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0125_user_0007_scene_0004_cfg_0004_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull up the lever', and now you are performing subtask 'pull up the lever.'\nThe trajectory starts at [0.45, 0.0]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.0], [0.45, 0.25], [0.46, 0.09], [0.43, 0.1], [0.35, 0.02], [0.46, 0.11], [0.37, 0.07], [0.29, 0.09], [0.43, 0.02], [0.45, 0.08]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0125_user_0007_scene_0004_cfg_0004_104122062823#0.jpg", + "gt": "[[289, 0], [291, 91], [293, 31], [277, 35], [222, 6], [294, 41], [237, 25], [187, 34], [276, 7], [290, 28]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0035_user_0010_scene_0001_cfg_0003_104122062295#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up a small block', and now you are performing subtask 'pick up a small block.'\nFrom start point [0.6, 0.19], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.19], [0.61, 0.33], [0.57, 0.29], [0.63, 0.29], [0.53, 0.27], [0.4, 0.16], [0.34, 0.33], [0.39, 0.29], [0.42, 0.39], [0.4, 0.21]]}" + } + ], + "images": "RH20T_cfg3_task_0035_user_0010_scene_0001_cfg_0003_104122062295#0", + "gt": "[[381, 70], [389, 120], [365, 104], [402, 106], [341, 98], [258, 59], [215, 118], [252, 104], [271, 139], [256, 76]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0103_user_0010_scene_0005_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the plate to the goal post', and now you are performing subtask 'move the plate to the goal post.'\nWith start location [0.63, 0.35], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.35], [0.62, 0.5], [0.59, 0.57], [0.61, 0.56], [0.63, 0.49], [0.65, 0.19], [0.61, 0.12], [0.66, 0.18], [0.68, 0.32], [0.71, 0.32]]}" + } + ], + "images": "RH20T_cfg1_task_0103_user_0010_scene_0005_cfg_0001_038522062547#0", + "gt": "[[404, 127], [395, 181], [379, 207], [392, 203], [401, 177], [415, 67], [391, 44], [420, 66], [435, 116], [455, 117]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "71329_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker in the mug and then remove it and put it back on the table', and now you are performing subtask 'place the marker in the mug.'\nWith start location [0.24, 0.68], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.24, 0.68], [0.25, 0.58], [0.29, 0.46], [0.31, 0.36], [0.33, 0.33], [0.39, 0.37], [0.4, 0.39], [0.41, 0.42], [0.4, 0.39], [0.4, 0.39]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/71329_exterior_image_2_left#121.jpg", + "gt": "[[78, 122], [80, 105], [94, 82], [100, 64], [107, 60], [124, 66], [129, 71], [130, 76], [129, 71], [129, 71]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0047_user_0020_scene_0004_cfg_0007_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the cup off the cup rack', and now you are performing subtask 'take the cup off the cup rack.'\nStarting at [0.53, 0.32], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.32], [0.58, 0.36], [0.6, 0.32], [0.62, 0.34], [0.63, 0.41], [0.63, 0.31], [0.6, 0.24], [0.59, 0.28], [0.59, 0.21], [0.58, 0.23]]}" + } + ], + "images": "RH20T_cfg7_task_0047_user_0020_scene_0004_cfg_0007_104122064161#0", + "gt": "[[337, 115], [369, 131], [381, 114], [398, 124], [403, 146], [406, 111], [382, 85], [378, 99], [379, 75], [372, 83]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "84471_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the lid from the pot and place it on the table', and now you are performing subtask 'remove the lid from the pot and place it on the table.'\nFrom start point [0.02, 0.16], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.02, 0.16], [0.2, 0.29], [0.32, 0.41], [0.39, 0.49], [0.43, 0.59], [0.41, 0.52], [0.37, 0.47], [0.34, 0.46], [0.37, 0.53], [0.32, 0.49]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/84471_exterior_image_1_left#0.jpg", + "gt": "[[7, 29], [64, 52], [104, 73], [125, 89], [139, 106], [131, 93], [117, 84], [109, 83], [118, 96], [103, 89]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0015_user_0014_scene_0005_cfg_0006_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'play the drum', and now you are performing subtask 'play the drum.'\nWith start location [0.53, 0.31], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.31], [0.62, 0.43], [0.63, 0.56], [0.61, 0.59], [0.6, 0.46], [0.57, 0.33], [0.55, 0.31], [0.57, 0.38], [0.55, 0.3], [0.55, 0.3]]}" + } + ], + "images": "RH20T_cfg6_task_0015_user_0014_scene_0005_cfg_0006_104122064161#0", + "gt": "[[338, 110], [395, 155], [402, 200], [392, 211], [383, 166], [365, 119], [355, 113], [363, 138], [350, 109], [354, 108]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "73339_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker in the mug', and now you are performing subtask 'put the marker in the mug.'\nYou begin at [0.81, 0.16]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.81, 0.16], [0.84, 0.65], [0.78, 0.74], [0.81, 0.7], [0.81, 0.61], [0.75, 0.54], [0.72, 0.57], [0.73, 0.64], [0.73, 0.58], [0.78, 0.53]]}" + } + ], + "images": "73339_exterior_image_1_left#0", + "gt": "[[258.0, 28.5], [269.0, 116.5], [249.0, 133.5], [260.0, 126.5], [260.0, 109.5], [240.0, 97.5], [231.0, 102.5], [234.0, 115.5], [234.0, 104.5], [249.0, 95.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0104_user_0007_scene_0007_cfg_0005_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'drag the plate back after holding it down', and now you are performing subtask 'drag the plate back after holding it down.'\nYour initial coordinate is [0.45, 0.3]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.3], [0.44, 0.25], [0.45, 0.37], [0.44, 0.38], [0.44, 0.42], [0.42, 0.43], [0.44, 0.4], [0.44, 0.42], [0.39, 0.42], [0.43, 0.31]]}" + } + ], + "images": "RH20T_cfg5_task_0104_user_0007_scene_0007_cfg_0005_036422060215#0", + "gt": "[[286, 108], [280, 90], [287, 133], [280, 136], [281, 150], [270, 154], [282, 145], [283, 150], [249, 152], [277, 113]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0033_user_0020_scene_0001_cfg_0007_104122061330#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pushing another small block to the designated position while grasping one small block', and now you are performing subtask 'pick up the front square from the cardboard, transfer the front square from the position to the position aligned with the second square, place the front square in the position aligned with the second square.'\nYou begin at [0.42, 0.29]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.29], [0.45, 0.28], [0.47, 0.31], [0.54, 0.2], [0.58, 0.17], [0.62, 0.22], [0.61, 0.27], [0.62, 0.26], [0.63, 0.28], [0.63, 0.29]]}" + } + ], + "images": "RH20T_cfg7_task_0033_user_0020_scene_0001_cfg_0007_104122061330#0", + "gt": "[[266, 103], [288, 100], [303, 113], [345, 72], [372, 63], [396, 79], [392, 98], [399, 95], [405, 102], [403, 106]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0031_user_0020_scene_0007_cfg_0007_104122063633#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'cover the box', and now you are performing subtask 'cover the box.'\nStarting at [0.37, 0.34], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.34], [0.33, 0.41], [0.33, 0.39], [0.31, 0.39], [0.31, 0.43], [0.3, 0.35], [0.3, 0.41], [0.27, 0.59], [0.3, 0.41], [0.32, 0.38]]}" + } + ], + "images": "RH20T_cfg7_task_0031_user_0020_scene_0007_cfg_0007_104122063633#0", + "gt": "[[234, 121], [212, 149], [210, 141], [199, 141], [200, 155], [192, 127], [192, 147], [170, 211], [192, 148], [202, 138]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0058_user_0010_scene_0010_cfg_0003_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn off the power strips', and now you are performing subtask 'press the buttons on a row of four power strips to turn off the power strips.'\nBegin at [0.38, 0.0] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.31, 0.0], [0.34, 0.02], [0.35, 0.14], [0.4, 0.04], [0.47, 0.1], [0.47, 0.14], [0.49, 0.09], [0.49, 0.11], [0.46, 0.01]]}" + } + ], + "images": "RH20T_cfg3_task_0058_user_0010_scene_0010_cfg_0003_104122062823#0", + "gt": "[[246, 0], [196, 0], [219, 8], [224, 50], [254, 13], [299, 35], [300, 51], [315, 31], [311, 41], [294, 4]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0070_user_0014_scene_0008_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a sliding window', and now you are performing subtask 'open a sliding window.'\nYour starting position is [0.63, 0.37]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.67, 0.4], [0.68, 0.43], [0.69, 0.42], [0.69, 0.45], [0.7, 0.43], [0.69, 0.5], [0.68, 0.33], [0.66, 0.28], [0.64, 0.28]]}" + } + ], + "images": "RH20T_cfg1_task_0070_user_0014_scene_0008_cfg_0001_038522062547#0", + "gt": "[[404, 132], [431, 145], [434, 155], [439, 151], [441, 162], [445, 155], [443, 179], [438, 120], [423, 99], [409, 101]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0032_user_0016_scene_0002_cfg_0004_038522062288#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'slide the outer casing onto the gift box', and now you are performing subtask 'slide the outer casing onto the gift box.'\nYour starting position is [0.38, 0.66]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.66], [0.4, 0.67], [0.41, 0.66], [0.37, 0.62], [0.31, 0.64], [0.32, 0.67], [0.32, 0.64], [0.34, 0.61], [0.41, 0.58], [0.43, 0.59]]}" + } + ], + "images": "RH20T_cfg4_task_0032_user_0016_scene_0002_cfg_0004_038522062288#175", + "gt": "[[240, 236], [259, 240], [264, 236], [235, 225], [198, 231], [205, 241], [204, 230], [218, 221], [264, 210], [276, 214]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "20030_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer', and now you are performing subtask 'open the drawer.'\nYou begin at [0.42, 0.0]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.0], [0.44, 0.41], [0.46, 0.47], [0.5, 0.53], [0.52, 0.52], [0.47, 0.54], [0.43, 0.59], [0.41, 0.49], [0.41, 0.47], [0.43, 0.51]]}" + } + ], + "images": "20030_exterior_image_1_left#0", + "gt": "[[133, 0], [142, 74], [147, 84], [160, 96], [165, 94], [152, 98], [138, 107], [132, 88], [132, 85], [136, 91]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0015_user_0020_scene_0007_cfg_0007_104122063633#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'play the drum', and now you are performing subtask 'play the drum.'\nWith start location [0.37, 0.34], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.34], [0.35, 0.51], [0.29, 0.64], [0.33, 0.52], [0.31, 0.41], [0.29, 0.45], [0.3, 0.4], [0.27, 0.56], [0.28, 0.69], [0.34, 0.32]]}" + } + ], + "images": "RH20T_cfg7_task_0015_user_0020_scene_0007_cfg_0007_104122063633#0", + "gt": "[[234, 121], [222, 183], [184, 231], [212, 187], [201, 149], [187, 163], [191, 144], [170, 203], [179, 247], [217, 115]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0038_user_0020_scene_0001_cfg_0006_104122061018#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the cup', and now you are performing subtask 'pick up the cup.'\nStarting at [0.57, 0.08], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.08], [0.55, 0.15], [0.6, 0.27], [0.57, 0.32], [0.48, 0.37], [0.47, 0.39], [0.5, 0.4], [0.48, 0.4], [0.5, 0.28], [0.51, 0.32]]}" + } + ], + "images": "RH20T_cfg6_task_0038_user_0020_scene_0001_cfg_0006_104122061018#0", + "gt": "[[367, 28], [355, 53], [382, 96], [366, 117], [308, 132], [304, 140], [319, 143], [305, 143], [322, 102], [329, 116]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "17163_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick the marker and put it in the cup', and now you are performing subtask 'pick the marker and put it in the cup.'\nStarting at [0.71, 0.57], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.71, 0.57], [0.63, 0.57], [0.62, 0.63], [0.63, 0.59], [0.61, 0.53], [0.56, 0.56], [0.53, 0.54], [0.52, 0.57], [0.53, 0.59], [0.55, 0.57]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/17163_exterior_image_1_left#0.jpg", + "gt": "[[228, 102], [203, 103], [200, 114], [202, 107], [195, 95], [180, 101], [171, 98], [167, 103], [171, 106], [175, 103]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "36076_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'switch off the mini light on the right side of the adapter', and now you are performing subtask 'switch off the mini light on the right side of the adapter.'\nWith start location [0.21, 0.15], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.21, 0.15], [0.41, 0.34], [0.44, 0.41], [0.51, 0.64], [0.5, 0.67], [0.53, 0.68], [0.56, 0.75], [0.55, 0.9], [0.5, 0.77], [0.51, 0.7]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/36076_exterior_image_2_left#0.jpg", + "gt": "[[67, 27], [130, 62], [141, 73], [164, 115], [161, 121], [171, 122], [179, 135], [176, 162], [161, 138], [162, 126]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0071_user_0014_scene_0003_cfg_0002_036422060215#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close a sliding window', and now you are performing subtask 'push the handle of the sliding window to close the window.'\nYou start from [0.46, 0.46]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.46], [0.46, 0.46], [0.43, 0.46], [0.45, 0.42], [0.46, 0.41], [0.46, 0.4], [0.47, 0.41], [0.47, 0.42], [0.47, 0.44], [0.47, 0.48]]}" + } + ], + "images": "RH20T_cfg2_task_0071_user_0014_scene_0003_cfg_0002_036422060215#236", + "gt": "[[296, 164], [294, 166], [277, 166], [287, 152], [293, 146], [296, 145], [301, 149], [300, 152], [303, 159], [303, 173]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0045_user_0020_scene_0010_cfg_0006_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the brush on the pen rack', and now you are performing subtask 'place the brush on the pen rack.'\nYou begin at [0.53, 0.31]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.31], [0.54, 0.47], [0.59, 0.55], [0.59, 0.62], [0.62, 0.41], [0.6, 0.3], [0.63, 0.28], [0.62, 0.24], [0.6, 0.24], [0.64, 0.25]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0045_user_0020_scene_0010_cfg_0006_104122064161#0.jpg", + "gt": "[[338, 110], [343, 170], [377, 198], [377, 224], [400, 147], [387, 109], [402, 100], [399, 87], [383, 87], [410, 90]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "55187_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the pen out of the dark blue cup', and now you are performing subtask 'take the pen out of the dark blue cup.'\nYour starting position is [0.29, 0.4]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.29, 0.4], [0.36, 0.38], [0.42, 0.43], [0.47, 0.58], [0.5, 0.59], [0.48, 0.76], [0.5, 0.7], [0.54, 0.78], [0.54, 0.88], [0.55, 0.72]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/55187_exterior_image_2_left#0.jpg", + "gt": "[[93, 72], [116, 68], [135, 77], [152, 105], [159, 107], [155, 137], [161, 126], [174, 140], [174, 158], [176, 130]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0044_user_0010_scene_0008_cfg_0005_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up a bag of things', and now you are performing subtask 'pick up a bag of things.'\nThe trajectory starts at [0.47, 0.4]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.4], [0.4, 0.54], [0.37, 0.53], [0.34, 0.56], [0.36, 0.57], [0.35, 0.58], [0.36, 0.6], [0.38, 0.57], [0.38, 0.3], [0.39, 0.31]]}" + } + ], + "images": "RH20T_cfg5_task_0044_user_0010_scene_0008_cfg_0005_037522062165#0", + "gt": "[[298, 143], [259, 194], [239, 192], [216, 202], [228, 205], [225, 210], [231, 216], [240, 206], [241, 107], [250, 111]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "44059_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the blue and white plush toy on the right side of the table', and now you are performing subtask 'place the blue and white plush toy on the right side of the table.'\nYour initial coordinate is [0.32, 0.01]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.01], [0.39, 0.21], [0.44, 0.54], [0.42, 0.59], [0.44, 0.57], [0.47, 0.59], [0.44, 0.68], [0.43, 0.68], [0.38, 0.44], [0.4, 0.42]]}" + } + ], + "images": "44059_exterior_image_1_left#0", + "gt": "[[103.5, 1.5], [123.5, 38.5], [140.5, 97.5], [134.5, 106.5], [140.5, 102.5], [150.5, 106.5], [141.5, 122.5], [137.5, 121.5], [121.5, 79.5], [126.5, 75.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "12617_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take out the bag and the plastic container from the box and then put them on the countertop', and now you are performing subtask 'place the plastic container on the kitchen countertop.'\nFrom start point [0.5, 0.38], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.38], [0.54, 0.22], [0.53, 0.19], [0.43, 0.38], [0.4, 0.48], [0.39, 0.68], [0.44, 0.63], [0.49, 0.66], [0.49, 0.63], [0.47, 0.62]]}" + } + ], + "images": "12617_exterior_image_2_left#326", + "gt": "[[160, 69], [173, 39], [169, 35], [136, 68], [129, 86], [125, 122], [142, 114], [156, 119], [156, 114], [150, 112]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0081_user_0010_scene_0009_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the leg of the glasses (with one hand)', and now you are performing subtask 'unfold the leg of the glasses (with one hand).'\nStart at [0.63, 0.35]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.35], [0.59, 0.16], [0.64, 0.07], [0.63, 0.13], [0.64, 0.06], [0.63, 0.17], [0.64, 0.13], [0.62, 0.15], [0.63, 0.12], [0.63, 0.19]]}" + } + ], + "images": "RH20T_cfg1_task_0081_user_0010_scene_0009_cfg_0001_038522062547#0", + "gt": "[[404, 127], [378, 59], [409, 26], [404, 46], [408, 22], [405, 60], [407, 48], [397, 55], [401, 45], [403, 67]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "19146_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the blue towel from the left to the right', and now you are performing subtask 'move the blue towel from the left to the right.'\nStart at [0.71, 0.02]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.71, 0.02], [0.61, 0.28], [0.59, 0.36], [0.59, 0.38], [0.61, 0.37], [0.6, 0.39], [0.55, 0.45], [0.5, 0.58], [0.47, 0.63], [0.6, 0.42]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/19146_exterior_image_2_left#0.jpg", + "gt": "[[226, 3], [194, 51], [189, 64], [189, 69], [194, 67], [193, 70], [176, 81], [161, 105], [152, 113], [192, 76]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "44415_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn the tap nozzle to the left', and now you are performing subtask 'turn the tap nozzle clockwise.'\nYou start from [0.6, 0.37]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.37], [0.58, 0.39], [0.57, 0.37], [0.57, 0.33], [0.55, 0.24], [0.55, 0.2], [0.53, 0.17], [0.47, 0.16], [0.52, 0.18], [0.49, 0.16]]}" + } + ], + "images": "44415_exterior_image_1_left#79", + "gt": "[[193, 67], [187, 70], [183, 66], [184, 59], [175, 43], [176, 36], [170, 31], [152, 28], [168, 33], [158, 29]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0077_user_0014_scene_0006_cfg_0002_105422061350#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stacking blocks (small Lego) randomly one at a time', and now you are performing subtask 'place the blue block at the grid center, pick up the red block on the table, transfer the red block from the table to the top of the blue block, place the red block on the blue block.'\nBegin at [0.31, 0.29] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.31, 0.29], [0.26, 0.17], [0.3, 0.14], [0.27, 0.16], [0.31, 0.23], [0.3, 0.26], [0.31, 0.29], [0.3, 0.3], [0.32, 0.28], [0.32, 0.28]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg2_task_0077_user_0014_scene_0006_cfg_0002_105422061350#281.jpg", + "gt": "[[200, 104], [169, 61], [190, 50], [175, 58], [197, 83], [191, 94], [199, 103], [192, 107], [203, 101], [203, 101]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_104122064161#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'place the object on the tabletop, pick up the cube in the box, transfer the cube from the box to the tabletop, place the cube on the tabletop.'\nFrom start point [0.58, 0.63], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.63], [0.6, 0.52], [0.58, 0.45], [0.57, 0.44], [0.56, 0.49], [0.57, 0.41], [0.57, 0.47], [0.54, 0.59], [0.57, 0.33], [0.57, 0.31]]}" + } + ], + "images": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_104122064161#453", + "gt": "[[372, 226], [381, 189], [373, 161], [366, 158], [360, 176], [366, 149], [364, 171], [347, 212], [368, 119], [362, 113]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "79403_exterior_image_2_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the cauliflower, red ball, and light blue object into the silver bowl', and now you are performing subtask 'place the blue object in the transparent box.'\nYou start from [0.64, 0.49]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.64, 0.49], [0.61, 0.52], [0.62, 0.46], [0.66, 0.46], [0.68, 0.52], [0.63, 0.6], [0.6, 0.62], [0.62, 0.46], [0.76, 0.39], [0.76, 0.39]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/79403_exterior_image_2_left#267.jpg", + "gt": "[[206, 88], [196, 94], [199, 82], [210, 83], [217, 93], [201, 108], [193, 111], [198, 82], [244, 70], [244, 70]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122061018#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'place the item on the table, take out the contents of the gift box, transfer the object from the container to the surface, place the object on the surface.'\nStarting at [0.47, 0.31], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.31], [0.46, 0.24], [0.47, 0.31], [0.55, 0.49], [0.47, 0.16], [0.48, 0.14], [0.44, 0.13], [0.43, 0.23], [0.47, 0.06], [0.53, 0.0]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122061018#440.jpg", + "gt": "[[298, 111], [293, 88], [299, 112], [351, 175], [303, 58], [309, 52], [283, 47], [278, 84], [299, 22], [341, 0]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0014_user_0010_scene_0005_cfg_0004_036422060909#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the object from the scale', and now you are performing subtask 'remove the object from the scale.'\nWith start location [0.53, 0.22], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.22], [0.5, 0.21], [0.5, 0.26], [0.51, 0.27], [0.44, 0.29], [0.48, 0.38], [0.38, 0.32], [0.32, 0.44], [0.35, 0.31], [0.39, 0.26]]}" + } + ], + "images": "RH20T_cfg4_task_0014_user_0010_scene_0005_cfg_0004_036422060909#0", + "gt": "[[337, 78], [321, 75], [317, 93], [327, 96], [284, 104], [305, 138], [241, 117], [206, 160], [225, 110], [250, 93]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0038_user_0010_scene_0010_cfg_0003_038522062288#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the cup', and now you are performing subtask 'pick up the cup.'\nYour starting position is [0.37, 0.36]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.36], [0.41, 0.33], [0.34, 0.43], [0.31, 0.54], [0.26, 0.56], [0.23, 0.67], [0.21, 0.63], [0.19, 0.66], [0.24, 0.56], [0.29, 0.41]]}" + } + ], + "images": "RH20T_cfg3_task_0038_user_0010_scene_0010_cfg_0003_038522062288#0", + "gt": "[[238, 128], [263, 120], [220, 154], [196, 193], [169, 201], [146, 241], [136, 228], [121, 237], [156, 202], [184, 149]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "48366_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the black pen in the grey bowl', and now you are performing subtask 'put the black pen in the grey bowl.'\nThe trajectory starts at [0.89, 0.11]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.89, 0.11], [0.86, 0.12], [0.79, 0.19], [0.66, 0.31], [0.63, 0.37], [0.62, 0.43], [0.62, 0.39], [0.6, 0.27], [0.56, 0.14], [0.56, 0.17]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/48366_exterior_image_2_left#0.jpg", + "gt": "[[285, 19], [274, 22], [253, 35], [211, 56], [202, 67], [199, 78], [199, 70], [193, 48], [179, 26], [179, 31]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "13889_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'shift the garment to the right then use the spoon to stir the contents of the bowl', and now you are performing subtask 'pick up the spoon on the table.'\nStart at [0.31, 0.52]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.31, 0.52], [0.4, 0.53], [0.45, 0.64], [0.58, 0.57], [0.5, 0.52], [0.47, 0.53], [0.47, 0.57], [0.5, 0.63], [0.5, 0.7], [0.51, 0.74]]}" + } + ], + "images": "13889_exterior_image_1_left#0", + "gt": "[[98, 93], [129, 96], [143, 115], [187, 102], [161, 93], [152, 96], [150, 102], [159, 113], [159, 126], [164, 133]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0123_user_0010_scene_0004_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the knife on the knife rack', and now you are performing subtask 'pick up the knife on the table, transfer the knife from the table to the knife rack, place the knife on the knife rack.'\nYour initial coordinate is [0.33, 0.33]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.33, 0.33], [0.31, 0.45], [0.34, 0.45], [0.36, 0.49], [0.35, 0.52], [0.41, 0.44], [0.41, 0.53], [0.4, 0.57], [0.4, 0.58], [0.41, 0.52]]}" + } + ], + "images": "RH20T_cfg5_task_0123_user_0010_scene_0004_cfg_0005_104122063678#0", + "gt": "[[212, 120], [197, 161], [220, 162], [230, 176], [221, 186], [260, 160], [264, 190], [253, 206], [259, 208], [260, 187]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0020_user_0008_scene_0002_cfg_0001_819112070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the desk lamp by pressing the button', and now you are performing subtask 'turn on the desk lamp by pressing the button.'\nThe trajectory starts at [0.62, 0.46]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.46], [0.68, 0.41], [0.73, 0.4], [0.73, 0.43], [0.7, 0.47], [0.7, 0.49], [0.7, 0.47], [0.68, 0.43], [0.64, 0.42], [0.63, 0.44]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0020_user_0008_scene_0002_cfg_0001_819112070044#0.jpg", + "gt": "[[398, 167], [436, 149], [465, 145], [467, 154], [449, 168], [448, 176], [449, 168], [433, 154], [412, 152], [403, 159]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0010_user_0003_scene_0002_cfg_0002_104122061850#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the handset of the telephone on the corresponding phone cradle', and now you are performing subtask 'pick up the telephone receiver on the table, transfer the telephone receiver from the table to the phone cradle, place the telephone receiver on the phone cradle.'\nBegin at [0.65, 0.4] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.4], [0.67, 0.39], [0.62, 0.33], [0.59, 0.35], [0.54, 0.51], [0.55, 0.54], [0.53, 0.57], [0.56, 0.58], [0.48, 0.51], [0.43, 0.51]]}" + } + ], + "images": "RH20T_cfg2_task_0010_user_0003_scene_0002_cfg_0002_104122061850#504", + "gt": "[[418, 144], [428, 140], [396, 118], [377, 125], [343, 183], [350, 195], [340, 204], [356, 210], [309, 184], [274, 184]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "22632_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the towel then fold it again', and now you are performing subtask 'fold the towel.'\nStart at [0.65, 0.72]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.72], [0.67, 0.7], [0.68, 0.71], [0.7, 0.65], [0.73, 0.66], [0.72, 0.64], [0.75, 0.61], [0.76, 0.59], [0.73, 0.59], [0.72, 0.6]]}" + } + ], + "images": "22632_exterior_image_1_left#304", + "gt": "[[208, 129], [214, 126], [219, 127], [225, 117], [232, 118], [230, 116], [241, 109], [242, 107], [233, 106], [230, 108]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "14309_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pour the contents of the orange cup into the bowl', and now you are performing subtask 'pour the contents of the orange cup into the bowl.'\nWith start location [0.4, 0.58], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.58], [0.43, 0.58], [0.46, 0.62], [0.47, 0.64], [0.47, 0.67], [0.45, 0.64], [0.43, 0.63], [0.42, 0.59], [0.38, 0.57], [0.38, 0.52]]}" + } + ], + "images": "14309_exterior_image_2_left#0", + "gt": "[[128, 104], [136, 105], [146, 111], [149, 116], [149, 120], [145, 115], [139, 113], [134, 106], [122, 103], [120, 93]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0029_user_0001_scene_0004_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp the handle and open the drawer', and now you are performing subtask 'grasp the handle and open the drawer.'\nYou start from [0.63, 0.37]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.68, 0.33], [0.76, 0.18], [0.8, 0.14], [0.79, 0.2], [0.72, 0.27], [0.7, 0.26], [0.81, 0.21], [0.79, 0.19], [0.75, 0.22]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0029_user_0001_scene_0004_cfg_0001_038522062547#0.jpg", + "gt": "[[403, 134], [437, 119], [489, 66], [514, 49], [504, 72], [462, 96], [448, 94], [520, 76], [506, 70], [481, 79]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0016_user_0014_scene_0001_cfg_0001_038522062547#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'hit the pool ball', and now you are performing subtask 'pick up the pool ball from the surface.'\nStart at [0.61, 0.14]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.61, 0.14], [0.61, 0.15], [0.61, 0.16], [0.61, 0.17], [0.61, 0.17], [0.61, 0.17], [0.61, 0.17], [0.61, 0.17], [0.61, 0.17], [0.61, 0.17]]}" + } + ], + "images": "RH20T_cfg1_task_0016_user_0014_scene_0001_cfg_0001_038522062547#409", + "gt": "[[391, 51], [390, 54], [390, 57], [388, 63], [388, 62], [390, 62], [390, 61], [392, 61], [392, 60], [392, 61]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "86984_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the spoon in the mug', and now you are performing subtask 'put the spoon in the mug.'\nYour initial coordinate is [0.07, 0.45]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.07, 0.45], [0.31, 0.63], [0.33, 0.76], [0.28, 0.94], [0.31, 0.98], [0.36, 0.88], [0.3, 0.61], [0.31, 0.71], [0.31, 0.64], [0.36, 0.6]]}" + } + ], + "images": "86984_exterior_image_2_left#0", + "gt": "[[23, 81], [98, 113], [107, 137], [90, 169], [99, 176], [114, 159], [95, 110], [100, 128], [99, 116], [115, 108]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0045_user_0010_scene_0004_cfg_0005_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the brush on the pen rack', and now you are performing subtask 'place the brush on the pen rack.'\nStart at [0.49, 0.14]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.14], [0.6, 0.44], [0.6, 0.41], [0.54, 0.34], [0.47, 0.12], [0.45, 0.18], [0.48, 0.21], [0.49, 0.27], [0.52, 0.14], [0.52, 0.08]]}" + } + ], + "images": "RH20T_cfg5_task_0045_user_0010_scene_0004_cfg_0005_104122061850#0", + "gt": "[[311, 50], [386, 160], [382, 146], [344, 121], [303, 45], [290, 65], [309, 74], [314, 97], [334, 51], [332, 27]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "3579_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the orange and pineapple plushies on the black tray', and now you are performing subtask 'pick up the pineapple plush toy in the green bowl.'\nBegin at [0.85, 0.16] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.85, 0.16], [0.72, 0.22], [0.7, 0.37], [0.72, 0.48], [0.67, 0.48], [0.7, 0.43], [0.74, 0.58], [0.61, 0.47], [0.6, 0.42], [0.57, 0.43]]}" + } + ], + "images": "3579_exterior_image_1_left#0", + "gt": "[[272.0, 29.0], [230.0, 40.0], [224.0, 66.0], [229.0, 86.0], [215.0, 86.0], [224.0, 78.0], [238.0, 105.0], [196.0, 85.0], [191.0, 76.0], [181.0, 78.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0091_user_0007_scene_0006_cfg_0004_038522062288#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move an object from one box to another', and now you are performing subtask 'move an object from one box to another.'\nYour starting position is [0.37, 0.33]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.33], [0.35, 0.37], [0.34, 0.33], [0.28, 0.34], [0.29, 0.35], [0.3, 0.41], [0.29, 0.34], [0.35, 0.52], [0.4, 0.48], [0.21, 0.24]]}" + } + ], + "images": "RH20T_cfg4_task_0091_user_0007_scene_0006_cfg_0004_038522062288#0", + "gt": "[[234, 118], [225, 133], [215, 118], [176, 124], [188, 126], [192, 148], [186, 123], [225, 186], [255, 172], [135, 88]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "73339_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker in the mug', and now you are performing subtask 'put the marker in the mug.'\nWith start location [0.29, 0.09], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.29, 0.09], [0.43, 0.28], [0.51, 0.52], [0.47, 0.61], [0.51, 0.59], [0.49, 0.46], [0.41, 0.42], [0.39, 0.48], [0.43, 0.51], [0.46, 0.39]]}" + } + ], + "images": "73339_exterior_image_2_left#0", + "gt": "[[93, 17], [136, 50], [164, 94], [152, 110], [162, 106], [156, 82], [131, 76], [125, 87], [136, 92], [146, 71]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "2900_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the lid on the table then pour the contents from the clear jar into the black pot', and now you are performing subtask 'pick up the clear jar from the table.'\nBegin at [0.71, 0.25] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.71, 0.25], [0.66, 0.23], [0.57, 0.44], [0.57, 0.34], [0.61, 0.41], [0.61, 0.62], [0.56, 0.42], [0.53, 0.48], [0.55, 0.52], [0.49, 0.55]]}" + } + ], + "images": "2900_exterior_image_2_left#0", + "gt": "[[227, 45], [212, 42], [183, 80], [181, 62], [196, 74], [196, 112], [179, 76], [169, 87], [177, 93], [158, 99]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "33799_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker in the mug', and now you are performing subtask 'put the marker in the mug.'\nThe trajectory starts at [0.61, 0.34]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.61, 0.34], [0.52, 0.54], [0.51, 0.65], [0.48, 0.61], [0.45, 0.71], [0.44, 0.74], [0.45, 0.75], [0.45, 0.69], [0.44, 0.71], [0.44, 0.71]]}" + } + ], + "images": "33799_exterior_image_2_left#0", + "gt": "[[196, 61], [165, 98], [163, 117], [155, 109], [143, 127], [142, 133], [144, 135], [143, 125], [140, 128], [140, 128]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "49601_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick the bottle up and move it to the right side of the table', and now you are performing subtask 'place the bottle on the table.'\nWith start location [0.68, 0.3], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.3], [0.71, 0.22], [0.66, 0.34], [0.68, 0.35], [0.67, 0.35], [0.68, 0.34], [0.67, 0.35], [0.64, 0.3], [0.63, 0.31], [0.66, 0.27]]}" + } + ], + "images": "49601_exterior_image_1_left#125", + "gt": "[[219, 54], [227, 40], [212, 61], [218, 63], [213, 63], [216, 62], [213, 63], [204, 54], [203, 56], [210, 49]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "36160_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the cloth on the table', and now you are performing subtask 'unfold the cloth on the table.'\nWith start location [0.37, 0.43], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.43], [0.42, 0.35], [0.5, 0.34], [0.48, 0.36], [0.5, 0.38], [0.51, 0.51], [0.5, 0.56], [0.49, 0.53], [0.58, 0.63], [0.62, 0.84]]}" + } + ], + "images": "36160_exterior_image_1_left#0", + "gt": "[[118, 78], [134, 63], [159, 62], [155, 64], [161, 69], [162, 91], [161, 100], [157, 96], [186, 114], [197, 152]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "70082_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the silver spoon in the box', and now you are performing subtask 'place the silver spoon in the box.'\nThe trajectory starts at [0.33, 0.83]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.33, 0.83], [0.32, 0.83], [0.32, 0.81], [0.35, 0.86], [0.32, 0.97], [0.32, 0.89], [0.29, 0.83], [0.31, 0.83], [0.3, 0.84], [0.3, 0.86]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/70082_exterior_image_2_left#248.jpg", + "gt": "[[105, 149], [102, 150], [101, 146], [113, 155], [101, 174], [102, 160], [94, 149], [100, 150], [96, 152], [97, 154]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0130_user_0014_scene_0004_cfg_0006_104122063633#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb', and now you are performing subtask 'pick up the lamp holder and the light bulb from the table, transfer the lamp holder from the table to the socket, insert the lamp holder into the socket.'\nBegin at [0.36, 0.33] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.33], [0.3, 0.46], [0.27, 0.52], [0.26, 0.57], [0.29, 0.51], [0.3, 0.46], [0.31, 0.46], [0.31, 0.44], [0.31, 0.48], [0.31, 0.48]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0130_user_0014_scene_0004_cfg_0006_104122063633#0.jpg", + "gt": "[[232, 119], [195, 165], [170, 189], [167, 205], [188, 182], [193, 166], [200, 164], [201, 160], [197, 172], [197, 172]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0014_user_0010_scene_0005_cfg_0004_038522062288#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the object from the scale', and now you are performing subtask 'remove the object from the scale.'\nYou start from [0.36, 0.33]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.33], [0.35, 0.38], [0.38, 0.43], [0.35, 0.59], [0.38, 0.56], [0.39, 0.74], [0.46, 0.86], [0.32, 0.71], [0.31, 0.73], [0.31, 0.73]]}" + } + ], + "images": "RH20T_cfg4_task_0014_user_0010_scene_0005_cfg_0004_038522062288#0", + "gt": "[[232, 120], [227, 137], [241, 156], [222, 211], [242, 200], [248, 265], [294, 310], [204, 254], [201, 262], [201, 262]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0071_user_0020_scene_0007_cfg_0006_104122061018#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close a sliding window', and now you are performing subtask 'push the handle of the sliding window to close the upper window, push the handle of the sliding window to close the lower window.'\nThe trajectory starts at [0.44, 0.26]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.26], [0.44, 0.17], [0.45, 0.23], [0.48, 0.24], [0.43, 0.21], [0.43, 0.23], [0.45, 0.23], [0.46, 0.17], [0.54, 0.15], [0.54, 0.13]]}" + } + ], + "images": "RH20T_cfg6_task_0071_user_0020_scene_0007_cfg_0006_104122061018#298", + "gt": "[[280, 93], [279, 61], [291, 83], [305, 85], [278, 75], [277, 83], [291, 84], [294, 63], [348, 54], [345, 46]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "45040_exterior_image_1_left#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer and put some of the plush toys in the drawer and close it', and now you are performing subtask 'push the drawer handle to close it.'\nStarting at [0.42, 0.31], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.31], [0.32, 0.17], [0.4, 0.14], [0.49, 0.27], [0.54, 0.27], [0.6, 0.34], [0.57, 0.39], [0.41, 0.38], [0.54, 0.34], [0.54, 0.34]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/45040_exterior_image_1_left#580.jpg", + "gt": "[[134.5, 55.5], [102.5, 30.5], [126.5, 25.5], [155.5, 48.5], [172.5, 48.5], [190.5, 60.5], [183.5, 70.5], [131.5, 68.5], [171.5, 60.5], [171.5, 60.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0065_user_0014_scene_0002_cfg_0001_750612070851#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with small Lego blocks for the nth time with n blocks', and now you are performing subtask 'pick up the pink block on the desktop, transfer the pink block from the desktop to the building area, place the pink block in the building area.'\nStart at [0.49, 0.47]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.47], [0.48, 0.44], [0.4, 0.51], [0.35, 0.64], [0.33, 0.64], [0.32, 0.73], [0.32, 0.75], [0.36, 0.63], [0.39, 0.6], [0.39, 0.64]]}" + } + ], + "images": "RH20T_cfg1_task_0065_user_0014_scene_0002_cfg_0001_750612070851#0", + "gt": "[[312, 168], [306, 157], [256, 185], [224, 230], [214, 231], [202, 261], [205, 271], [230, 228], [251, 216], [248, 229]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0061_user_0007_scene_0008_cfg_0005_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the first block out of the box and place it on the table, then take the second block and place it on top of the first block', and now you are performing subtask 'pick up the first block in the box, transfer the first block from the box to above the table, place the first block on the table.'\nFrom start point [0.45, 0.42], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.42], [0.45, 0.47], [0.44, 0.51], [0.44, 0.55], [0.42, 0.63], [0.42, 0.58], [0.43, 0.57], [0.5, 0.58], [0.54, 0.63], [0.52, 0.72]]}" + } + ], + "images": "RH20T_cfg5_task_0061_user_0007_scene_0008_cfg_0005_037522062165#0", + "gt": "[[289, 152], [287, 171], [280, 183], [281, 198], [268, 226], [269, 209], [275, 205], [318, 208], [344, 226], [336, 260]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0034_user_0007_scene_0008_cfg_0005_104122063678#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'pick up the fourth square from the table, transfer the fourth square from the table to the position above the third square, place the fourth square on the third square.'\nYou begin at [0.63, 0.55]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.55], [0.66, 0.59], [0.67, 0.62], [0.66, 0.68], [0.66, 0.62], [0.65, 0.59], [0.6, 0.53], [0.56, 0.52], [0.56, 0.47], [0.56, 0.47]]}" + } + ], + "images": "RH20T_cfg5_task_0034_user_0007_scene_0008_cfg_0005_104122063678#894", + "gt": "[[402, 198], [425, 213], [430, 223], [421, 244], [422, 225], [416, 213], [383, 192], [359, 186], [361, 171], [361, 171]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "84706_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the pee plush toy to the counter, then pick up the cloth and place it on the counter, then pick up the toy from the sink and put it in the pot', and now you are performing subtask 'pick up the pink toy from the sink.'\nStarting at [0.33, 0.64], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.33, 0.64], [0.3, 0.64], [0.33, 0.5], [0.31, 0.31], [0.39, 0.17], [0.51, 0.19], [0.43, 0.33], [0.38, 0.32], [0.33, 0.48], [0.36, 0.5]]}" + } + ], + "images": "84706_exterior_image_1_left#105", + "gt": "[[106.5, 116.0], [95.5, 116.0], [105.5, 90.0], [100.5, 55.0], [123.5, 30.0], [163.5, 34.0], [138.5, 59.0], [120.5, 57.0], [105.5, 86.0], [116.5, 90.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "351_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'touch the lid of the cup', and now you are performing subtask 'touch the lid of the cup.'\nWith start location [0.7, 0.7], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.7], [0.83, 0.54], [0.98, 0.48], [1.0, 0.51], [0.79, 0.67], [0.76, 0.72], [0.75, 0.86], [0.68, 0.92], [0.7, 0.94], [0.71, 0.83]]}" + } + ], + "images": "351_exterior_image_1_left#0", + "gt": "[[222.5, 126.0], [265.5, 98.0], [314.5, 86.0], [320.0, 91.0], [252.5, 121.0], [243.5, 130.0], [238.5, 155.0], [218.5, 165.0], [222.5, 170.0], [227.5, 149.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0047_user_0014_scene_0014_cfg_0004_104422070011#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the cup off the cup rack', and now you are performing subtask 'take the cup off the cup rack.'\nThe trajectory starts at [0.53, 0.29]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.29], [0.7, 0.43], [0.7, 0.39], [0.53, 0.38], [0.45, 0.29], [0.43, 0.35], [0.45, 0.41], [0.61, 0.36], [0.65, 0.41], [0.65, 0.41]]}" + } + ], + "images": "RH20T_cfg4_task_0047_user_0014_scene_0014_cfg_0004_104422070011#0", + "gt": "[[337, 103], [451, 155], [450, 141], [340, 137], [291, 103], [275, 125], [288, 146], [393, 130], [417, 148], [417, 148]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "2437_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the tulip plush toy inside the sink', and now you are performing subtask 'put the tulip plush toy inside the sink.'\nStarting at [0.75, 0.01], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.75, 0.01], [0.52, 0.44], [0.48, 0.56], [0.51, 0.66], [0.63, 0.49], [0.69, 0.49], [0.69, 0.53], [0.65, 0.59], [0.62, 0.47], [0.76, 0.36]]}" + } + ], + "images": "2437_exterior_image_2_left#0", + "gt": "[[241, 2], [166, 79], [154, 100], [164, 119], [201, 89], [221, 88], [222, 96], [209, 107], [199, 84], [244, 65]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "84821_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer', and now you are performing subtask 'open the drawer.'\nYou begin at [0.43, 0.39]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.39], [0.45, 0.36], [0.45, 0.51], [0.5, 0.52], [0.49, 0.55], [0.53, 0.56], [0.49, 0.53], [0.39, 0.59], [0.35, 0.3], [0.37, 0.27]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/84821_exterior_image_1_left#0.jpg", + "gt": "[[137, 71], [143, 64], [144, 91], [160, 93], [158, 99], [169, 100], [157, 95], [124, 106], [112, 54], [118, 49]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "41140_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'spread the silver containers on the counter', and now you are performing subtask 'place the silver container on the counter.'\nYou begin at [0.59, 0.67]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.67], [0.55, 0.63], [0.46, 0.67], [0.47, 0.68], [0.41, 0.72], [0.4, 0.76], [0.41, 0.83], [0.42, 0.74], [0.47, 0.77], [0.44, 0.72]]}" + } + ], + "images": "41140_exterior_image_2_left#0", + "gt": "[[190, 120], [176, 114], [146, 120], [150, 123], [132, 129], [127, 136], [130, 149], [135, 133], [151, 139], [140, 130]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "14075_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the object into the pot, then open the drawer and take the object from the pot and put it into the drawer', and now you are performing subtask 'place the object in the drawer.'\nWith start location [0.63, 0.85], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.85], [0.68, 0.91], [0.68, 0.97], [0.75, 0.81], [0.53, 0.9], [0.5, 1.0], [0.51, 0.76], [0.57, 0.69], [0.68, 0.73], [0.68, 0.73]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/14075_exterior_image_1_left#252.jpg", + "gt": "[[201.0, 153.0], [218.0, 163.0], [216.0, 174.0], [240.0, 146.0], [171.0, 162.0], [159.0, 180.0], [162.0, 136.0], [182.0, 125.0], [217.0, 132.0], [217.0, 132.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "89810_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the blue cloth from the back of the chair', and now you are performing subtask 'pick up the blue fabric on the back of the chair.'\nFrom start point [0.75, 0.66], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.75, 0.66], [0.68, 0.59], [0.64, 0.58], [0.65, 0.59], [0.63, 0.63], [0.61, 0.63], [0.62, 0.69], [0.61, 0.71], [0.6, 0.67], [0.61, 0.7]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/89810_exterior_image_1_left#0.jpg", + "gt": "[[241, 119], [216, 106], [206, 105], [208, 107], [202, 113], [196, 113], [197, 125], [195, 128], [193, 121], [195, 126]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "33072_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn off the left light switch', and now you are performing subtask 'turn off the left light switch.'\nStart at [0.66, 0.49]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.66, 0.49], [0.61, 0.49], [0.56, 0.42], [0.54, 0.44], [0.52, 0.41], [0.49, 0.33], [0.5, 0.3], [0.44, 0.31], [0.46, 0.39], [0.49, 0.41]]}" + } + ], + "images": "33072_exterior_image_1_left#0", + "gt": "[[212, 88], [196, 88], [180, 75], [173, 79], [165, 74], [156, 59], [159, 54], [140, 56], [146, 71], [158, 73]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "69663_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the pen from the table and put it in the yellow cup', and now you are performing subtask 'pick up the pen from the table and put it in the yellow cup.'\nStart at [0.97, 0.0]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.97, 0.0], [0.93, 0.09], [0.97, 0.08], [0.79, 0.27], [0.75, 0.42], [0.78, 0.17], [0.81, 0.14], [0.79, 0.31], [0.74, 0.43], [0.69, 0.29]]}" + } + ], + "images": "69663_exterior_image_1_left#0", + "gt": "[[311, 0], [298, 17], [311, 15], [254, 48], [239, 75], [250, 31], [259, 26], [254, 56], [237, 77], [222, 52]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0004_user_0006_scene_0003_cfg_0002_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the small block on the left and move it to the right', and now you are performing subtask 'pick up the small block on the left and move it to the right.'\nFrom start point [0.44, 0.48], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.48], [0.32, 0.44], [0.24, 0.34], [0.28, 0.3], [0.29, 0.43], [0.28, 0.36], [0.3, 0.36], [0.35, 0.4], [0.36, 0.47], [0.38, 0.37]]}" + } + ], + "images": "RH20T_cfg2_task_0004_user_0006_scene_0003_cfg_0002_037522062165#0", + "gt": "[[282, 174], [208, 157], [151, 123], [178, 109], [188, 153], [180, 129], [195, 130], [227, 143], [233, 168], [240, 133]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "58558_exterior_image_2_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the clothes', and now you are performing subtask 'fold the garment once more on the table.'\nYour initial coordinate is [0.53, 0.6]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.6], [0.48, 0.64], [0.5, 0.68], [0.5, 0.66], [0.48, 0.68], [0.48, 0.63], [0.61, 0.52], [0.68, 0.49], [0.58, 0.56], [0.54, 0.63]]}" + } + ], + "images": "58558_exterior_image_2_left#361", + "gt": "[[171, 108], [154, 116], [160, 122], [160, 119], [154, 123], [153, 114], [194, 93], [219, 89], [185, 100], [173, 113]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0037_user_0007_scene_0006_cfg_0002_036422060215#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the second block on the third block, pick up the third block on the table, transfer the third block from the table to the top of the fourth block, place the third block on the fourth block.'\nStart at [0.49, 0.68]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.68], [0.47, 0.66], [0.44, 0.69], [0.45, 0.72], [0.45, 0.69], [0.46, 0.64], [0.48, 0.62], [0.49, 0.62], [0.49, 0.65], [0.5, 0.64]]}" + } + ], + "images": "RH20T_cfg2_task_0037_user_0007_scene_0006_cfg_0002_036422060215#471", + "gt": "[[312, 243], [301, 239], [281, 250], [287, 260], [285, 249], [296, 230], [309, 224], [314, 224], [316, 235], [319, 230]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "92163_exterior_image_1_left#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the towel', and now you are performing subtask 'pull the towel to the inside of the table.'\nYour initial coordinate is [0.36, 0.61]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.61], [0.4, 0.61], [0.38, 0.56], [0.29, 0.62], [0.41, 0.53], [0.3, 0.86], [0.41, 0.69], [0.46, 0.48], [0.41, 0.46], [0.41, 0.46]]}" + } + ], + "images": "92163_exterior_image_1_left#801", + "gt": "[[115, 110], [127, 109], [120, 100], [93, 111], [130, 96], [95, 155], [132, 124], [146, 86], [132, 83], [132, 83]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "51220_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the orange object inside the sink', and now you are performing subtask 'put the orange object inside the sink.'\nYour initial coordinate is [0.72, 0.49]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.72, 0.49], [0.58, 0.67], [0.57, 0.81], [0.57, 0.68], [0.63, 0.63], [0.68, 0.63], [0.7, 0.68], [0.7, 0.81], [0.68, 0.72], [0.71, 0.72]]}" + } + ], + "images": "51220_exterior_image_2_left#0", + "gt": "[[231, 89], [185, 121], [184, 145], [182, 123], [203, 114], [219, 114], [225, 123], [225, 146], [216, 130], [226, 129]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "26372_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the washing machine lid', and now you are performing subtask 'open the washing machine lid.'\nYou begin at [0.32, 0.33]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.33], [0.41, 0.48], [0.42, 0.62], [0.43, 0.64], [0.45, 0.63], [0.44, 0.66], [0.39, 0.74], [0.41, 0.62], [0.37, 0.43], [0.33, 0.31]]}" + } + ], + "images": "26372_exterior_image_1_left#0", + "gt": "[[103, 60], [132, 86], [135, 111], [139, 116], [143, 114], [142, 118], [125, 133], [130, 111], [119, 78], [106, 55]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "86720_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take out a straw from the open drawer and put it on the countertop, then close the open drawer', and now you are performing subtask 'push the drawer to the closed position.'\nStarting at [0.52, 0.5], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.5], [0.49, 0.53], [0.51, 0.53], [0.47, 0.57], [0.44, 0.64], [0.44, 0.71], [0.46, 0.74], [0.49, 0.74], [0.57, 0.61], [0.44, 0.56]]}" + } + ], + "images": "86720_exterior_image_1_left#188", + "gt": "[[167, 90], [157, 95], [163, 95], [150, 102], [141, 116], [141, 128], [148, 134], [157, 134], [182, 109], [142, 101]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0041_user_0007_scene_0006_cfg_0005_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'clean the tabletop with a sponge', and now you are performing subtask 'clean the tabletop with a sponge.'\nFrom start point [0.45, 0.3], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.3], [0.43, 0.36], [0.38, 0.42], [0.34, 0.43], [0.32, 0.49], [0.37, 0.44], [0.44, 0.46], [0.38, 0.49], [0.43, 0.48], [0.43, 0.48]]}" + } + ], + "images": "RH20T_cfg5_task_0041_user_0007_scene_0006_cfg_0005_036422060215#0", + "gt": "[[286, 108], [278, 129], [245, 152], [218, 154], [205, 175], [238, 157], [282, 167], [245, 178], [276, 172], [276, 172]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0081_user_0015_scene_0002_cfg_0004_104422070011#TrajectoryPrediction_Task#14", + "task": "TrajectoryPrediction_Task", + "p_index": 14, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the leg of the glasses (with one hand)', and now you are performing subtask 'pick up the glasses on the table.'\nWith start location [0.42, 0.33], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.33], [0.43, 0.36], [0.42, 0.41], [0.42, 0.49], [0.41, 0.5], [0.41, 0.49], [0.42, 0.49], [0.43, 0.47], [0.44, 0.56], [0.43, 0.54]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0081_user_0015_scene_0002_cfg_0004_104422070011#422.jpg", + "gt": "[[270, 119], [272, 131], [267, 148], [266, 175], [261, 181], [262, 177], [267, 178], [276, 171], [283, 202], [274, 196]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "78490_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull the drawer handle to open the drawer', and now you are performing subtask 'pull the drawer handle to open the drawer.'\nYour starting position is [0.41, 0.54]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.54], [0.42, 0.5], [0.44, 0.56], [0.46, 0.53], [0.46, 0.69], [0.49, 0.73], [0.57, 0.68], [0.51, 0.8], [0.45, 0.73], [0.44, 0.51]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/78490_exterior_image_1_left#0.jpg", + "gt": "[[131, 97], [135, 90], [140, 100], [147, 96], [147, 124], [157, 132], [184, 123], [163, 144], [143, 131], [141, 92]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg3_task_0018_user_0011_scene_0002_cfg_0003_104122063550#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'play jenga', and now you are performing subtask 'play jenga.'\nYour initial coordinate is [0.43, 0.53]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.53], [0.46, 0.54], [0.46, 0.51], [0.5, 0.53], [0.49, 0.49], [0.53, 0.55], [0.47, 0.61], [0.48, 0.8], [0.5, 0.77], [0.47, 0.65]]}" + } + ], + "images": "RH20T_cfg3_task_0018_user_0011_scene_0002_cfg_0003_104122063550#0", + "gt": "[[277, 192], [297, 193], [292, 183], [317, 191], [312, 176], [339, 198], [303, 218], [305, 287], [318, 278], [302, 233]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "3618_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the marker in the mug and put it on the table', and now you are performing subtask 'pick up the marker from the cup.'\nWith start location [0.96, 0.0], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.96, 0.0], [0.84, 0.09], [0.79, 0.17], [0.8, 0.18], [0.76, 0.38], [0.74, 0.39], [0.73, 0.36], [0.73, 0.38], [0.74, 0.32], [0.74, 0.34]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/3618_exterior_image_1_left#0.jpg", + "gt": "[[308.5, 0.0], [268.5, 16.0], [252.5, 31.0], [256.5, 32.0], [243.5, 68.0], [236.5, 70.0], [233.5, 65.0], [234.5, 69.0], [235.5, 58.0], [237.5, 61.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0037_user_0007_scene_0004_cfg_0005_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the first block from the surface, place the first block on the vertical stack on the surface, pick up the second block from the surface.'\nYou start from [0.45, 0.42]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.42], [0.39, 0.43], [0.33, 0.55], [0.31, 0.56], [0.32, 0.6], [0.32, 0.57], [0.36, 0.54], [0.41, 0.54], [0.43, 0.56], [0.43, 0.62]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0037_user_0007_scene_0004_cfg_0005_037522062165#0.jpg", + "gt": "[[289, 152], [249, 156], [214, 199], [201, 200], [204, 216], [208, 205], [231, 194], [261, 193], [278, 202], [272, 225]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0094_user_0007_scene_0009_cfg_0004_104422070011#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'push the box to open the box.'\nWith start location [0.51, 0.35], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.35], [0.54, 0.35], [0.52, 0.32], [0.52, 0.23], [0.5, 0.15], [0.49, 0.18], [0.5, 0.17], [0.52, 0.2], [0.53, 0.33], [0.55, 0.36]]}" + } + ], + "images": "RH20T_cfg4_task_0094_user_0007_scene_0009_cfg_0004_104422070011#190", + "gt": "[[329, 127], [344, 126], [330, 116], [333, 84], [320, 54], [315, 65], [323, 63], [331, 72], [341, 119], [351, 131]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0034_user_0014_scene_0001_cfg_0007_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'pick up the square, transfer the square from the table to the table, pick up the square.'\nYour starting position is [0.53, 0.32]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.32], [0.54, 0.53], [0.52, 0.66], [0.52, 0.71], [0.51, 0.74], [0.52, 0.47], [0.53, 0.45], [0.55, 0.47], [0.55, 0.54], [0.57, 0.57]]}" + } + ], + "images": "RH20T_cfg7_task_0034_user_0014_scene_0001_cfg_0007_104122064161#0", + "gt": "[[337, 115], [345, 192], [330, 239], [335, 254], [325, 268], [332, 168], [339, 163], [351, 171], [354, 195], [364, 204]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122063633#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'place the item on the table, take out the contents of the gift box, transfer the object from the container to the surface, place the object on the surface.'\nYou begin at [0.29, 0.53]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.29, 0.53], [0.29, 0.59], [0.31, 0.62], [0.29, 0.42], [0.3, 0.38], [0.33, 0.36], [0.33, 0.33], [0.32, 0.41], [0.32, 0.29], [0.34, 0.26]]}" + } + ], + "images": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122063633#466", + "gt": "[[183, 191], [186, 214], [198, 225], [186, 151], [189, 137], [212, 128], [212, 120], [208, 147], [204, 105], [216, 94]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0089_user_0020_scene_0009_cfg_0006_104122064161#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push a block with obstacles', and now you are performing subtask 'push a block to the right, push a block backward.'\nWith start location [0.67, 0.47], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.67, 0.47], [0.68, 0.49], [0.7, 0.46], [0.73, 0.51], [0.68, 0.53], [0.65, 0.59], [0.68, 0.51], [0.66, 0.38], [0.62, 0.3], [0.58, 0.29]]}" + } + ], + "images": "RH20T_cfg6_task_0089_user_0020_scene_0009_cfg_0006_104122064161#291", + "gt": "[[430, 169], [438, 177], [446, 167], [467, 182], [434, 192], [415, 214], [433, 183], [425, 136], [398, 107], [370, 103]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'pick up the block on the flat surface, transfer the block from the flat surface to the block board, place the block on the block board.'\nFrom start point [0.53, 0.32], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.32], [0.62, 0.44], [0.67, 0.55], [0.67, 0.62], [0.65, 0.66], [0.65, 0.69], [0.62, 0.55], [0.57, 0.51], [0.55, 0.55], [0.56, 0.52]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122064161#0.jpg", + "gt": "[[337, 115], [398, 160], [428, 199], [426, 225], [413, 237], [414, 247], [396, 197], [365, 182], [351, 199], [360, 189]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0034_user_0020_scene_0001_cfg_0006_104122064161#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'pick up the fourth square on the top, transfer the fourth square from the top to above the third square, place the fourth square above the third square.'\nWith start location [0.57, 0.25], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.25], [0.6, 0.19], [0.6, 0.21], [0.6, 0.26], [0.6, 0.23], [0.59, 0.28], [0.54, 0.38], [0.56, 0.46], [0.58, 0.39], [0.57, 0.33]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0034_user_0020_scene_0001_cfg_0006_104122064161#660.jpg", + "gt": "[[363, 90], [383, 70], [386, 75], [384, 95], [382, 83], [378, 100], [347, 135], [356, 167], [370, 139], [368, 118]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "74748_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the lid on the black and yellow boxes', and now you are performing subtask 'close the lid on the black and yellow boxes.'\nFrom start point [0.68, 0.29], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.29], [0.76, 0.3], [0.95, 0.24], [0.96, 0.27], [0.99, 0.26], [1.0, 0.28], [1.0, 0.25], [0.92, 0.28], [0.74, 0.43], [0.72, 0.37]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/74748_exterior_image_2_left#0.jpg", + "gt": "[[218, 52], [244, 54], [303, 43], [307, 49], [317, 46], [320, 51], [320, 45], [293, 50], [236, 77], [229, 66]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0004_user_0010_scene_0001_cfg_0005_105422061350#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the small block on the left and move it to the right', and now you are performing subtask 'pick up the small block on the left and move it to the right.'\nYour initial coordinate is [0.43, 0.13]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.13], [0.43, 0.16], [0.41, 0.1], [0.41, 0.0], [0.45, 0.0], [0.45, 0.1], [0.42, 0.07], [0.39, 0.16], [0.41, 0.36], [0.36, 0.46]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0004_user_0010_scene_0001_cfg_0005_105422061350#0.jpg", + "gt": "[[276, 47], [278, 58], [264, 36], [264, 0], [286, 0], [289, 37], [267, 24], [250, 56], [262, 128], [231, 165]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0096_user_0020_scene_0003_cfg_0006_104122064161#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take something out of a drawer', and now you are performing subtask 'pull the handle of the drawer.'\nYou begin at [0.52, 0.62]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.62], [0.52, 0.64], [0.52, 0.64], [0.54, 0.69], [0.53, 0.64], [0.55, 0.59], [0.53, 0.61], [0.53, 0.66], [0.54, 0.63], [0.54, 0.51]]}" + } + ], + "images": "RH20T_cfg6_task_0096_user_0020_scene_0003_cfg_0006_104122064161#435", + "gt": "[[331, 223], [332, 231], [334, 230], [344, 248], [337, 230], [349, 212], [340, 220], [338, 239], [343, 228], [343, 184]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "30342_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the garlic on the counter into the glass bowl one by one', and now you are performing subtask 'place the third garlic into the glass bowl.'\nYour starting position is [0.65, 0.72]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.72], [0.67, 0.68], [0.7, 0.68], [0.71, 0.76], [0.7, 0.78], [0.7, 0.73], [0.67, 0.71], [0.64, 0.8], [0.68, 0.69], [0.67, 0.69]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/30342_exterior_image_1_left#233.jpg", + "gt": "[[209, 129], [215, 123], [225, 123], [228, 137], [223, 140], [223, 131], [214, 128], [205, 144], [218, 124], [213, 125]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "59697_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the objects from the orange bowl to the black pot, then cover the black pot with the glass lid', and now you are performing subtask 'place the glass lid on the black pot.'\nStart at [0.53, 0.27]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.27], [0.62, 0.4], [0.62, 0.49], [0.58, 0.53], [0.65, 0.41], [0.64, 0.24], [0.58, 0.08], [0.54, 0.26], [0.5, 0.27], [0.53, 0.12]]}" + } + ], + "images": "59697_exterior_image_1_left#333", + "gt": "[[170, 49], [198, 72], [200, 89], [187, 96], [208, 74], [204, 44], [187, 14], [174, 46], [160, 49], [170, 22]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "42894_exterior_image_1_left#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull the white bowl to the left, take out the fork and the spoon from the bowl, then place them on the table, move the white cup to the left', and now you are performing subtask 'place the cup on the left side of the table.'\nThe trajectory starts at [0.49, 0.57]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.57], [0.55, 0.62], [0.57, 0.81], [0.57, 0.71], [0.54, 0.57], [0.51, 0.49], [0.43, 0.41], [0.45, 0.47], [0.45, 0.54], [0.44, 0.43]]}" + } + ], + "images": "42894_exterior_image_1_left#671", + "gt": "[[156, 103], [175, 112], [182, 146], [182, 127], [174, 103], [164, 88], [138, 73], [145, 84], [144, 98], [142, 78]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "70655_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the yellow mug to the center of the table', and now you are performing subtask 'move the yellow mug to the center of the table.'\nFrom start point [0.98, 0.13], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.98, 0.13], [0.84, 0.07], [0.76, 0.09], [0.66, 0.22], [0.58, 0.24], [0.62, 0.31], [0.55, 0.4], [0.54, 0.46], [0.54, 0.38], [0.54, 0.38]]}" + } + ], + "images": "70655_exterior_image_1_left#0", + "gt": "[[314, 23], [268, 13], [244, 17], [210, 40], [186, 44], [199, 55], [176, 72], [174, 82], [173, 69], [173, 69]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0201_user_0015_scene_0007_cfg_0001_035622060973#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'the medium pipette tip is inserted into the holder for medium pipette tips', and now you are performing subtask 'the medium pipette tip is inserted into the holder for medium pipette tips.'\nYou start from [0.38, 0.36]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.36], [0.36, 0.34], [0.33, 0.37], [0.36, 0.43], [0.38, 0.38], [0.44, 0.35], [0.43, 0.41], [0.45, 0.37], [0.42, 0.44], [0.4, 0.33]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0201_user_0015_scene_0007_cfg_0001_035622060973#0.jpg", + "gt": "[[246, 131], [228, 122], [212, 134], [230, 156], [240, 138], [281, 126], [275, 146], [288, 134], [266, 158], [254, 119]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "90957_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the green pen out of the white bowl', and now you are performing subtask 'take the green pen out of the white bowl.'\nStart at [0.2, 0.03]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.2, 0.03], [0.34, 0.19], [0.39, 0.29], [0.38, 0.35], [0.41, 0.35], [0.49, 0.47], [0.47, 0.42], [0.35, 0.37], [0.36, 0.46], [0.41, 0.5]]}" + } + ], + "images": "90957_exterior_image_1_left#0", + "gt": "[[64.5, 5.5], [109.5, 33.5], [123.5, 51.5], [122.5, 63.5], [132.5, 63.5], [155.5, 85.5], [151.5, 75.5], [110.5, 66.5], [115.5, 83.5], [130.5, 89.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "87631_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the green block on the yellow block', and now you are performing subtask 'put the green block on the yellow block.'\nStarting at [0.32, 0.01], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.01], [0.39, 0.2], [0.55, 0.4], [0.52, 0.47], [0.49, 0.34], [0.55, 0.53], [0.53, 0.66], [0.52, 0.34], [0.49, 0.31], [0.45, 0.41]]}" + } + ], + "images": "87631_exterior_image_1_left#0", + "gt": "[[102, 1], [125, 36], [176, 72], [167, 85], [158, 61], [175, 96], [171, 119], [167, 62], [158, 55], [145, 74]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0127_user_0020_scene_0006_cfg_0006_104122064161#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'plug in the power cord of the desk lamp, turn on the socket, and light up the desk lamp', and now you are performing subtask 'press the switch on the desk lamp to turn it on.'\nStart at [0.63, 0.29]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.29], [0.64, 0.28], [0.65, 0.28], [0.65, 0.3], [0.66, 0.29], [0.64, 0.28], [0.65, 0.27], [0.63, 0.28], [0.62, 0.26], [0.63, 0.27]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0127_user_0020_scene_0006_cfg_0006_104122064161#522.jpg", + "gt": "[[405, 105], [409, 101], [416, 102], [417, 107], [420, 103], [412, 100], [414, 96], [402, 99], [397, 92], [402, 97]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "66066_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the fabric', and now you are performing subtask 'pick up the fabric on the table.'\nYou begin at [0.7, 0.02]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.02], [0.64, 0.13], [0.68, 0.13], [0.65, 0.16], [0.64, 0.29], [0.7, 0.28], [0.69, 0.34], [0.69, 0.27], [0.68, 0.23], [0.68, 0.26]]}" + } + ], + "images": "66066_exterior_image_1_left#0", + "gt": "[[225, 3], [206, 24], [216, 24], [208, 28], [206, 52], [225, 50], [222, 61], [222, 48], [216, 42], [217, 46]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "15679_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the orange plush toy on the counter then close the drawer', and now you are performing subtask 'push the drawer handle to close the drawer.'\nStart at [0.65, 0.44]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.44], [0.54, 0.44], [0.47, 0.53], [0.47, 0.63], [0.55, 0.62], [0.54, 0.53], [0.53, 0.63], [0.6, 0.59], [0.56, 0.53], [0.59, 0.59]]}" + } + ], + "images": "15679_exterior_image_2_left#210", + "gt": "[[208, 80], [174, 79], [152, 95], [152, 114], [175, 112], [173, 95], [170, 114], [193, 106], [180, 95], [190, 107]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0041_user_0001_scene_0006_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'clean the tabletop with a sponge', and now you are performing subtask 'clean the tabletop with a sponge.'\nBegin at [0.63, 0.37] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.62, 0.4], [0.63, 0.37], [0.63, 0.42], [0.65, 0.43], [0.65, 0.49], [0.63, 0.45], [0.65, 0.47], [0.65, 0.53], [0.65, 0.44]]}" + } + ], + "images": "RH20T_cfg1_task_0041_user_0001_scene_0006_cfg_0001_038522062547#0", + "gt": "[[405, 134], [398, 143], [404, 132], [402, 150], [418, 155], [419, 178], [406, 162], [419, 169], [418, 191], [418, 158]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0029_user_0016_scene_0009_cfg_0004_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp the handle and open the drawer', and now you are performing subtask 'grasp the handle and open the drawer.'\nThe trajectory starts at [0.45, 0.0]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.0], [0.51, 0.01], [0.49, 0.02], [0.5, 0.02], [0.49, 0.09], [0.53, 0.24], [0.52, 0.25], [0.41, 0.14], [0.4, 0.0], [0.4, 0.0]]}" + } + ], + "images": "RH20T_cfg4_task_0029_user_0016_scene_0009_cfg_0004_104122062823#0", + "gt": "[[290, 0], [324, 2], [316, 6], [321, 8], [312, 32], [339, 85], [336, 90], [263, 49], [254, 0], [254, 0]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "64758_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the black clothes on the chair', and now you are performing subtask 'put the black clothes on the chair.'\nBegin at [0.42, 0.65] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.65], [0.52, 0.6], [0.54, 0.69], [0.61, 0.73], [0.57, 0.57], [0.51, 0.51], [0.43, 0.54], [0.38, 0.64], [0.49, 0.49], [0.52, 0.49]]}" + } + ], + "images": "64758_exterior_image_1_left#0", + "gt": "[[135.0, 116.5], [165.0, 108.5], [174.0, 123.5], [195.0, 131.5], [184.0, 102.5], [163.0, 92.5], [137.0, 96.5], [122.0, 114.5], [156.0, 88.5], [165.0, 88.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0064_user_0020_scene_0009_cfg_0007_104122061018#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the roll of paper down from the shelf', and now you are performing subtask 'take the roll of paper down from the shelf.'\nStart at [0.57, 0.09]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.09], [0.56, 0.23], [0.63, 0.43], [0.54, 0.63], [0.55, 0.37], [0.56, 0.4], [0.54, 0.35], [0.52, 0.36], [0.53, 0.3], [0.59, 0.19]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0064_user_0020_scene_0009_cfg_0007_104122061018#0.jpg", + "gt": "[[363, 31], [358, 84], [403, 155], [344, 226], [352, 134], [358, 143], [344, 126], [330, 128], [337, 108], [379, 69]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0044_user_0014_scene_0005_cfg_0007_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up a bag of things', and now you are performing subtask 'pick up a bag of things.'\nFrom start point [0.53, 0.32], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.32], [0.54, 0.35], [0.55, 0.42], [0.58, 0.51], [0.57, 0.63], [0.57, 0.57], [0.56, 0.53], [0.55, 0.4], [0.53, 0.31], [0.53, 0.26]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0044_user_0014_scene_0005_cfg_0007_104122064161#0.jpg", + "gt": "[[337, 115], [346, 126], [355, 152], [373, 184], [364, 226], [365, 207], [360, 192], [352, 144], [338, 112], [340, 95]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0096_user_0020_scene_0002_cfg_0006_037522061512#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take something out of a drawer', and now you are performing subtask 'pick up the item from the drawer, transfer the item from the drawer to the surface, place the item on the surface.'\nYour initial coordinate is [0.35, 0.32]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.32], [0.37, 0.36], [0.36, 0.41], [0.35, 0.33], [0.35, 0.3], [0.37, 0.15], [0.39, 0.14], [0.39, 0.18], [0.41, 0.27], [0.41, 0.27]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0096_user_0020_scene_0002_cfg_0006_037522061512#322.jpg", + "gt": "[[222, 115], [234, 131], [232, 146], [221, 120], [221, 108], [239, 55], [250, 52], [251, 66], [265, 98], [265, 98]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "9198_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the orange object from the yellow object and put it on the table', and now you are performing subtask 'pick up the orange object from the yellow object and put it on the table.'\nYour initial coordinate is [0.4, 0.0]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.0], [0.48, 0.13], [0.46, 0.21], [0.46, 0.26], [0.55, 0.24], [0.59, 0.31], [0.57, 0.37], [0.54, 0.18], [0.53, 0.14], [0.55, 0.19]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/9198_exterior_image_1_left#0.jpg", + "gt": "[[129, 0], [153, 24], [146, 38], [148, 46], [176, 44], [189, 55], [183, 67], [172, 32], [171, 26], [177, 35]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "36754_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the carrot plushie in the pot, then turn the tap nozzle to the left, put the carrot plushie back on the table', and now you are performing subtask 'pick up the carrot plushie from the pot.'\nStarting at [0.46, 0.49], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.49], [0.47, 0.48], [0.47, 0.5], [0.47, 0.47], [0.51, 0.44], [0.6, 0.42], [0.58, 0.47], [0.56, 0.41], [0.46, 0.46], [0.47, 0.53]]}" + } + ], + "images": "36754_exterior_image_1_left#158", + "gt": "[[146, 88], [150, 86], [149, 90], [152, 84], [163, 79], [192, 75], [187, 84], [180, 74], [148, 82], [149, 95]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "7985_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the towel', and now you are performing subtask 'fold the towel.'\nYour initial coordinate is [0.56, 0.48]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.56, 0.48], [0.55, 0.49], [0.55, 0.51], [0.53, 0.54], [0.53, 0.54], [0.53, 0.52], [0.54, 0.5], [0.59, 0.44], [0.59, 0.43], [0.58, 0.42]]}" + } + ], + "images": "7985_exterior_image_1_left#155", + "gt": "[[178, 87], [175, 89], [175, 91], [171, 97], [170, 97], [170, 94], [174, 90], [190, 79], [190, 77], [186, 76]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0065_user_0010_scene_0003_cfg_0004_104122062295#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with small Lego blocks for the nth time with n blocks', and now you are performing subtask 'pick up the red small Lego block from the surface, transfer the red small Lego block from the surface to the gray Lego base plate, place the red small Lego block on the gray Lego base plate.'\nStart at [0.41, 0.17]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.17], [0.37, 0.2], [0.38, 0.33], [0.41, 0.41], [0.45, 0.37], [0.28, 0.56], [0.3, 0.68], [0.35, 0.56], [0.41, 0.54], [0.4, 0.59]]}" + } + ], + "images": "RH20T_cfg4_task_0065_user_0010_scene_0003_cfg_0004_104122062295#0", + "gt": "[[262, 63], [236, 71], [243, 119], [264, 148], [286, 134], [179, 200], [189, 246], [226, 201], [260, 195], [257, 212]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "68797_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the peach cloth and sweatshirt from the backrest of the chair', and now you are performing subtask 'pick up the peach clothing from the back of the chair.'\nThe trajectory starts at [0.71, 0.18]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.71, 0.18], [0.66, 0.21], [0.63, 0.29], [0.63, 0.37], [0.66, 0.39], [0.52, 0.26], [0.51, 0.22], [0.51, 0.21], [0.55, 0.3], [0.6, 0.31]]}" + } + ], + "images": "68797_exterior_image_1_left#0", + "gt": "[[228.0, 32.5], [212.0, 38.5], [203.0, 51.5], [203.0, 66.5], [211.0, 70.5], [168.0, 46.5], [163.0, 40.5], [164.0, 37.5], [175.0, 53.5], [191.0, 56.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "39440_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the green bottle in the black bin', and now you are performing subtask 'put the green bottle in the black bin.'\nFrom start point [0.83, 0.22], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.83, 0.22], [0.71, 0.36], [0.67, 0.45], [0.65, 0.45], [0.64, 0.48], [0.61, 0.32], [0.52, 0.11], [0.51, 0.0], [0.53, 0.01], [0.57, 0.19]]}" + } + ], + "images": "39440_exterior_image_1_left#0", + "gt": "[[265, 39], [228, 65], [213, 81], [207, 81], [205, 86], [195, 58], [165, 20], [164, 0], [170, 1], [182, 34]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0029_user_0003_scene_0005_cfg_0002_105422061350#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp the handle and open the drawer', and now you are performing subtask 'pull the drawer out by pulling the handle.'\nYou start from [0.58, 0.24]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.24], [0.58, 0.23], [0.58, 0.24], [0.58, 0.26], [0.56, 0.29], [0.55, 0.34], [0.54, 0.36], [0.54, 0.37], [0.53, 0.37], [0.55, 0.36]]}" + } + ], + "images": "RH20T_cfg2_task_0029_user_0003_scene_0005_cfg_0002_105422061350#231", + "gt": "[[372, 85], [369, 84], [373, 85], [371, 92], [359, 103], [352, 123], [345, 130], [347, 132], [341, 132], [349, 128]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "10383_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'partially unfold the towel on the table', and now you are performing subtask 'fold the towel partially unfolded.'\nYour starting position is [0.78, 0.29]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.77, 0.29], [0.62, 0.41], [0.63, 0.41], [0.62, 0.41], [0.62, 0.47], [0.64, 0.63], [0.63, 0.61], [0.63, 0.63], [0.54, 0.67], [0.55, 0.6]]}" + } + ], + "images": "10383_exterior_image_1_left#313", + "gt": "[[248, 52], [198, 74], [203, 73], [199, 74], [198, 84], [205, 113], [201, 110], [201, 113], [173, 121], [176, 108]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_819112070044#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn them on', and now you are performing subtask 'press the first button on the right side of the power strip, press the second button on the right side of the power strip.'\nBegin at [0.8, 0.51] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.8, 0.51], [0.75, 0.5], [0.75, 0.53], [0.76, 0.56], [0.75, 0.56], [0.75, 0.56], [0.75, 0.57], [0.75, 0.55], [0.76, 0.53], [0.76, 0.53]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_819112070044#113.jpg", + "gt": "[[509, 184], [482, 180], [480, 191], [484, 201], [482, 201], [483, 202], [480, 204], [480, 199], [484, 192], [484, 192]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "83331_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the blue object from the cabinet and place it in the open drawer', and now you are performing subtask 'remove the blue object from the cabinet and place it in the open drawer.'\nYou start from [0.62, 0.45]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.45], [0.58, 0.22], [0.54, 0.16], [0.47, 0.24], [0.53, 0.23], [0.57, 0.33], [0.6, 0.5], [0.57, 0.73], [0.62, 0.67], [0.62, 0.67]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/83331_exterior_image_2_left#0.jpg", + "gt": "[[197, 81], [186, 40], [174, 28], [152, 43], [170, 42], [184, 60], [192, 90], [183, 131], [200, 120], [200, 120]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "68771_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the marker from the cup and put it on the left side of the table', and now you are performing subtask 'remove the marker from the cup and put it on the left side of the table.'\nWith start location [0.75, 0.23], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.75, 0.22], [0.56, 0.21], [0.54, 0.26], [0.58, 0.36], [0.57, 0.43], [0.55, 0.4], [0.55, 0.36], [0.53, 0.28], [0.54, 0.44], [0.53, 0.42]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/68771_exterior_image_1_left#0.jpg", + "gt": "[[240.5, 40.5], [180.5, 38.5], [173.5, 47.5], [184.5, 64.5], [183.5, 77.5], [175.5, 72.5], [177.5, 65.5], [170.5, 49.5], [172.5, 79.5], [168.5, 75.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "44059_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the blue and white plush toy on the right side of the table', and now you are performing subtask 'place the blue and white plush toy on the right side of the table.'\nYou start from [0.96, 0.04]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.96, 0.04], [0.79, 0.37], [0.61, 0.53], [0.56, 0.65], [0.53, 0.64], [0.61, 0.54], [0.64, 0.56], [0.68, 0.63], [0.62, 0.53], [0.61, 0.45]]}" + } + ], + "images": "44059_exterior_image_2_left#0", + "gt": "[[308, 7], [254, 67], [196, 96], [178, 117], [171, 115], [194, 98], [204, 101], [216, 114], [198, 95], [196, 81]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "5432_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker into the cup', and now you are performing subtask 'put the marker into the cup.'\nYour starting position is [0.14, 0.0]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.14, 0.0], [0.44, 0.29], [0.45, 0.41], [0.39, 0.52], [0.4, 0.68], [0.39, 0.52], [0.35, 0.43], [0.32, 0.41], [0.31, 0.49], [0.33, 0.47]]}" + } + ], + "images": "5432_exterior_image_2_left#0", + "gt": "[[45, 0], [141, 53], [143, 73], [125, 94], [129, 123], [125, 93], [112, 77], [102, 74], [100, 89], [106, 85]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "10901_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the marker and put it in the cup', and now you are performing subtask 'pick up the marker and put it in the cup.'\nYou begin at [0.49, 0.53]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.53], [0.38, 0.63], [0.4, 0.67], [0.39, 0.72], [0.38, 0.66], [0.38, 0.69], [0.39, 0.57], [0.35, 0.54], [0.38, 0.57], [0.39, 0.53]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/10901_exterior_image_2_left#0.jpg", + "gt": "[[158, 96], [121, 114], [129, 120], [126, 130], [123, 118], [120, 124], [125, 103], [112, 98], [123, 102], [125, 96]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0061_user_0001_scene_0005_cfg_0001_035622060973#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'pick up the small square from the gift box, transfer the small square from the gift box to above the fabric, place the small square on the fabric.'\nFrom start point [0.37, 0.35], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.35], [0.37, 0.32], [0.31, 0.36], [0.31, 0.33], [0.3, 0.33], [0.32, 0.3], [0.34, 0.32], [0.32, 0.43], [0.34, 0.39], [0.33, 0.55]]}" + } + ], + "images": "RH20T_cfg1_task_0061_user_0001_scene_0005_cfg_0001_035622060973#274", + "gt": "[[234, 125], [236, 114], [201, 131], [200, 119], [190, 119], [203, 108], [215, 117], [204, 153], [216, 141], [210, 198]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0065_user_0014_scene_0002_cfg_0001_819112070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with small Lego blocks for the nth time with n blocks', and now you are performing subtask 'pick up the pink block on the desktop, transfer the pink block from the desktop to the building area, place the pink block in the building area.'\nStart at [0.62, 0.47]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.47], [0.63, 0.5], [0.66, 0.51], [0.66, 0.5], [0.67, 0.51], [0.66, 0.49], [0.67, 0.49], [0.65, 0.54], [0.66, 0.59], [0.65, 0.53]]}" + } + ], + "images": "RH20T_cfg1_task_0065_user_0014_scene_0002_cfg_0001_819112070044#0", + "gt": "[[398, 168], [406, 180], [425, 184], [423, 181], [427, 182], [423, 178], [427, 177], [419, 196], [422, 214], [413, 191]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0096_user_0014_scene_0001_cfg_0001_039422060546#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take something out of a drawer', and now you are performing subtask 'pick up the object from the drawer, transfer the object from the drawer to the top of the table, place the object on the table.'\nWith start location [0.46, 0.08], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.08], [0.46, 0.09], [0.45, 0.12], [0.44, 0.07], [0.45, 0.01], [0.44, 0.0], [0.46, 0.3], [0.45, 0.38], [0.44, 0.39], [0.4, 0.12]]}" + } + ], + "images": "RH20T_cfg1_task_0096_user_0014_scene_0001_cfg_0001_039422060546#149", + "gt": "[[293, 29], [295, 34], [286, 43], [284, 26], [291, 5], [284, 1], [294, 109], [289, 137], [282, 139], [259, 45]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "52926_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the two pillows and move them to the right side of the couch', and now you are performing subtask 'transfer the second pillow from the sofa to the right side of the sofa.'\nBegin at [0.48, 0.67] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.67], [0.44, 0.76], [0.47, 0.81], [0.45, 0.8], [0.45, 0.83], [0.43, 0.81], [0.48, 0.79], [0.53, 0.67], [0.59, 0.65], [0.51, 0.68]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/52926_exterior_image_1_left#245.jpg", + "gt": "[[153, 121], [140, 137], [149, 145], [143, 144], [143, 149], [138, 145], [153, 143], [171, 121], [189, 117], [164, 123]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_750612070853#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn them on', and now you are performing subtask 'press the third button on the right side of the power strip, press the fourth button on the right side of the power strip.'\nYou start from [0.4, 0.32]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.32], [0.41, 0.34], [0.41, 0.34], [0.42, 0.35], [0.41, 0.32], [0.41, 0.3], [0.41, 0.28], [0.41, 0.3], [0.4, 0.34], [0.4, 0.37]]}" + } + ], + "images": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_750612070853#231", + "gt": "[[253, 116], [262, 124], [264, 123], [266, 127], [262, 115], [264, 107], [261, 102], [261, 109], [256, 123], [257, 134]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0128_user_0007_scene_0003_cfg_0002_104422070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'plug in the USB drive to the port', and now you are performing subtask 'plug in the USB drive to the port.'\nBegin at [0.5, 0.52] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.52], [0.52, 0.41], [0.52, 0.45], [0.53, 0.4], [0.55, 0.41], [0.6, 0.51], [0.59, 0.49], [0.59, 0.51], [0.57, 0.47], [0.52, 0.5]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg2_task_0128_user_0007_scene_0003_cfg_0002_104422070044#0.jpg", + "gt": "[[317, 186], [332, 148], [335, 162], [340, 145], [352, 146], [381, 182], [380, 175], [377, 184], [365, 171], [331, 181]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0073_user_0001_scene_0003_cfg_0001_039422060546#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the things in the drawer', and now you are performing subtask 'pick up the block on the drawer, transfer the block from above the drawer to inside the drawer, place the block in the drawer.'\nYour starting position is [0.41, 0.35]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.35], [0.4, 0.27], [0.26, 0.23], [0.22, 0.28], [0.24, 0.32], [0.2, 0.42], [0.22, 0.45], [0.23, 0.39], [0.31, 0.26], [0.33, 0.38]]}" + } + ], + "images": "RH20T_cfg1_task_0073_user_0001_scene_0003_cfg_0001_039422060546#0", + "gt": "[[261, 126], [259, 97], [167, 84], [144, 100], [151, 114], [129, 151], [143, 163], [146, 142], [200, 93], [214, 137]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "44324_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unstack the cans and put them beside each other', and now you are performing subtask 'place the second can next to the other cans.'\nFrom start point [0.51, 0.54], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.54], [0.53, 0.51], [0.57, 0.54], [0.6, 0.53], [0.6, 0.56], [0.58, 0.53], [0.58, 0.56], [0.56, 0.51], [0.54, 0.52], [0.61, 0.51]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/44324_exterior_image_1_left#236.jpg", + "gt": "[[164, 97], [171, 91], [183, 98], [193, 95], [192, 101], [186, 96], [185, 100], [178, 91], [172, 94], [196, 91]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "22982_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the white cloth on the orange chair', and now you are performing subtask 'put the white cloth on the orange chair.'\nYour initial coordinate is [0.61, 0.24]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.61, 0.24], [0.51, 0.39], [0.48, 0.82], [0.44, 0.81], [0.39, 0.56], [0.39, 0.28], [0.49, 0.11], [0.59, 0.35], [0.47, 0.26], [0.47, 0.26]]}" + } + ], + "images": "22982_exterior_image_1_left#0", + "gt": "[[194, 43], [163, 70], [153, 147], [142, 146], [125, 100], [126, 50], [158, 19], [188, 63], [150, 47], [150, 47]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "64549_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the orange ball to the bottom of the cabinet', and now you are performing subtask 'move the orange ball to the bottom of the cabinet.'\nYour initial coordinate is [0.86, 0.07]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.86, 0.07], [0.45, 0.24], [0.4, 0.31], [0.39, 0.36], [0.49, 0.37], [0.49, 0.39], [0.47, 0.43], [0.45, 0.42], [0.45, 0.38], [0.41, 0.28]]}" + } + ], + "images": "64549_exterior_image_2_left#0", + "gt": "[[274, 12], [143, 44], [128, 55], [125, 64], [156, 67], [158, 70], [152, 77], [145, 76], [145, 68], [131, 50]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "65014_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the toaster', and now you are performing subtask 'turn on the toaster.'\nYou begin at [0.02, 0.27]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.02, 0.27], [0.19, 0.24], [0.35, 0.26], [0.41, 0.32], [0.42, 0.49], [0.29, 0.42], [0.27, 0.33], [0.22, 0.27], [0.27, 0.3], [0.23, 0.28]]}" + } + ], + "images": "65014_exterior_image_1_left#0", + "gt": "[[6, 49], [61, 44], [112, 47], [132, 58], [135, 88], [92, 75], [85, 60], [72, 49], [85, 54], [75, 50]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "20669_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'slide the green towel on the table', and now you are performing subtask 'slide the green towel on the table.'\nStarting at [0.4, 0.02], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.02], [0.43, 0.11], [0.47, 0.35], [0.48, 0.47], [0.45, 0.51], [0.43, 0.58], [0.44, 0.58], [0.45, 0.57], [0.48, 0.45], [0.5, 0.44]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/20669_exterior_image_1_left#0.jpg", + "gt": "[[128, 3], [138, 19], [152, 63], [153, 85], [143, 92], [137, 104], [140, 105], [143, 102], [154, 81], [160, 79]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0094_user_0007_scene_0009_cfg_0005_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'press the cardboard box on the side, press the cardboard box on the side.'\nYour starting position is [0.48, 0.14]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.14], [0.48, 0.15], [0.5, 0.24], [0.53, 0.4], [0.53, 0.43], [0.52, 0.47], [0.52, 0.52], [0.52, 0.5], [0.53, 0.44], [0.54, 0.4]]}" + } + ], + "images": "RH20T_cfg5_task_0094_user_0007_scene_0009_cfg_0005_104122061850#0", + "gt": "[[305, 50], [309, 54], [321, 85], [338, 144], [338, 154], [333, 170], [336, 188], [334, 179], [338, 157], [343, 145]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0079_user_0015_scene_0003_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the microwave door', and now you are performing subtask 'pull the handle of the microwave door to open the door.'\nStarting at [0.63, 0.37], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.7, 0.45], [0.72, 0.49], [0.66, 0.39], [0.64, 0.34], [0.67, 0.39], [0.62, 0.31], [0.61, 0.11], [0.57, 0.12], [0.62, 0.11]]}" + } + ], + "images": "RH20T_cfg1_task_0079_user_0015_scene_0003_cfg_0001_038522062547#0", + "gt": "[[405, 132], [450, 161], [458, 178], [420, 142], [411, 123], [430, 141], [398, 110], [391, 39], [364, 42], [394, 39]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0085_user_0007_scene_0006_cfg_0005_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the bubble ring from the assembled bubble ring and ball and place it on the table', and now you are performing subtask 'pick up the bubble ring that is assembled with the ball.'\nStarting at [0.45, 0.42], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.42], [0.47, 0.44], [0.48, 0.5], [0.5, 0.54], [0.51, 0.72], [0.5, 0.75], [0.49, 0.77], [0.5, 0.81], [0.51, 0.78], [0.51, 0.79]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0085_user_0007_scene_0006_cfg_0005_037522062165#0.jpg", + "gt": "[[289, 152], [300, 158], [309, 181], [319, 196], [327, 259], [318, 269], [316, 279], [323, 290], [324, 282], [326, 284]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "26153_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the salt cellar to the right of the table', and now you are performing subtask 'move the salt cellar to the right of the table.'\nBegin at [0.84, 0.17] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.84, 0.17], [0.73, 0.3], [0.63, 0.46], [0.61, 0.55], [0.66, 0.32], [0.7, 0.28], [0.76, 0.31], [0.76, 0.46], [0.76, 0.33], [0.76, 0.33]]}" + } + ], + "images": "26153_exterior_image_2_left#0", + "gt": "[[268, 31], [232, 54], [201, 83], [196, 99], [212, 58], [225, 50], [244, 56], [242, 82], [244, 59], [244, 59]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "20551_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the marker from the desk and put it into the pot', and now you are performing subtask 'pick up the marker from the desk and put it into the pot.'\nWith start location [0.23, 0.09], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.22, 0.09], [0.3, 0.22], [0.56, 0.38], [0.64, 0.58], [0.61, 0.6], [0.62, 0.43], [0.58, 0.28], [0.51, 0.36], [0.52, 0.47], [0.51, 0.39]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/20551_exterior_image_1_left#0.jpg", + "gt": "[[72.0, 17.0], [97.0, 39.0], [179.0, 68.0], [205.0, 105.0], [196.0, 108.0], [197.0, 78.0], [187.0, 50.0], [162.0, 64.0], [168.0, 85.0], [164.0, 70.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "4936_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer, place the black clip in the open drawer', and now you are performing subtask 'pick up the black clip from the countertop.'\nYour initial coordinate is [0.4, 0.22]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.22], [0.37, 0.32], [0.41, 0.45], [0.45, 0.48], [0.49, 0.43], [0.42, 0.5], [0.41, 0.44], [0.43, 0.41], [0.43, 0.3], [0.47, 0.27]]}" + } + ], + "images": "4936_exterior_image_1_left#0", + "gt": "[[127.5, 40.0], [118.5, 58.0], [131.5, 81.0], [142.5, 86.0], [155.5, 78.0], [134.5, 90.0], [130.5, 80.0], [137.5, 73.0], [138.5, 54.0], [151.5, 49.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "8647_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the bottle from the table and put it in the clear container', and now you are performing subtask 'pick up the bottle from the table and put it in the clear container.'\nFrom start point [0.42, 0.12], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.12], [0.52, 0.31], [0.58, 0.54], [0.56, 0.6], [0.59, 0.66], [0.59, 0.77], [0.57, 0.73], [0.55, 0.59], [0.49, 0.61], [0.52, 0.57]]}" + } + ], + "images": "8647_exterior_image_2_left#0", + "gt": "[[133, 21], [167, 55], [187, 98], [179, 108], [189, 119], [190, 139], [181, 131], [177, 107], [156, 110], [165, 102]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "22767_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the lid and put it on the open cup', and now you are performing subtask 'pick up the lid and put it on the open cup.'\nYou begin at [0.69, 0.55]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.55], [0.63, 0.59], [0.63, 0.69], [0.66, 0.67], [0.62, 0.88], [0.6, 0.88], [0.61, 0.79], [0.56, 0.86], [0.58, 0.91], [0.62, 0.73]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/22767_exterior_image_2_left#0.jpg", + "gt": "[[220, 99], [203, 107], [203, 125], [212, 121], [197, 159], [192, 158], [195, 142], [179, 155], [185, 164], [197, 132]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "34784_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the purple plush toy on the beige plate', and now you are performing subtask 'put the purple plush toy on the beige plate.'\nYou start from [0.68, 0.53]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.53], [0.75, 0.63], [0.73, 0.62], [0.73, 0.58], [0.74, 0.51], [0.73, 0.47], [0.71, 0.48], [0.62, 0.59], [0.61, 0.66], [0.65, 0.6]]}" + } + ], + "images": "34784_exterior_image_2_left#0", + "gt": "[[216, 96], [241, 113], [234, 112], [232, 104], [237, 92], [235, 85], [226, 86], [198, 106], [194, 118], [209, 108]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0066_user_0015_scene_0001_cfg_0004_038522062288#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks', and now you are performing subtask 'build with large Megabloks.'\nYou start from [0.37, 0.33]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.33], [0.35, 0.43], [0.37, 0.36], [0.35, 0.36], [0.37, 0.47], [0.46, 0.71], [0.41, 0.56], [0.39, 0.61], [0.39, 0.55], [0.3, 0.52]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0066_user_0015_scene_0001_cfg_0004_038522062288#0.jpg", + "gt": "[[236, 118], [222, 156], [238, 128], [227, 129], [234, 168], [292, 257], [260, 202], [252, 221], [247, 199], [189, 189]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0058_user_0014_scene_0003_cfg_0004_036422060909#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn off the power strips', and now you are performing subtask 'press the button on the third power strip from the left, press the button on the right power strip.'\nBegin at [0.53, 0.37] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.37], [0.51, 0.36], [0.5, 0.36], [0.52, 0.38], [0.52, 0.4], [0.52, 0.38], [0.5, 0.34], [0.49, 0.28], [0.49, 0.27], [0.49, 0.29]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0058_user_0014_scene_0003_cfg_0004_036422060909#188.jpg", + "gt": "[[338, 134], [329, 130], [323, 131], [330, 137], [332, 144], [331, 137], [320, 123], [313, 99], [312, 98], [312, 103]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "26514_exterior_image_1_left#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the towel in half then fold it in half again', and now you are performing subtask 'fold the towel in half and place it on the table.'\nYou start from [0.49, 0.51]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.51], [0.5, 0.73], [0.49, 0.85], [0.51, 0.83], [0.51, 0.86], [0.55, 0.62], [0.53, 0.59], [0.48, 0.57], [0.44, 0.61], [0.44, 0.59]]}" + } + ], + "images": "26514_exterior_image_1_left#440", + "gt": "[[158, 91], [161, 132], [157, 153], [164, 149], [164, 155], [175, 112], [170, 106], [153, 102], [142, 110], [142, 107]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0029_user_0020_scene_0002_cfg_0007_037522061512#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp the handle and open the drawer', and now you are performing subtask 'grasp the handle and open the drawer.'\nYou begin at [0.41, 0.24]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.24], [0.39, 0.28], [0.35, 0.28], [0.31, 0.31], [0.29, 0.35], [0.28, 0.41], [0.29, 0.34], [0.33, 0.26], [0.32, 0.13], [0.34, 0.15]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0029_user_0020_scene_0002_cfg_0007_037522061512#0.jpg", + "gt": "[[261, 86], [249, 99], [221, 99], [197, 111], [184, 125], [179, 146], [187, 122], [210, 93], [208, 46], [217, 54]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "25326_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the lid on the clear container on the left then turn on the extension cable', and now you are performing subtask 'put the lid on the clear container.'\nBegin at [0.38, 0.0] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.47, 0.02], [0.53, 0.1], [0.53, 0.48], [0.47, 0.35], [0.41, 0.33], [0.36, 0.36], [0.37, 0.32], [0.33, 0.29], [0.33, 0.29]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/25326_exterior_image_2_left#0.jpg", + "gt": "[[122, 0], [150, 4], [170, 18], [170, 86], [151, 63], [130, 59], [114, 65], [117, 58], [107, 52], [107, 52]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "44145_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the purple object in the bowl', and now you are performing subtask 'put the purple object in the bowl.'\nFrom start point [0.35, 0.54], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.54], [0.51, 0.59], [0.55, 0.7], [0.54, 0.78], [0.6, 0.61], [0.69, 0.51], [0.73, 0.52], [0.71, 0.64], [0.74, 0.7], [0.78, 0.53]]}" + } + ], + "images": "44145_exterior_image_2_left#0", + "gt": "[[111, 98], [163, 107], [176, 126], [174, 140], [193, 110], [221, 92], [234, 94], [227, 115], [238, 126], [249, 96]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0094_user_0014_scene_0003_cfg_0004_104422070011#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'push the lid on the bottom of the box.'\nYour initial coordinate is [0.46, 0.5]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.5], [0.43, 0.45], [0.43, 0.51], [0.44, 0.51], [0.45, 0.33], [0.41, 0.27], [0.41, 0.29], [0.43, 0.26], [0.48, 0.33], [0.53, 0.35]]}" + } + ], + "images": "RH20T_cfg4_task_0094_user_0014_scene_0003_cfg_0004_104422070011#365", + "gt": "[[292, 181], [278, 162], [273, 182], [280, 183], [285, 120], [262, 96], [262, 105], [276, 94], [308, 118], [341, 126]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "79766_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'straighten the cable', and now you are performing subtask 'place the thread on the table.'\nThe trajectory starts at [0.78, 0.01]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.78, 0.01], [0.74, 0.18], [0.58, 0.33], [0.58, 0.38], [0.57, 0.5], [0.62, 0.45], [0.8, 0.43], [0.77, 0.39], [0.73, 0.43], [0.73, 0.48]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/79766_exterior_image_1_left#0.jpg", + "gt": "[[250, 1], [236, 33], [187, 60], [185, 69], [182, 90], [197, 81], [255, 77], [246, 71], [232, 78], [235, 87]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0066_user_0014_scene_0007_cfg_0007_104122064161#TrajectoryPrediction_Task#10", + "task": "TrajectoryPrediction_Task", + "p_index": 10, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'place the building blocks on the pile of building blocks on the building board, pick up the building blocks on the table, transfer the building blocks from the desktop to the pile of building blocks on the building board, place the building blocks on the pile of building blocks on the building board.'\nYour starting position is [0.58, 0.48]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.48], [0.61, 0.41], [0.65, 0.38], [0.67, 0.43], [0.67, 0.39], [0.58, 0.38], [0.56, 0.43], [0.53, 0.42], [0.54, 0.45], [0.55, 0.4]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0066_user_0014_scene_0007_cfg_0007_104122064161#709.jpg", + "gt": "[[371, 173], [388, 147], [414, 137], [427, 154], [427, 139], [373, 138], [356, 153], [341, 152], [344, 163], [350, 145]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0076_user_0010_scene_0001_cfg_0001_750612070853#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stacking blocks', and now you are performing subtask 'pick up the first small Lego block from the table, transfer the Lego from the table to the Lego board, place the Lego on the Lego board.'\nYou start from [0.52, 0.29]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.29], [0.32, 0.54], [0.29, 0.54], [0.3, 0.57], [0.35, 0.5], [0.29, 0.68], [0.3, 0.61], [0.34, 0.68], [0.45, 0.64], [0.47, 0.71]]}" + } + ], + "images": "RH20T_cfg1_task_0076_user_0010_scene_0001_cfg_0001_750612070853#0", + "gt": "[[333, 105], [207, 196], [185, 195], [190, 206], [227, 181], [186, 243], [190, 220], [215, 243], [289, 232], [299, 255]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "10503_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the white towel then set it down again', and now you are performing subtask 'press the white towel.'\nStart at [0.44, 0.52]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.52], [0.47, 0.32], [0.43, 0.41], [0.42, 0.6], [0.4, 0.57], [0.43, 0.58], [0.42, 0.61], [0.44, 0.63], [0.42, 0.5], [0.4, 0.49]]}" + } + ], + "images": "10503_exterior_image_2_left#172", + "gt": "[[142, 94], [151, 57], [139, 74], [133, 108], [129, 103], [137, 104], [134, 109], [140, 114], [133, 90], [129, 89]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0094_user_0001_scene_0009_cfg_0001_038522062547#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'pull the lid of the box upwards to open the box.'\nYou start from [0.66, 0.82]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.66, 0.82], [0.59, 0.73], [0.57, 0.75], [0.55, 0.66], [0.54, 0.66], [0.58, 0.77], [0.55, 0.64], [0.5, 0.61], [0.58, 0.69], [0.6, 0.49]]}" + } + ], + "images": "RH20T_cfg1_task_0094_user_0001_scene_0009_cfg_0001_038522062547#466", + "gt": "[[423, 296], [378, 261], [363, 270], [355, 237], [343, 237], [369, 276], [354, 230], [318, 218], [371, 248], [384, 175]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0003_user_0014_scene_0006_cfg_0005_105422061350#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the red button on the black and white base, press the green button on the gray and white base.'\nYour initial coordinate is [0.35, 0.08]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.08], [0.36, 0.16], [0.39, 0.16], [0.38, 0.14], [0.37, 0.15], [0.37, 0.14], [0.37, 0.18], [0.36, 0.18], [0.38, 0.19], [0.38, 0.15]]}" + } + ], + "images": "RH20T_cfg5_task_0003_user_0014_scene_0006_cfg_0005_105422061350#289", + "gt": "[[227, 30], [230, 59], [247, 57], [242, 50], [237, 54], [236, 51], [237, 66], [233, 66], [240, 67], [242, 55]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0077_user_0014_scene_0004_cfg_0006_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'insert the red Lego block onto the gray base plate', and now you are performing subtask 'insert the red Lego block onto the gray base plate.'\nThe trajectory starts at [0.53, 0.31]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.31], [0.6, 0.32], [0.63, 0.37], [0.65, 0.44], [0.64, 0.54], [0.61, 0.5], [0.58, 0.49], [0.57, 0.51], [0.57, 0.49], [0.58, 0.49]]}" + } + ], + "images": "RH20T_cfg6_task_0077_user_0014_scene_0004_cfg_0006_104122064161#0", + "gt": "[[338, 111], [383, 117], [401, 133], [413, 158], [407, 196], [389, 180], [369, 177], [368, 183], [365, 175], [369, 178]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_036422060215#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the block on the other blocks, pick up the block on the plane, transfer the block from the plane to the other blocks, place the block on the other blocks.'\nStart at [0.36, 0.3]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.3], [0.3, 0.35], [0.24, 0.49], [0.25, 0.54], [0.29, 0.47], [0.29, 0.34], [0.32, 0.23], [0.41, 0.2], [0.43, 0.26], [0.44, 0.17]]}" + } + ], + "images": "RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_036422060215#984", + "gt": "[[229, 109], [190, 127], [155, 175], [159, 195], [183, 171], [188, 123], [208, 82], [261, 71], [273, 92], [282, 60]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0037_user_0007_scene_0004_cfg_0005_104122063678#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the third block on top of the second block, pick up the fourth block from the surface, place the fourth block on top of the third block.'\nYou start from [0.54, 0.42]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.42], [0.56, 0.31], [0.56, 0.36], [0.52, 0.41], [0.56, 0.29], [0.57, 0.28], [0.56, 0.25], [0.58, 0.3], [0.56, 0.36], [0.58, 0.29]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0037_user_0007_scene_0004_cfg_0005_104122063678#924.jpg", + "gt": "[[345, 150], [361, 111], [357, 131], [333, 147], [357, 103], [365, 101], [361, 90], [373, 108], [358, 129], [371, 104]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "66771_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the yellow item and put it on the right side of the table', and now you are performing subtask 'pick up the yellow item and put it on the right side of the table.'\nFrom start point [0.23, 0.01], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.23, 0.01], [0.32, 0.2], [0.35, 0.5], [0.37, 0.52], [0.43, 0.38], [0.48, 0.36], [0.57, 0.41], [0.59, 0.52], [0.52, 0.28], [0.39, 0.29]]}" + } + ], + "images": "66771_exterior_image_1_left#0", + "gt": "[[72.5, 2.5], [101.5, 35.5], [113.5, 90.5], [118.5, 93.5], [137.5, 67.5], [153.5, 64.5], [181.5, 74.5], [187.5, 93.5], [167.5, 50.5], [123.5, 52.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "52193_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'cover the pot', and now you are performing subtask 'cover the pot.'\nFrom start point [0.53, 0.1], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.1], [0.47, 0.13], [0.42, 0.24], [0.41, 0.42], [0.4, 0.28], [0.49, 0.26], [0.49, 0.37], [0.52, 0.39], [0.47, 0.41], [0.46, 0.38]]}" + } + ], + "images": "52193_exterior_image_2_left#0", + "gt": "[[169, 18], [152, 24], [134, 43], [132, 75], [127, 50], [156, 46], [156, 67], [166, 71], [149, 74], [146, 68]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0058_user_0015_scene_0003_cfg_0001_819112070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn off the power strips', and now you are performing subtask 'press the first button on the power socket, press the second button on the power socket.'\nYour initial coordinate is [0.62, 0.47]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.47], [0.63, 0.47], [0.71, 0.48], [0.77, 0.5], [0.78, 0.49], [0.79, 0.51], [0.78, 0.56], [0.79, 0.54], [0.76, 0.53], [0.75, 0.54]]}" + } + ], + "images": "RH20T_cfg1_task_0058_user_0015_scene_0003_cfg_0001_819112070044#0", + "gt": "[[400, 168], [401, 170], [452, 172], [495, 180], [497, 178], [506, 182], [501, 203], [505, 195], [488, 192], [483, 194]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "70647_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the sweatshirt from the backrest of the black chair', and now you are performing subtask 'remove the sweatshirt from the backrest of the black chair.'\nFrom start point [0.55, 0.34], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.34], [0.84, 0.53], [0.89, 0.77], [0.83, 0.54], [0.69, 0.26], [0.81, 0.38], [0.8, 0.46], [0.76, 0.51], [0.77, 0.45], [0.77, 0.45]]}" + } + ], + "images": "70647_exterior_image_2_left#0", + "gt": "[[176, 62], [270, 96], [286, 139], [266, 97], [220, 46], [260, 69], [256, 83], [243, 91], [248, 81], [248, 81]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "51834_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick the marker from the plate and put it on the table', and now you are performing subtask 'pick the marker from the plate and put it on the table.'\nStart at [0.36, 0.3]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.3], [0.42, 0.31], [0.45, 0.34], [0.5, 0.67], [0.5, 0.56], [0.49, 0.5], [0.47, 0.46], [0.45, 0.47], [0.45, 0.55], [0.46, 0.54]]}" + } + ], + "images": "51834_exterior_image_1_left#0", + "gt": "[[115.5, 54.5], [133.5, 55.5], [142.5, 61.5], [159.5, 120.5], [159.5, 101.5], [156.5, 89.5], [149.5, 82.5], [142.5, 84.5], [145.5, 99.5], [147.5, 96.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "69741_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the grey salt cellar then set it down again', and now you are performing subtask 'pick up the grey salt cellar then set it down again.'\nThe trajectory starts at [0.49, 0.0]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.0], [0.5, 0.14], [0.56, 0.17], [0.63, 0.4], [0.57, 0.58], [0.6, 0.48], [0.61, 0.52], [0.59, 0.48], [0.64, 0.49], [0.61, 0.56]]}" + } + ], + "images": "69741_exterior_image_1_left#0", + "gt": "[[157, 0], [160, 26], [180, 31], [201, 72], [183, 104], [192, 87], [195, 94], [189, 86], [204, 89], [195, 101]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "48888_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'after opening and then closing the flask lid, and then opening it again', and now you are performing subtask 'press the flask lid closed.'\nFrom start point [0.78, 0.19], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.78, 0.19], [0.7, 0.24], [0.67, 0.31], [0.68, 0.31], [0.67, 0.37], [0.71, 0.49], [0.67, 0.41], [0.65, 0.36], [0.68, 0.32], [0.74, 0.41]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/48888_exterior_image_1_left#0.jpg", + "gt": "[[249, 34], [224, 43], [214, 56], [219, 56], [215, 66], [227, 88], [213, 74], [209, 64], [217, 58], [238, 73]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "75830_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the orange packet from the green box', and now you are performing subtask 'remove the orange packet from the green box.'\nYour starting position is [0.65, 0.31]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.31], [0.52, 0.32], [0.49, 0.39], [0.55, 0.27], [0.5, 0.11], [0.44, 0.08], [0.38, 0.48], [0.43, 0.38], [0.53, 0.41], [0.53, 0.41]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/75830_exterior_image_2_left#0.jpg", + "gt": "[[209, 56], [166, 57], [157, 70], [175, 48], [161, 19], [140, 14], [123, 87], [139, 69], [170, 73], [170, 73]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0104_user_0020_scene_0004_cfg_0006_104122061602#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'drag the plate back after holding it down', and now you are performing subtask 'drag the plate back after holding it down.'\nYou begin at [0.7, 0.29]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.29], [0.64, 0.32], [0.63, 0.3], [0.62, 0.31], [0.62, 0.36], [0.76, 0.6], [0.74, 0.48], [0.63, 0.37], [0.64, 0.32], [0.66, 0.29]]}" + } + ], + "images": "RH20T_cfg6_task_0104_user_0020_scene_0004_cfg_0006_104122061602#0", + "gt": "[[445, 106], [407, 117], [403, 107], [394, 112], [398, 128], [489, 215], [476, 172], [403, 132], [407, 114], [423, 104]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0037_user_0007_scene_0004_cfg_0005_036422060215#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the third block on top of the second block, pick up the fourth block from the surface, place the fourth block on top of the third block.'\nBegin at [0.43, 0.3] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.3], [0.43, 0.26], [0.41, 0.24], [0.32, 0.28], [0.27, 0.45], [0.28, 0.48], [0.35, 0.24], [0.42, 0.17], [0.44, 0.22], [0.45, 0.15]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0037_user_0007_scene_0004_cfg_0005_036422060215#915.jpg", + "gt": "[[274, 109], [277, 92], [265, 86], [207, 102], [173, 162], [177, 172], [222, 86], [267, 61], [284, 80], [289, 53]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "74887_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'throw the building block tower onto the table, then push the green building blocks in the tower to the edge of the table', and now you are performing subtask 'push the green block from the block tower to the edge of the table.'\nYou start from [0.36, 0.56]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.56], [0.38, 0.49], [0.41, 0.59], [0.4, 0.62], [0.38, 0.62], [0.35, 0.64], [0.45, 0.57], [0.45, 0.55], [0.44, 0.52], [0.4, 0.53]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/74887_exterior_image_2_left#127.jpg", + "gt": "[[116, 100], [123, 89], [132, 107], [129, 111], [121, 111], [112, 115], [143, 103], [144, 99], [140, 94], [129, 95]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122061330#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the block on the surface, transfer the block from the surface to the other block, place the block on the other block.'\nStart at [0.42, 0.27]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.27], [0.44, 0.34], [0.44, 0.39], [0.46, 0.41], [0.45, 0.49], [0.44, 0.51], [0.45, 0.41], [0.47, 0.36], [0.46, 0.35], [0.46, 0.38]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122061330#0.jpg", + "gt": "[[266, 97], [284, 122], [283, 140], [292, 146], [289, 175], [280, 182], [285, 149], [301, 129], [296, 127], [294, 135]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0110_user_0007_scene_0009_cfg_0004_038522062288#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the pencil out from the pencil sharpener', and now you are performing subtask 'take the pencil out from the pencil sharpener.'\nBegin at [0.36, 0.33] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.33], [0.34, 0.36], [0.35, 0.41], [0.42, 0.51], [0.44, 0.48], [0.43, 0.6], [0.43, 0.44], [0.41, 0.4], [0.37, 0.39], [0.38, 0.41]]}" + } + ], + "images": "RH20T_cfg4_task_0110_user_0007_scene_0009_cfg_0004_038522062288#0", + "gt": "[[233, 119], [218, 130], [225, 146], [267, 182], [279, 174], [274, 215], [275, 157], [262, 145], [239, 139], [241, 148]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "78978_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the lid of the white bin', and now you are performing subtask 'open the lid of the white bin.'\nWith start location [0.6, 0.33], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.33], [0.58, 0.39], [0.51, 0.48], [0.49, 0.53], [0.44, 0.61], [0.47, 0.62], [0.48, 0.73], [0.46, 0.68], [0.44, 0.57], [0.46, 0.51]]}" + } + ], + "images": "78978_exterior_image_1_left#0", + "gt": "[[193, 60], [186, 70], [162, 86], [157, 96], [140, 110], [150, 112], [153, 131], [148, 123], [142, 102], [146, 91]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0038_user_0016_scene_0009_cfg_0001_035622060973#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the cup', and now you are performing subtask 'pick up the cup.'\nYour starting position is [0.39, 0.36]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.36], [0.39, 0.36], [0.38, 0.35], [0.36, 0.36], [0.36, 0.38], [0.36, 0.39], [0.35, 0.41], [0.36, 0.41], [0.35, 0.35], [0.36, 0.32]]}" + } + ], + "images": "RH20T_cfg1_task_0038_user_0016_scene_0009_cfg_0001_035622060973#0", + "gt": "[[247, 129], [250, 129], [246, 127], [228, 131], [228, 137], [230, 139], [226, 146], [228, 149], [226, 127], [228, 115]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "1714_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the stack of blocks to the right and stack two blue blocks on it', and now you are performing subtask 'place the two blue blocks on top of the block pile.'\nWith start location [0.57, 0.42], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.42], [0.55, 0.15], [0.55, 0.25], [0.57, 0.22], [0.58, 0.28], [0.56, 0.39], [0.6, 0.17], [0.58, 0.28], [0.61, 0.34], [0.62, 0.1]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/1714_exterior_image_1_left#534.jpg", + "gt": "[[182.0, 75.5], [175.0, 27.5], [175.0, 45.5], [181.0, 40.5], [185.0, 49.5], [178.0, 70.5], [192.0, 31.5], [186.0, 49.5], [194.0, 61.5], [197.0, 17.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "48910_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the towel twice', and now you are performing subtask 'fold the towel in half again.'\nThe trajectory starts at [0.41, 0.52]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.52], [0.42, 0.39], [0.59, 0.22], [0.58, 0.43], [0.53, 0.57], [0.62, 0.46], [0.48, 0.46], [0.53, 0.38], [0.51, 0.53], [0.54, 0.51]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/48910_exterior_image_1_left#168.jpg", + "gt": "[[132, 93], [133, 70], [189, 40], [185, 78], [170, 103], [197, 82], [155, 82], [170, 69], [163, 96], [174, 92]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "10601_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the mug cup to the left', and now you are performing subtask 'move the mug cup to the left.'\nStart at [0.66, 0.02]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.66, 0.02], [0.6, 0.12], [0.61, 0.11], [0.6, 0.16], [0.57, 0.25], [0.55, 0.32], [0.56, 0.44], [0.54, 0.45], [0.52, 0.41], [0.51, 0.44]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/10601_exterior_image_1_left#0.jpg", + "gt": "[[210.0, 3.5], [192.0, 21.5], [195.0, 20.5], [192.0, 28.5], [181.0, 45.5], [177.0, 58.5], [179.0, 78.5], [174.0, 80.5], [167.0, 73.5], [162.0, 79.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0018_user_0003_scene_0010_cfg_0002_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'play jenga', and now you are performing subtask 'play jenga.'\nThe trajectory starts at [0.44, 0.48]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.48], [0.32, 0.51], [0.28, 0.49], [0.26, 0.51], [0.27, 0.55], [0.28, 0.53], [0.32, 0.64], [0.36, 0.69], [0.32, 0.55], [0.32, 0.55]]}" + } + ], + "images": "RH20T_cfg2_task_0018_user_0003_scene_0010_cfg_0002_037522062165#0", + "gt": "[[284, 174], [207, 182], [180, 176], [169, 183], [174, 197], [182, 191], [204, 231], [231, 247], [207, 197], [207, 197]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_038522062547#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'place the nth Megablock on the green base.'\nWith start location [0.56, 0.33], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.56, 0.32], [0.63, 0.32], [0.63, 0.29], [0.71, 0.45], [0.72, 0.39], [0.62, 0.2], [0.59, 0.22], [0.59, 0.26], [0.54, 0.11], [0.54, 0.11]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_038522062547#632.jpg", + "gt": "[[358, 117], [404, 115], [403, 104], [455, 162], [463, 139], [397, 73], [375, 79], [378, 94], [344, 39], [344, 39]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "4309_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull out a tissue square from the pink box', and now you are performing subtask 'pull out a tissue square from the pink box.'\nBegin at [0.36, 0.82] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.82], [0.43, 0.76], [0.53, 0.82], [0.53, 0.76], [0.57, 0.74], [0.5, 0.67], [0.48, 0.53], [0.56, 0.43], [0.63, 0.42], [0.59, 0.41]]}" + } + ], + "images": "4309_exterior_image_2_left#0", + "gt": "[[115, 147], [138, 137], [170, 148], [169, 137], [181, 134], [161, 120], [153, 96], [178, 78], [202, 75], [188, 74]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "83180_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the pen from the black mug', and now you are performing subtask 'remove the pen from the black mug.'\nStart at [0.84, 0.09]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.84, 0.09], [0.71, 0.0], [0.66, 0.01], [0.63, 0.08], [0.65, 0.08], [0.56, 0.28], [0.6, 0.17], [0.65, 0.28], [0.66, 0.36], [0.64, 0.33]]}" + } + ], + "images": "83180_exterior_image_2_left#0", + "gt": "[[270, 17], [227, 0], [210, 1], [201, 14], [209, 15], [179, 51], [191, 31], [209, 51], [210, 64], [204, 60]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg3_task_0073_user_0010_scene_0010_cfg_0003_036422060909#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the things in the drawer', and now you are performing subtask 'pick up the small square from the table, transfer the small square from the table to the open drawer, place the small square in the drawer.'\nBegin at [0.53, 0.24] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.24], [0.47, 0.0], [0.43, 0.0], [0.63, 0.2], [0.64, 0.32], [0.67, 0.41], [0.6, 0.29], [0.44, 0.25], [0.4, 0.29], [0.4, 0.29]]}" + } + ], + "images": "RH20T_cfg3_task_0073_user_0010_scene_0010_cfg_0003_036422060909#0", + "gt": "[[341, 86], [300, 1], [273, 0], [402, 71], [407, 115], [431, 147], [387, 105], [283, 89], [257, 104], [257, 104]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "24199_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the towel to wipe the pan', and now you are performing subtask 'use the towel to wipe the pan.'\nFrom start point [0.57, 0.06], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.06], [0.45, 0.33], [0.44, 0.49], [0.43, 0.37], [0.42, 0.54], [0.36, 0.63], [0.34, 0.54], [0.43, 0.37], [0.46, 0.47], [0.47, 0.37]]}" + } + ], + "images": "24199_exterior_image_2_left#0", + "gt": "[[181, 11], [143, 60], [141, 89], [138, 67], [135, 97], [116, 114], [109, 98], [137, 67], [148, 84], [151, 67]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "46919_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the door of the washing machine and push the tray back in', and now you are performing subtask 'push the tray back into the washing machine.'\nYour starting position is [0.73, 0.68]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.73, 0.68], [0.73, 0.69], [0.73, 0.71], [0.69, 0.72], [0.69, 0.69], [0.66, 0.65], [0.64, 0.64], [0.65, 0.64], [0.66, 0.63], [0.67, 0.64]]}" + } + ], + "images": "46919_exterior_image_2_left#142", + "gt": "[[234, 122], [235, 124], [233, 128], [221, 129], [220, 125], [211, 117], [204, 115], [207, 116], [212, 114], [215, 115]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "42963_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'cover the pot, turn the faucet clockwise', and now you are performing subtask 'twist the faucet clockwise.'\nFrom start point [0.36, 0.42], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.42], [0.35, 0.41], [0.35, 0.43], [0.4, 0.46], [0.47, 0.32], [0.52, 0.36], [0.55, 0.32], [0.55, 0.36], [0.54, 0.25], [0.42, 0.32]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/42963_exterior_image_1_left#242.jpg", + "gt": "[[116, 75], [111, 73], [111, 78], [129, 82], [151, 57], [167, 65], [177, 58], [177, 64], [172, 45], [133, 57]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0050_user_0020_scene_0008_cfg_0007_104122063633#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the plate on the bowl rack', and now you are performing subtask 'put the plate on the bowl rack.'\nStarting at [0.37, 0.34], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.34], [0.28, 0.52], [0.3, 0.37], [0.27, 0.39], [0.27, 0.37], [0.28, 0.47], [0.28, 0.34], [0.32, 0.32], [0.34, 0.35], [0.33, 0.33]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0050_user_0020_scene_0008_cfg_0007_104122063633#0.jpg", + "gt": "[[234, 121], [182, 186], [193, 132], [173, 142], [173, 133], [181, 168], [180, 123], [202, 116], [217, 126], [211, 118]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0103_user_0010_scene_0005_cfg_0001_819112070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the plate to the goal post', and now you are performing subtask 'move the plate to the goal post.'\nBegin at [0.62, 0.47] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.47], [0.62, 0.53], [0.65, 0.54], [0.67, 0.59], [0.64, 0.56], [0.46, 0.62], [0.45, 0.68], [0.49, 0.56], [0.56, 0.49], [0.56, 0.46]]}" + } + ], + "images": "RH20T_cfg1_task_0103_user_0010_scene_0005_cfg_0001_819112070044#0", + "gt": "[[400, 169], [400, 190], [416, 193], [426, 213], [412, 202], [294, 223], [285, 243], [316, 201], [360, 176], [356, 165]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "1032_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the peach towel into a triangle thrice', and now you are performing subtask 'fold the peach towel into a triangle.'\nYour initial coordinate is [0.34, 0.57]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.34, 0.57], [0.46, 0.67], [0.53, 0.86], [0.57, 0.92], [0.57, 0.96], [0.52, 0.74], [0.58, 0.64], [0.54, 0.61], [0.55, 0.58], [0.49, 0.61]]}" + } + ], + "images": "1032_exterior_image_2_left#0", + "gt": "[[108, 102], [148, 120], [169, 155], [183, 165], [181, 172], [165, 133], [187, 115], [174, 110], [177, 104], [158, 110]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_104122064161#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'place the object on the tabletop, pick up the object in the box, transfer the object from the box to the tabletop, place the object on the tabletop.'\nThe trajectory starts at [0.61, 0.46]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.61, 0.46], [0.61, 0.44], [0.61, 0.45], [0.61, 0.49], [0.61, 0.45], [0.61, 0.41], [0.61, 0.41], [0.61, 0.47], [0.59, 0.58], [0.58, 0.63]]}" + } + ], + "images": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_104122064161#354", + "gt": "[[392, 165], [390, 158], [392, 163], [389, 177], [393, 163], [391, 147], [393, 148], [390, 171], [380, 210], [372, 226]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_039422060546#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'pick up the fifth square from the right side of the table, transfer the fifth square from the right side of the table to the top of the second square, place the fifth square on top of the second square.'\nWith start location [0.48, 0.49], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.49], [0.52, 0.52], [0.55, 0.53], [0.55, 0.57], [0.54, 0.52], [0.46, 0.48], [0.41, 0.49], [0.42, 0.52], [0.39, 0.48], [0.41, 0.45]]}" + } + ], + "images": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_039422060546#710", + "gt": "[[306, 175], [332, 188], [349, 190], [355, 204], [345, 187], [294, 173], [265, 178], [269, 189], [251, 174], [262, 162]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122063633#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'pick up the block on the flat surface, transfer the block from the flat surface to the other blocks, place the block on the other blocks.'\nFrom start point [0.29, 0.44], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.29, 0.44], [0.22, 0.52], [0.2, 0.58], [0.22, 0.52], [0.3, 0.44], [0.33, 0.45], [0.34, 0.5], [0.32, 0.49], [0.31, 0.46], [0.31, 0.45]]}" + } + ], + "images": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122063633#587", + "gt": "[[188, 160], [143, 189], [129, 210], [141, 189], [192, 157], [209, 163], [218, 180], [205, 177], [197, 164], [201, 161]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "78177_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the door', and now you are performing subtask 'open the door.'\nThe trajectory starts at [0.36, 0.56]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.56], [0.57, 0.56], [0.84, 0.65], [0.86, 0.71], [0.83, 0.81], [0.9, 0.88], [0.92, 0.84], [0.88, 0.79], [0.87, 0.56], [0.81, 0.54]]}" + } + ], + "images": "78177_exterior_image_1_left#0", + "gt": "[[115, 101], [184, 100], [268, 117], [275, 127], [266, 146], [289, 159], [294, 151], [283, 143], [278, 101], [260, 98]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "39227_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'spread the towel on the countertop', and now you are performing subtask 'fold the towel.'\nYou begin at [0.69, 0.59]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.59], [0.63, 0.69], [0.59, 0.58], [0.63, 0.71], [0.61, 0.73], [0.65, 0.67], [0.76, 0.73], [0.74, 0.66], [0.73, 0.43], [0.73, 0.43]]}" + } + ], + "images": "39227_exterior_image_1_left#237", + "gt": "[[220, 106], [203, 125], [189, 105], [203, 127], [195, 131], [208, 120], [243, 131], [236, 119], [235, 78], [235, 78]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0125_user_0020_scene_0004_cfg_0006_104122064161#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull up the lever', and now you are performing subtask 'pull the lever on the wooden table up, push the lever on the wooden table up, push the lever on the wooden table up, push the lever on the wooden table up.'\nYou begin at [0.68, 0.23]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.23], [0.73, 0.28], [0.67, 0.22], [0.69, 0.23], [0.67, 0.17], [0.62, 0.15], [0.57, 0.18], [0.6, 0.15], [0.58, 0.19], [0.59, 0.2]]}" + } + ], + "images": "RH20T_cfg6_task_0125_user_0020_scene_0004_cfg_0006_104122064161#573", + "gt": "[[433, 83], [467, 102], [428, 79], [439, 83], [428, 60], [399, 55], [366, 65], [382, 55], [369, 67], [377, 72]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "82985_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the lid on the small pot on the table', and now you are performing subtask 'put the lid on the small pot on the table.'\nBegin at [0.8, 0.0] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.8, 0.0], [0.82, 0.15], [0.77, 0.23], [0.71, 0.47], [0.71, 0.51], [0.73, 0.47], [0.77, 0.52], [0.77, 0.56], [0.74, 0.61], [0.75, 0.54]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/82985_exterior_image_2_left#0.jpg", + "gt": "[[256, 0], [264, 27], [248, 41], [227, 84], [226, 92], [233, 84], [248, 94], [248, 101], [236, 109], [239, 97]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0008_user_0014_scene_0007_cfg_0006_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grab the block and place it at the designated location', and now you are performing subtask 'grab the block and place it at the designated location.'\nStart at [0.53, 0.31]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.31], [0.54, 0.41], [0.56, 0.47], [0.56, 0.56], [0.56, 0.53], [0.53, 0.37], [0.52, 0.32], [0.53, 0.28], [0.53, 0.31], [0.52, 0.29]]}" + } + ], + "images": "RH20T_cfg6_task_0008_user_0014_scene_0007_cfg_0006_104122064161#0", + "gt": "[[338, 110], [346, 149], [356, 169], [357, 203], [357, 192], [337, 132], [336, 115], [341, 102], [340, 111], [335, 106]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0018_user_0010_scene_0009_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'throw a block pulled out from the middle of the Jenga tower onto the table', and now you are performing subtask 'throw a block pulled out from the middle of the Jenga tower onto the table.'\nStart at [0.5, 0.37]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.5, 0.43], [0.57, 0.51], [0.54, 0.57], [0.55, 0.56], [0.55, 0.54], [0.52, 0.48], [0.52, 0.43], [0.54, 0.4], [0.52, 0.4]]}" + } + ], + "images": "RH20T_cfg5_task_0018_user_0010_scene_0009_cfg_0005_104122063678#0", + "gt": "[[323, 132], [320, 153], [366, 183], [345, 205], [352, 202], [353, 193], [336, 174], [334, 156], [348, 145], [336, 144]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0043_user_0010_scene_0006_cfg_0003_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unscrew the lid from the cup', and now you are performing subtask 'twist the lid counterclockwise to unscrew it from the cup.'\nYour initial coordinate is [0.3, 0.12]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.3, 0.12], [0.23, 0.11], [0.29, 0.08], [0.26, 0.08], [0.27, 0.18], [0.32, 0.2], [0.37, 0.18], [0.31, 0.16], [0.36, 0.12], [0.43, 0.16]]}" + } + ], + "images": "RH20T_cfg3_task_0043_user_0010_scene_0006_cfg_0003_104122062823#0", + "gt": "[[193, 42], [149, 40], [184, 30], [167, 29], [171, 64], [202, 72], [238, 64], [196, 59], [229, 45], [275, 56]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "24041_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press down on the top of the silver flask', and now you are performing subtask 'press down on the top of the silver flask.'\nYou start from [0.72, 0.42]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.72, 0.42], [0.67, 0.61], [0.6, 0.72], [0.61, 0.72], [0.6, 0.74], [0.61, 0.79], [0.6, 0.82], [0.62, 0.86], [0.61, 0.81], [0.64, 0.71]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/24041_exterior_image_1_left#0.jpg", + "gt": "[[231.0, 76.0], [214.0, 110.0], [191.0, 130.0], [196.0, 129.0], [192.0, 134.0], [194.0, 142.0], [192.0, 148.0], [198.0, 154.0], [194.0, 146.0], [206.0, 128.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0014_user_0010_scene_0002_cfg_0004_104122063550#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'the cube on the scale is picked up and placed aside', and now you are performing subtask 'the cube on the scale is picked up and placed aside.'\nStart at [0.43, 0.5]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.5], [0.46, 0.54], [0.44, 0.48], [0.43, 0.5], [0.49, 0.47], [0.47, 0.55], [0.46, 0.34], [0.44, 0.39], [0.45, 0.36], [0.5, 0.47]]}" + } + ], + "images": "RH20T_cfg4_task_0014_user_0010_scene_0002_cfg_0004_104122063550#0", + "gt": "[[278, 181], [292, 195], [279, 172], [277, 179], [313, 168], [299, 198], [292, 121], [283, 139], [286, 130], [318, 171]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0058_user_0007_scene_0004_cfg_0005_105422061350#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn off the power strips', and now you are performing subtask 'press the buttons on a row of four power strips to turn off the power strips.'\nWith start location [0.43, 0.13], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.13], [0.38, 0.12], [0.36, 0.24], [0.34, 0.27], [0.34, 0.31], [0.33, 0.33], [0.37, 0.36], [0.39, 0.34], [0.4, 0.38], [0.44, 0.39]]}" + } + ], + "images": "RH20T_cfg5_task_0058_user_0007_scene_0004_cfg_0005_105422061350#0", + "gt": "[[276, 47], [244, 42], [232, 88], [217, 98], [220, 113], [212, 119], [237, 128], [247, 121], [258, 136], [280, 141]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0201_user_0015_scene_0007_cfg_0001_750612070853#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'the medium pipette tip is inserted into the holder for medium pipette tips', and now you are performing subtask 'place the tip of the medium pipette into the holder for medium pipette tips, push the tip of the medium pipette.'\nYour starting position is [0.63, 0.4]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.4], [0.62, 0.36], [0.62, 0.39], [0.63, 0.42], [0.62, 0.41], [0.61, 0.35], [0.6, 0.36], [0.61, 0.35], [0.54, 0.28], [0.54, 0.28]]}" + } + ], + "images": "RH20T_cfg1_task_0201_user_0015_scene_0007_cfg_0001_750612070853#347", + "gt": "[[405, 145], [400, 131], [397, 141], [401, 150], [398, 149], [388, 127], [386, 130], [390, 127], [346, 101], [346, 101]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0002_user_0013_scene_0006_cfg_0002_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull out a napkin, place the napkin on the table', and now you are performing subtask 'pull out a napkin, place the napkin on the table.'\nStarting at [0.53, 0.3], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.3], [0.52, 0.34], [0.57, 0.47], [0.57, 0.5], [0.55, 0.45], [0.48, 0.37], [0.47, 0.36], [0.47, 0.29], [0.45, 0.18], [0.45, 0.18]]}" + } + ], + "images": "RH20T_cfg2_task_0002_user_0013_scene_0006_cfg_0002_104122061850#0", + "gt": "[[337, 109], [336, 123], [367, 169], [368, 181], [349, 161], [309, 134], [304, 128], [303, 106], [286, 64], [286, 64]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0034_user_0015_scene_0008_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'pick up the first block from the table, transfer the first block from behind the table to in front of the table, place the first block in front of the table.'\nStarting at [0.63, 0.37], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.63, 0.45], [0.59, 0.46], [0.57, 0.51], [0.54, 0.52], [0.54, 0.53], [0.61, 0.56], [0.62, 0.55], [0.6, 0.55], [0.6, 0.54]]}" + } + ], + "images": "RH20T_cfg1_task_0034_user_0015_scene_0008_cfg_0001_038522062547#0", + "gt": "[[406, 134], [401, 161], [378, 164], [366, 182], [346, 186], [346, 192], [392, 200], [396, 199], [383, 197], [384, 194]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "40510_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the glue stick from the cup, place it on the box, then on the table, back into the cup, and lastly on the table', and now you are performing subtask 'place glue stick on the box.'\nYour starting position is [0.61, 0.21]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.61, 0.21], [0.63, 0.29], [0.57, 0.44], [0.59, 0.48], [0.57, 0.52], [0.61, 0.43], [0.67, 0.49], [0.68, 0.52], [0.67, 0.55], [0.68, 0.52]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/40510_exterior_image_2_left#0.jpg", + "gt": "[[195, 37], [202, 53], [183, 80], [190, 87], [182, 93], [195, 78], [214, 88], [218, 93], [215, 99], [216, 94]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0066_user_0016_scene_0003_cfg_0004_104122062823#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'building blocks', and now you are performing subtask 'place the green block on the block board, pick up the red block from the table, transfer the red block from the table to the block board, place the red block on the block board.'\nFrom start point [0.66, 0.26], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.66, 0.26], [0.7, 0.19], [0.68, 0.21], [0.68, 0.16], [0.66, 0.1], [0.63, 0.12], [0.59, 0.26], [0.6, 0.15], [0.59, 0.26], [0.56, 0.17]]}" + } + ], + "images": "RH20T_cfg4_task_0066_user_0016_scene_0003_cfg_0004_104122062823#154", + "gt": "[[421, 95], [448, 69], [437, 76], [438, 57], [422, 37], [402, 43], [379, 94], [387, 55], [377, 95], [361, 62]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "5583_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the blue marker into the bowl', and now you are performing subtask 'put the blue marker into the bowl.'\nYou begin at [0.44, 0.36]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.36], [0.44, 0.37], [0.48, 0.33], [0.52, 0.33], [0.57, 0.38], [0.6, 0.46], [0.7, 0.46], [0.72, 0.48], [0.68, 0.57], [0.65, 0.53]]}" + } + ], + "images": "5583_exterior_image_1_left#0", + "gt": "[[140, 65], [140, 67], [154, 60], [167, 60], [182, 68], [191, 82], [225, 82], [229, 86], [219, 103], [209, 96]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "11875_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the fourth switch from the left', and now you are performing subtask 'turn on the fourth switch from the left.'\nStart at [0.2, 0.23]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.2, 0.23], [0.26, 0.29], [0.33, 0.38], [0.38, 0.5], [0.38, 0.56], [0.38, 0.57], [0.38, 0.62], [0.39, 0.62], [0.39, 0.67], [0.38, 0.67]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/11875_exterior_image_2_left#0.jpg", + "gt": "[[64, 42], [84, 52], [105, 69], [120, 90], [123, 101], [120, 103], [122, 111], [125, 112], [126, 121], [123, 121]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_039422060546#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'fold the fourth lid on the box on the table.'\nYou start from [0.3, 0.52]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.3, 0.52], [0.33, 0.58], [0.37, 0.57], [0.37, 0.61], [0.37, 0.56], [0.35, 0.59], [0.38, 0.62], [0.44, 0.78], [0.39, 0.56], [0.3, 0.47]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_039422060546#174.jpg", + "gt": "[[194, 188], [214, 209], [236, 205], [236, 219], [235, 203], [227, 214], [243, 222], [283, 282], [250, 202], [193, 169]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0065_user_0014_scene_0003_cfg_0007_104122064161#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with small Lego blocks for the nth time with n blocks', and now you are performing subtask 'pick up the second Lego block from the right side of the workbench, transfer the second Lego block from the right side of the workbench to the center Lego baseplate, place the second Lego block on the Lego baseplate.'\nWith start location [0.64, 0.51], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.64, 0.51], [0.67, 0.57], [0.67, 0.61], [0.66, 0.52], [0.64, 0.48], [0.6, 0.45], [0.59, 0.48], [0.59, 0.46], [0.58, 0.47], [0.62, 0.47]]}" + } + ], + "images": "RH20T_cfg7_task_0065_user_0014_scene_0003_cfg_0007_104122064161#279", + "gt": "[[411, 182], [430, 204], [428, 218], [421, 187], [409, 173], [382, 162], [376, 173], [375, 164], [373, 169], [400, 169]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "37808_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the pen in the mug', and now you are performing subtask 'put the pen in the mug.'\nBegin at [0.73, 0.19] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.73, 0.19], [0.68, 0.19], [0.64, 0.42], [0.65, 0.41], [0.66, 0.65], [0.6, 0.46], [0.55, 0.48], [0.54, 0.54], [0.55, 0.29], [0.53, 0.29]]}" + } + ], + "images": "37808_exterior_image_1_left#0", + "gt": "[[233.5, 33.5], [217.5, 33.5], [204.5, 75.5], [208.5, 73.5], [212.5, 116.5], [190.5, 82.5], [177.5, 86.5], [173.5, 97.5], [176.5, 52.5], [169.5, 51.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0015_user_0008_scene_0002_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'play the drum', and now you are performing subtask 'play the drum.'\nYour starting position is [0.64, 0.38]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.64, 0.38], [0.61, 0.43], [0.65, 0.37], [0.7, 0.46], [0.68, 0.52], [0.73, 0.35], [0.72, 0.38], [0.68, 0.38], [0.65, 0.43], [0.67, 0.35]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0015_user_0008_scene_0002_cfg_0001_038522062547#0.jpg", + "gt": "[[407, 137], [392, 155], [418, 134], [447, 165], [438, 188], [468, 125], [461, 137], [432, 135], [418, 153], [430, 127]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0101_user_0007_scene_0002_cfg_0002_104422070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'cleaning the table with a cloth', and now you are performing subtask 'cleaning the table with a cloth.'\nYou begin at [0.49, 0.53]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.53], [0.52, 0.55], [0.57, 0.55], [0.57, 0.56], [0.6, 0.56], [0.5, 0.52], [0.52, 0.56], [0.52, 0.57], [0.52, 0.58], [0.52, 0.58]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg2_task_0101_user_0007_scene_0002_cfg_0002_104422070044#0.jpg", + "gt": "[[311, 190], [331, 199], [367, 198], [367, 203], [384, 201], [318, 187], [333, 200], [336, 205], [333, 209], [333, 209]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "230_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the pen and put it in the cup', and now you are performing subtask 'pick up the pen and put it in the cup.'\nYou begin at [0.67, 0.59]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.67, 0.59], [0.7, 0.55], [0.68, 0.51], [0.71, 0.62], [0.76, 0.59], [0.79, 0.64], [0.8, 0.71], [0.75, 0.74], [0.74, 0.81], [0.79, 0.66]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/230_exterior_image_1_left#0.jpg", + "gt": "[[213.5, 105.5], [225.5, 99.5], [218.5, 91.5], [226.5, 112.5], [244.5, 106.5], [252.5, 114.5], [256.5, 128.5], [238.5, 133.5], [237.5, 146.5], [252.5, 119.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0066_user_0016_scene_0001_cfg_0004_104122062823#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large megabloks for the nth time with n blocks', and now you are performing subtask 'pick up a large Megablok from the bottom right corner, transfer the large Megablok from the bottom right corner to the designated building area on the table, place the large Megablok in the building area, press the large Megablok.'\nWith start location [0.78, 0.27], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.78, 0.27], [0.75, 0.2], [0.71, 0.19], [0.68, 0.25], [0.67, 0.35], [0.68, 0.27], [0.66, 0.3], [0.69, 0.25], [0.66, 0.17], [0.63, 0.14]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0066_user_0016_scene_0001_cfg_0004_104122062823#151.jpg", + "gt": "[[501, 96], [479, 73], [452, 67], [436, 89], [429, 125], [434, 98], [424, 107], [442, 89], [422, 60], [402, 52]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "68184_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the white cloth from the black hanger and place it on the table', and now you are performing subtask 'take the white cloth from the black hanger and place it on the table.'\nStart at [0.44, 0.16]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.16], [0.45, 0.09], [0.46, 0.08], [0.46, 0.13], [0.47, 0.14], [0.47, 0.18], [0.43, 0.26], [0.45, 0.27], [0.45, 0.21], [0.48, 0.42]]}" + } + ], + "images": "68184_exterior_image_1_left#0", + "gt": "[[140.0, 28.0], [145.0, 16.0], [148.0, 15.0], [146.0, 23.0], [151.0, 25.0], [149.0, 32.0], [137.0, 47.0], [145.0, 49.0], [145.0, 37.0], [155.0, 76.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "31583_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the marker out of the cup and put it on the table', and now you are performing subtask 'take the marker out of the cup and put it on the table.'\nBegin at [0.72, 0.29] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.73, 0.29], [0.64, 0.32], [0.6, 0.38], [0.52, 0.62], [0.54, 0.54], [0.6, 0.56], [0.57, 0.56], [0.61, 0.62], [0.59, 0.71], [0.61, 0.56]]}" + } + ], + "images": "31583_exterior_image_2_left#0", + "gt": "[[232, 52], [205, 58], [192, 69], [165, 112], [173, 98], [191, 100], [184, 100], [194, 111], [189, 127], [195, 100]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0028_user_0020_scene_0003_cfg_0007_104122061602#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp the handle and close the drawer', and now you are performing subtask 'grasp the handle and close the drawer.'\nYou start from [0.7, 0.32]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.32], [0.69, 0.32], [0.7, 0.41], [0.71, 0.38], [0.69, 0.39], [0.61, 0.32], [0.66, 0.32], [0.65, 0.27], [0.68, 0.24], [0.7, 0.29]]}" + } + ], + "images": "RH20T_cfg7_task_0028_user_0020_scene_0003_cfg_0007_104122061602#0", + "gt": "[[448, 116], [443, 117], [447, 146], [454, 138], [444, 142], [391, 116], [421, 114], [413, 97], [438, 88], [450, 103]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "81550_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the mug from the coffee maker and put it on the counter, and pour the content from the transparent cup into the mug', and now you are performing subtask 'pick up the transparent cup from the countertop.'\nYou begin at [0.73, 0.53]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.73, 0.53], [0.64, 0.58], [0.66, 0.74], [0.61, 0.81], [0.54, 0.82], [0.53, 0.87], [0.59, 0.75], [0.53, 0.63], [0.55, 0.56], [0.53, 0.6]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/81550_exterior_image_1_left#0.jpg", + "gt": "[[235, 96], [206, 104], [210, 133], [194, 145], [173, 148], [170, 156], [188, 135], [171, 114], [177, 101], [170, 108]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0057_user_0014_scene_0008_cfg_0006_104122063633#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn on the power strips', and now you are performing subtask 'press the second button on the power strip, press the third button on the power strip.'\nYour starting position is [0.34, 0.45]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.34, 0.45], [0.33, 0.46], [0.32, 0.46], [0.32, 0.46], [0.32, 0.46], [0.33, 0.47], [0.32, 0.47], [0.32, 0.44], [0.31, 0.44], [0.31, 0.45]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0057_user_0014_scene_0008_cfg_0006_104122063633#178.jpg", + "gt": "[[216, 162], [210, 167], [203, 165], [206, 164], [206, 166], [209, 169], [208, 171], [203, 160], [199, 159], [197, 161]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0089_user_0014_scene_0001_cfg_0004_104122062823#TrajectoryPrediction_Task#12", + "task": "TrajectoryPrediction_Task", + "p_index": 12, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push an object with obstacles', and now you are performing subtask 'push the block to the left, push the block to the left.'\nYou start from [0.55, 0.1]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.1], [0.51, 0.11], [0.49, 0.08], [0.49, 0.07], [0.45, 0.12], [0.41, 0.13], [0.44, 0.08], [0.46, 0.04], [0.45, 0.03], [0.41, 0.07]]}" + } + ], + "images": "RH20T_cfg4_task_0089_user_0014_scene_0001_cfg_0004_104122062823#380", + "gt": "[[350, 35], [326, 38], [313, 29], [315, 25], [285, 44], [260, 47], [284, 28], [292, 16], [285, 12], [260, 24]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0013_user_0007_scene_0001_cfg_0005_105422061350#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the block on the scale', and now you are performing subtask 'place the block on the scale.'\nStarting at [0.43, 0.14], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.14], [0.41, 0.2], [0.32, 0.27], [0.22, 0.43], [0.22, 0.48], [0.25, 0.38], [0.29, 0.37], [0.39, 0.24], [0.41, 0.27], [0.39, 0.15]]}" + } + ], + "images": "RH20T_cfg5_task_0013_user_0007_scene_0001_cfg_0005_105422061350#0", + "gt": "[[273, 51], [260, 73], [207, 98], [141, 154], [144, 174], [163, 136], [187, 132], [250, 88], [262, 96], [252, 54]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_750612070851#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the fourth block on top of the third block, pick up the fifth block from the table, transfer the fifth block from the table to the top of the stack, place the fifth block on the top of the stack.'\nWith start location [0.65, 0.28], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.28], [0.61, 0.44], [0.46, 0.44], [0.43, 0.48], [0.43, 0.53], [0.56, 0.0], [0.7, 0.04], [0.77, 0.18], [0.76, 0.23], [0.79, 0.09]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_750612070851#538.jpg", + "gt": "[[416, 100], [389, 159], [292, 160], [275, 174], [276, 191], [356, 1], [450, 16], [495, 64], [485, 82], [504, 31]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "66631_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the towel to wipe the table', and now you are performing subtask 'fold the towel.'\nBegin at [0.68, 0.4] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.4], [0.62, 0.42], [0.61, 0.46], [0.64, 0.22], [0.63, 0.07], [0.62, 0.23], [0.64, 0.4], [0.64, 0.41], [0.65, 0.43], [0.69, 0.37]]}" + } + ], + "images": "66631_exterior_image_1_left#155", + "gt": "[[216, 72], [197, 76], [194, 83], [204, 40], [202, 13], [200, 42], [206, 72], [204, 74], [207, 77], [220, 66]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "52069_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the yellow plate to the front right side of the table', and now you are performing subtask 'transfer the yellow plate from the center of the table to the front right side of the table.'\nStarting at [0.67, 0.01], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.67, 0.01], [0.68, 0.15], [0.68, 0.49], [0.69, 0.5], [0.69, 0.48], [0.67, 0.5], [0.69, 0.53], [0.68, 0.56], [0.69, 0.55], [0.7, 0.56]]}" + } + ], + "images": "52069_exterior_image_1_left#0", + "gt": "[[214, 2], [219, 27], [218, 88], [221, 90], [221, 86], [213, 90], [221, 96], [217, 100], [222, 99], [223, 101]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "74617_exterior_image_1_left#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the fries plush toy and put it on the plate', and now you are performing subtask 'place the fries toy in the plate.'\nFrom start point [0.51, 0.54], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.54], [0.45, 0.62], [0.52, 0.53], [0.47, 0.67], [0.52, 0.77], [0.52, 0.57], [0.59, 0.52], [0.5, 0.64], [0.49, 0.76], [0.47, 0.62]]}" + } + ], + "images": "74617_exterior_image_1_left#352", + "gt": "[[162, 97], [143, 111], [166, 95], [152, 120], [166, 139], [166, 103], [189, 93], [160, 115], [157, 137], [152, 111]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "63511_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unravel the charger', and now you are performing subtask 'place the charger at the front of the tabletop.'\nYour starting position is [0.57, 0.4]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.4], [0.49, 0.4], [0.46, 0.48], [0.48, 0.49], [0.49, 0.4], [0.57, 0.41], [0.53, 0.24], [0.47, 0.49], [0.5, 0.55], [0.52, 0.48]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/63511_exterior_image_1_left#176.jpg", + "gt": "[[181, 72], [157, 72], [148, 86], [155, 89], [158, 72], [182, 73], [171, 43], [152, 88], [159, 99], [166, 86]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0008_user_0010_scene_0010_cfg_0005_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grab the block and place it at the black dot', and now you are performing subtask 'grab the block and place it at the black dot.'\nYour starting position is [0.45, 0.42]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.42], [0.39, 0.47], [0.33, 0.41], [0.22, 0.37], [0.22, 0.48], [0.3, 0.49], [0.41, 0.64], [0.41, 0.67], [0.38, 0.58], [0.36, 0.41]]}" + } + ], + "images": "RH20T_cfg5_task_0008_user_0010_scene_0010_cfg_0005_037522062165#0", + "gt": "[[289, 152], [252, 170], [212, 146], [139, 132], [143, 172], [189, 177], [260, 229], [261, 241], [244, 210], [233, 148]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "1544_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'hang the cloth on the screen', and now you are performing subtask 'hang the cloth on the screen.'\nStarting at [0.18, 0.0], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.18, 0.0], [0.25, 0.16], [0.2, 0.03], [0.27, 0.77], [0.32, 0.36], [0.33, 0.42], [0.38, 0.36], [0.42, 0.13], [0.46, 0.11], [0.51, 0.23]]}" + } + ], + "images": "1544_exterior_image_2_left#0", + "gt": "[[59, 0], [80, 29], [63, 5], [85, 138], [104, 65], [106, 76], [120, 64], [135, 23], [147, 19], [162, 42]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "9510_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pour the water out of the mug on the right side and move the mug to the left side of the sink', and now you are performing subtask 'pour the water out of the mug on the right side and move the mug to the left side of the sink.'\nStart at [0.64, 0.32]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.64, 0.32], [0.54, 0.47], [0.5, 0.32], [0.48, 0.43], [0.3, 0.57], [0.24, 0.57], [0.14, 0.73], [0.1, 0.93], [0.14, 0.71], [0.24, 0.51]]}" + } + ], + "images": "9510_exterior_image_1_left#0", + "gt": "[[206, 57], [173, 84], [160, 57], [155, 77], [95, 103], [76, 102], [44, 131], [31, 167], [45, 128], [78, 92]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0073_user_0007_scene_0001_cfg_0005_104122063678#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the things in the drawer', and now you are performing subtask 'grab the handle to pull the drawer open, pick up the cube on the table, transfer the cube from the table to the drawer, place the cube in the drawer.'\nFrom start point [0.51, 0.47], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.47], [0.52, 0.44], [0.55, 0.46], [0.56, 0.47], [0.56, 0.53], [0.57, 0.45], [0.55, 0.42], [0.54, 0.51], [0.51, 0.53], [0.51, 0.59]]}" + } + ], + "images": "RH20T_cfg5_task_0073_user_0007_scene_0001_cfg_0005_104122063678#539", + "gt": "[[328, 169], [333, 160], [352, 164], [359, 171], [356, 192], [362, 163], [351, 152], [345, 183], [329, 192], [328, 213]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "21213_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the two paper bags out of the wooden crate', and now you are performing subtask 'pick up the second bag on the table.'\nBegin at [0.4, 0.59] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.59], [0.42, 0.62], [0.41, 0.66], [0.43, 0.61], [0.44, 0.53], [0.45, 0.74], [0.43, 0.67], [0.44, 0.56], [0.48, 0.57], [0.49, 0.61]]}" + } + ], + "images": "21213_exterior_image_2_left#0", + "gt": "[[127, 107], [133, 111], [132, 118], [139, 109], [140, 95], [143, 133], [139, 121], [141, 100], [155, 103], [158, 110]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0037_user_0014_scene_0002_cfg_0007_104122063633#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the fourth block from the tabletop, transfer the fourth block from the tabletop to the third block, place the fourth block on the third block.'\nWith start location [0.28, 0.45], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.28, 0.45], [0.25, 0.49], [0.25, 0.55], [0.25, 0.56], [0.29, 0.37], [0.32, 0.35], [0.34, 0.38], [0.32, 0.34], [0.28, 0.39], [0.27, 0.43]]}" + } + ], + "images": "RH20T_cfg7_task_0037_user_0014_scene_0002_cfg_0007_104122063633#267", + "gt": "[[177, 162], [160, 178], [157, 197], [162, 200], [184, 134], [203, 126], [220, 136], [204, 123], [180, 140], [171, 154]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "68570_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull open the bottom drawer of the beige filing cabinet', and now you are performing subtask 'pull open the bottom drawer of the beige filing cabinet.'\nStart at [0.59, 0.99]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.99], [0.54, 0.89], [0.54, 0.81], [0.61, 0.74], [0.63, 0.77], [0.68, 0.69], [0.66, 0.77], [0.7, 0.86], [0.71, 0.74], [0.71, 0.74]]}" + } + ], + "images": "68570_exterior_image_1_left#0", + "gt": "[[188, 178], [173, 160], [174, 145], [195, 133], [203, 139], [218, 125], [211, 139], [223, 154], [227, 133], [227, 133]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "22578_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the lid from the glass jar and place it on the table, then put it back on the jar', and now you are performing subtask 'place the lid from the table on the right side of the glass jar onto the glass jar.'\nYou begin at [0.11, 0.25]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.11, 0.25], [0.16, 0.29], [0.35, 0.32], [0.43, 0.43], [0.38, 0.51], [0.44, 0.49], [0.49, 0.57], [0.49, 0.74], [0.5, 0.54], [0.5, 0.54]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/22578_exterior_image_2_left#0.jpg", + "gt": "[[34, 45], [52, 53], [112, 57], [136, 78], [122, 92], [140, 88], [156, 103], [158, 133], [160, 98], [160, 98]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "10286_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the keys from the nightstand and put them on the windowsill then place the keys back on the nightstand', and now you are performing subtask 'place the key back on the nightstand.'\nBegin at [0.51, 0.57] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.57], [0.53, 0.44], [0.51, 0.61], [0.46, 0.48], [0.52, 0.61], [0.55, 0.58], [0.5, 0.58], [0.44, 0.41], [0.22, 0.41], [0.27, 0.47]]}" + } + ], + "images": "10286_exterior_image_2_left#186", + "gt": "[[163, 102], [170, 79], [163, 109], [148, 86], [166, 109], [175, 104], [159, 104], [141, 73], [71, 74], [85, 85]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg3_task_0058_user_0010_scene_0006_cfg_0003_104122062823#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn off the power strips', and now you are performing subtask 'press the first button of the power strip, press the second button of the power strip.'\nYour initial coordinate is [0.23, 0.16]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.23, 0.16], [0.23, 0.16], [0.23, 0.16], [0.23, 0.17], [0.23, 0.17], [0.23, 0.18], [0.22, 0.15], [0.22, 0.14], [0.21, 0.13], [0.21, 0.13]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg3_task_0058_user_0010_scene_0006_cfg_0003_104122062823#168.jpg", + "gt": "[[149, 58], [146, 56], [148, 59], [148, 63], [147, 63], [147, 64], [144, 55], [140, 49], [136, 46], [136, 46]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0046_user_0020_scene_0004_cfg_0006_104122061018#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the cup on the cup rack', and now you are performing subtask 'put the cup on the cup rack.'\nFrom start point [0.57, 0.08], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.08], [0.52, 0.19], [0.47, 0.17], [0.49, 0.25], [0.5, 0.2], [0.47, 0.16], [0.49, 0.22], [0.45, 0.22], [0.48, 0.27], [0.45, 0.21]]}" + } + ], + "images": "RH20T_cfg6_task_0046_user_0020_scene_0004_cfg_0006_104122061018#0", + "gt": "[[367, 28], [330, 69], [300, 60], [312, 89], [317, 71], [300, 59], [313, 78], [287, 79], [310, 98], [286, 75]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0081_user_0020_scene_0004_cfg_0006_104122060811#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the leg of the glasses (with one hand)', and now you are performing subtask 'fold the leg of the glasses out.'\nYour initial coordinate is [0.56, 0.38]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.56, 0.38], [0.55, 0.4], [0.55, 0.43], [0.54, 0.46], [0.54, 0.47], [0.54, 0.47], [0.55, 0.46], [0.55, 0.45], [0.55, 0.47], [0.56, 0.49]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0081_user_0020_scene_0004_cfg_0006_104122060811#420.jpg", + "gt": "[[357, 138], [353, 144], [352, 155], [344, 165], [344, 169], [348, 169], [353, 165], [354, 161], [351, 168], [357, 175]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0073_user_0020_scene_0003_cfg_0006_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the things in the drawer', and now you are performing subtask 'pull the drawer to the open position.'\nThe trajectory starts at [0.53, 0.31]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.31], [0.54, 0.29], [0.53, 0.28], [0.55, 0.24], [0.54, 0.22], [0.56, 0.27], [0.55, 0.32], [0.57, 0.33], [0.58, 0.39], [0.55, 0.36]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0073_user_0020_scene_0003_cfg_0006_104122064161#0.jpg", + "gt": "[[338, 111], [345, 105], [341, 101], [350, 87], [348, 79], [356, 98], [355, 114], [364, 119], [372, 142], [355, 130]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0065_user_0007_scene_0010_cfg_0002_105422061350#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with small Lego blocks for the nth time with n blocks', and now you are performing subtask 'pick up the green Lego block from the table, transfer the green Lego block from the table to the building area, place the green Lego block in the building area, press the green Lego block in the building area, pick up the black Lego block from the table.'\nYou start from [0.47, 0.3]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.3], [0.48, 0.38], [0.41, 0.27], [0.27, 0.22], [0.26, 0.26], [0.35, 0.3], [0.35, 0.35], [0.34, 0.31], [0.3, 0.31], [0.3, 0.31]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg2_task_0065_user_0007_scene_0010_cfg_0002_105422061350#0.jpg", + "gt": "[[304, 109], [305, 135], [262, 97], [171, 79], [169, 93], [227, 108], [224, 126], [216, 111], [189, 112], [189, 112]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0073_user_0001_scene_0003_cfg_0001_819112070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the things in the drawer', and now you are performing subtask 'pick up the block on the drawer, transfer the block from above the drawer to inside the drawer, place the block in the drawer.'\nWith start location [0.62, 0.46], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.46], [0.63, 0.39], [0.73, 0.36], [0.75, 0.33], [0.75, 0.38], [0.74, 0.36], [0.68, 0.4], [0.67, 0.5], [0.67, 0.45], [0.66, 0.45]]}" + } + ], + "images": "RH20T_cfg1_task_0073_user_0001_scene_0003_cfg_0001_819112070044#0", + "gt": "[[400, 167], [406, 142], [469, 129], [483, 119], [480, 136], [476, 131], [435, 145], [428, 179], [431, 161], [422, 163]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "1714_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the stack of blocks to the right and stack two blue blocks on it', and now you are performing subtask 'pick up two blue blocks from the left side of the pile.'\nWith start location [0.46, 0.35], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.35], [0.54, 0.31], [0.59, 0.36], [0.58, 0.26], [0.6, 0.14], [0.59, 0.19], [0.6, 0.19], [0.58, 0.28], [0.57, 0.42], [0.57, 0.42]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/1714_exterior_image_1_left#383.jpg", + "gt": "[[146.0, 62.5], [173.0, 55.5], [188.0, 64.5], [187.0, 47.5], [193.0, 25.5], [188.0, 34.5], [191.0, 34.5], [185.0, 50.5], [182.0, 75.5], [182.0, 75.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0127_user_0020_scene_0006_cfg_0006_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'plug in the power cord of the desk lamp, turn on the socket, and light up the desk lamp', and now you are performing subtask 'pick up the power cable on the table, transfer the power cable from the table to the socket on the desktop.'\nYour starting position is [0.53, 0.31]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.31], [0.54, 0.28], [0.55, 0.28], [0.67, 0.29], [0.66, 0.42], [0.67, 0.37], [0.64, 0.35], [0.61, 0.37], [0.6, 0.41], [0.6, 0.41]]}" + } + ], + "images": "RH20T_cfg6_task_0127_user_0020_scene_0006_cfg_0006_104122064161#0", + "gt": "[[338, 111], [343, 101], [354, 99], [429, 106], [422, 151], [427, 132], [412, 127], [392, 134], [385, 147], [385, 147]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0037_user_0014_scene_0001_cfg_0007_104122061330#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line', and now you are performing subtask 'pick up the third block from the table, transfer the third block from the table to the stacking position, place the third block on top of the second block, pick up the fourth block from the table.'\nWith start location [0.49, 0.61], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.61], [0.5, 0.64], [0.46, 0.57], [0.45, 0.54], [0.41, 0.54], [0.4, 0.55], [0.41, 0.53], [0.42, 0.53], [0.45, 0.57], [0.48, 0.59]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0037_user_0014_scene_0001_cfg_0007_104122061330#282.jpg", + "gt": "[[314, 221], [320, 229], [294, 205], [288, 195], [263, 195], [255, 199], [261, 192], [270, 192], [289, 205], [307, 213]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_104122061018#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'place the object on the tabletop, pick up the cube in the box, transfer the cube from the box to the tabletop, place the cube on the tabletop.'\nStarting at [0.42, 0.17], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.17], [0.42, 0.22], [0.43, 0.18], [0.49, 0.3], [0.49, 0.38], [0.5, 0.25], [0.47, 0.16], [0.47, 0.24], [0.48, 0.13], [0.53, 0.11]]}" + } + ], + "images": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_104122061018#510", + "gt": "[[270, 61], [269, 78], [272, 64], [313, 107], [316, 137], [319, 91], [303, 57], [303, 86], [309, 48], [339, 39]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "74887_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'throw the building block tower onto the table, then push the green building blocks in the tower to the edge of the table', and now you are performing subtask 'push the green block from the block tower to the edge of the table.'\nYour initial coordinate is [0.61, 0.56]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.61, 0.56], [0.61, 0.59], [0.63, 0.66], [0.61, 0.6], [0.58, 0.56], [0.57, 0.53], [0.57, 0.51], [0.57, 0.51], [0.57, 0.5], [0.59, 0.51]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/74887_exterior_image_1_left#203.jpg", + "gt": "[[194, 101], [196, 107], [203, 118], [196, 108], [186, 101], [182, 96], [182, 91], [181, 91], [182, 90], [190, 91]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0037_user_0014_scene_0010_cfg_0001_819112070044#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the block from the table onto the three blocks standing upright, pick up the block from the right side of the table, transfer the block from the table to the four blocks standing upright, place the block from the table onto the four blocks standing upright.'\nYour initial coordinate is [0.5, 0.49]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.49], [0.47, 0.47], [0.46, 0.5], [0.48, 0.57], [0.47, 0.51], [0.5, 0.46], [0.5, 0.45], [0.5, 0.47], [0.5, 0.43], [0.5, 0.43]]}" + } + ], + "images": "RH20T_cfg1_task_0037_user_0014_scene_0010_cfg_0001_819112070044#321", + "gt": "[[323, 177], [302, 171], [294, 179], [308, 205], [303, 184], [323, 165], [321, 161], [317, 169], [321, 153], [321, 153]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "86920_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'cover the pot and fold the cloth', and now you are performing subtask 'place the lid on the pot.'\nBegin at [1.0, 0.04] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[1.0, 0.04], [0.94, 0.11], [0.9, 0.24], [0.9, 0.52], [0.87, 0.57], [0.93, 0.42], [0.94, 0.41], [0.94, 0.47], [0.93, 0.48], [0.95, 0.44]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/86920_exterior_image_2_left#0.jpg", + "gt": "[[319, 8], [302, 20], [289, 43], [289, 93], [279, 102], [297, 76], [302, 73], [301, 84], [297, 87], [304, 80]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0094_user_0007_scene_0006_cfg_0005_104122063678#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'press the part of the box.'\nThe trajectory starts at [0.59, 0.52]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.52], [0.59, 0.54], [0.6, 0.54], [0.6, 0.56], [0.61, 0.63], [0.6, 0.65], [0.6, 0.65], [0.6, 0.62], [0.59, 0.57], [0.59, 0.55]]}" + } + ], + "images": "RH20T_cfg5_task_0094_user_0007_scene_0006_cfg_0005_104122063678#232", + "gt": "[[379, 188], [379, 193], [382, 196], [384, 202], [388, 227], [382, 234], [381, 233], [382, 225], [378, 204], [375, 198]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0037_user_0020_scene_0006_cfg_0007_104122061018#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the third block from the table, transfer the third block from the table to above the first block, pick up the fourth block from the table.'\nStart at [0.42, 0.25]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.25], [0.41, 0.34], [0.42, 0.25], [0.43, 0.24], [0.45, 0.32], [0.47, 0.26], [0.53, 0.35], [0.54, 0.46], [0.53, 0.51], [0.52, 0.52]]}" + } + ], + "images": "RH20T_cfg7_task_0037_user_0020_scene_0006_cfg_0007_104122061018#201", + "gt": "[[267, 91], [263, 123], [267, 91], [274, 85], [285, 115], [303, 95], [338, 125], [346, 166], [340, 183], [331, 187]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0016_user_0011_scene_0010_cfg_0003_104122062823#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'hit the pool ball', and now you are performing subtask 'twist the cue clockwise to hit the pool ball, push the ball with the cue, place the cue on the pool table.'\nYou begin at [0.52, 0.36]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.36], [0.45, 0.26], [0.48, 0.32], [0.48, 0.36], [0.5, 0.42], [0.5, 0.37], [0.49, 0.32], [0.41, 0.18], [0.37, 0.2], [0.4, 0.24]]}" + } + ], + "images": "RH20T_cfg3_task_0016_user_0011_scene_0010_cfg_0003_104122062823#360", + "gt": "[[335, 131], [286, 95], [310, 115], [306, 130], [321, 152], [323, 133], [311, 114], [260, 64], [239, 71], [257, 88]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "4134_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'insert the small object into the green building block', and now you are performing subtask 'pick up the blue block on the table.'\nThe trajectory starts at [0.76, 0.29]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.76, 0.29], [0.6, 0.24], [0.51, 0.31], [0.51, 0.47], [0.54, 0.51], [0.58, 0.51], [0.63, 0.54], [0.65, 0.51], [0.52, 0.33], [0.49, 0.32]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/4134_exterior_image_1_left#0.jpg", + "gt": "[[242.5, 52.0], [192.5, 44.0], [163.5, 55.0], [163.5, 85.0], [171.5, 91.0], [186.5, 91.0], [202.5, 98.0], [207.5, 92.0], [167.5, 60.0], [157.5, 58.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "48008_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the sauce bottle out of the blue cup and put it on the counter', and now you are performing subtask 'take the sauce bottle out of the blue cup and put it on the counter.'\nStarting at [0.1, 0.1], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.1, 0.1], [0.35, 0.23], [0.51, 0.58], [0.46, 0.56], [0.51, 0.53], [0.57, 0.57], [0.55, 0.67], [0.53, 0.58], [0.59, 0.58], [0.54, 0.48]]}" + } + ], + "images": "48008_exterior_image_1_left#0", + "gt": "[[31.5, 18.0], [110.5, 42.0], [164.5, 104.0], [148.5, 101.0], [162.5, 96.0], [181.5, 103.0], [175.5, 120.0], [169.5, 104.0], [188.5, 105.0], [173.5, 87.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_035622060973#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the third block on top of the second block, pick up the fourth block from the surface.'\nBegin at [0.33, 0.41] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.33, 0.41], [0.34, 0.38], [0.33, 0.41], [0.32, 0.47], [0.32, 0.44], [0.33, 0.4], [0.3, 0.42], [0.3, 0.43], [0.3, 0.49], [0.29, 0.51]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_035622060973#484.jpg", + "gt": "[[214, 147], [216, 138], [209, 149], [207, 170], [207, 160], [213, 145], [194, 150], [190, 155], [190, 177], [186, 184]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with small Lego blocks for the nth time with n blocks', and now you are performing subtask 'pick up the black small Lego block from the table, transfer the black small Lego block from the table to the gray base plate, place the black small Lego block on the gray base plate, press the black small Lego block.'\nBegin at [0.63, 0.36] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.36], [0.64, 0.32], [0.68, 0.43], [0.73, 0.43], [0.7, 0.46], [0.71, 0.55], [0.73, 0.46], [0.72, 0.6], [0.71, 0.54], [0.64, 0.51]]}" + } + ], + "images": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_038522062547#0", + "gt": "[[405, 129], [411, 115], [438, 154], [468, 156], [449, 166], [454, 199], [466, 167], [460, 217], [452, 194], [408, 183]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0029_user_0001_scene_0004_cfg_0001_750612070853#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp the handle and open the drawer', and now you are performing subtask 'grasp the handle and open the drawer.'\nThe trajectory starts at [0.46, 0.32]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.32], [0.37, 0.23], [0.33, 0.26], [0.3, 0.43], [0.3, 0.72], [0.36, 1.0], [0.54, 0.99], [0.34, 1.0], [0.4, 1.0], [0.4, 0.69]]}" + } + ], + "images": "RH20T_cfg1_task_0029_user_0001_scene_0004_cfg_0001_750612070853#0", + "gt": "[[292, 115], [234, 83], [210, 94], [190, 154], [195, 259], [232, 360], [344, 356], [216, 360], [254, 360], [255, 248]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "31362_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the towel twice', and now you are performing subtask 'fold the towel.'\nYour starting position is [0.35, 0.01]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.01], [0.63, 0.44], [0.65, 0.63], [0.6, 0.66], [0.68, 0.59], [0.62, 0.49], [0.54, 0.45], [0.37, 0.53], [0.37, 0.59], [0.44, 0.58]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/31362_exterior_image_1_left#0.jpg", + "gt": "[[113, 1], [203, 79], [208, 114], [193, 118], [219, 106], [200, 88], [173, 81], [119, 96], [119, 106], [141, 104]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "10020_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the black lid from the top of the can and put it on the table', and now you are performing subtask 'take the black lid from the top of the can and put it on the table.'\nThe trajectory starts at [0.8, 0.0]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.8, 0.0], [0.61, 0.14], [0.57, 0.28], [0.54, 0.29], [0.52, 0.22], [0.49, 0.33], [0.5, 0.48], [0.49, 0.52], [0.47, 0.52], [0.51, 0.37]]}" + } + ], + "images": "10020_exterior_image_1_left#0", + "gt": "[[257, 0], [196, 26], [183, 50], [172, 53], [165, 40], [158, 59], [159, 87], [156, 93], [152, 93], [163, 67]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0037_user_0007_scene_0004_cfg_0005_105422061350#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the third block from the surface, place the third block on top of the second block.'\nFrom start point [0.38, 0.21], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.21], [0.31, 0.16], [0.25, 0.24], [0.21, 0.32], [0.22, 0.36], [0.24, 0.16], [0.29, 0.13], [0.32, 0.06], [0.35, 0.06], [0.38, 0.13]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0037_user_0007_scene_0004_cfg_0005_105422061350#637.jpg", + "gt": "[[243, 75], [196, 56], [157, 86], [136, 117], [139, 129], [156, 58], [183, 47], [202, 20], [225, 22], [246, 47]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0061_user_0001_scene_0006_cfg_0001_035622060973#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'place the large block on the table, pick up the small block from the gift box, transfer the small block from the gift box to the table, place the small block on the table.'\nFrom start point [0.23, 0.39], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.23, 0.39], [0.3, 0.36], [0.36, 0.45], [0.38, 0.4], [0.41, 0.41], [0.37, 0.45], [0.38, 0.47], [0.33, 0.32], [0.3, 0.33], [0.31, 0.39]]}" + } + ], + "images": "RH20T_cfg1_task_0061_user_0001_scene_0006_cfg_0001_035622060973#188", + "gt": "[[148, 141], [190, 131], [232, 161], [240, 144], [261, 148], [234, 161], [242, 170], [212, 114], [192, 120], [201, 140]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'pick up the first Megablock from the left side of the table, place the first Megablock on the green base.'\nYou begin at [0.63, 0.37]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.68, 0.61], [0.74, 0.76], [0.72, 0.75], [0.73, 0.72], [0.7, 0.65], [0.66, 0.61], [0.68, 0.61], [0.62, 0.6], [0.69, 0.62]]}" + } + ], + "images": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_038522062547#0", + "gt": "[[406, 132], [432, 220], [473, 272], [460, 270], [468, 260], [451, 234], [423, 219], [436, 220], [400, 216], [442, 223]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0003_user_0010_scene_0005_cfg_0005_104422070044#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the green button on the right, press the red button in the middle.'\nYour starting position is [0.63, 0.45]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.45], [0.62, 0.45], [0.62, 0.46], [0.62, 0.47], [0.63, 0.46], [0.62, 0.46], [0.62, 0.44], [0.62, 0.42], [0.6, 0.4], [0.59, 0.39]]}" + } + ], + "images": "RH20T_cfg5_task_0003_user_0010_scene_0005_cfg_0005_104422070044#132", + "gt": "[[403, 162], [400, 163], [400, 166], [399, 168], [401, 167], [400, 167], [398, 158], [395, 150], [387, 144], [375, 142]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0057_user_0014_scene_0008_cfg_0004_038522062288#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn on the power strips', and now you are performing subtask 'press the button on the first power strip, press the button on the second power strip.'\nStart at [0.41, 0.42]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.42], [0.41, 0.42], [0.41, 0.43], [0.42, 0.44], [0.42, 0.45], [0.42, 0.46], [0.43, 0.46], [0.44, 0.44], [0.44, 0.43], [0.44, 0.43]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0057_user_0014_scene_0008_cfg_0004_038522062288#135.jpg", + "gt": "[[265, 151], [263, 150], [265, 153], [266, 158], [268, 163], [271, 165], [273, 165], [279, 158], [280, 155], [281, 154]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "81842_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place one block on top of the stacked blue and green blocks', and now you are performing subtask 'pick up the blue block on the table.'\nFrom start point [0.68, 0.19], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.19], [0.61, 0.18], [0.56, 0.25], [0.51, 0.48], [0.52, 0.52], [0.56, 0.54], [0.55, 0.39], [0.46, 0.18], [0.43, 0.2], [0.46, 0.21]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/81842_exterior_image_1_left#0.jpg", + "gt": "[[217.5, 34.0], [196.5, 33.0], [180.5, 45.0], [164.5, 86.0], [167.5, 93.0], [180.5, 98.0], [176.5, 71.0], [146.5, 32.0], [137.5, 36.0], [147.5, 37.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "25326_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the lid on the clear container on the left then turn on the extension cable', and now you are performing subtask 'press the button on the extension cord to turn on the power.'\nYou start from [0.33, 0.29]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.33, 0.29], [0.53, 0.33], [0.58, 0.52], [0.63, 0.56], [0.63, 0.66], [0.58, 0.62], [0.62, 0.6], [0.62, 0.67], [0.54, 0.48], [0.54, 0.12]]}" + } + ], + "images": "25326_exterior_image_2_left#150", + "gt": "[[107, 52], [169, 59], [186, 93], [203, 101], [201, 119], [187, 111], [197, 108], [200, 120], [172, 87], [173, 21]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "47445_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the marker from the box and place it on the table', and now you are performing subtask 'remove the marker from the box and place it on the table.'\nYour initial coordinate is [0.8, 0.36]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.8, 0.36], [0.73, 0.32], [0.66, 0.33], [0.57, 0.43], [0.55, 0.49], [0.57, 0.45], [0.54, 0.38], [0.53, 0.42], [0.51, 0.6], [0.51, 0.6]]}" + } + ], + "images": "47445_exterior_image_1_left#0", + "gt": "[[257, 65], [235, 57], [212, 59], [181, 78], [177, 89], [183, 81], [174, 69], [169, 76], [164, 108], [164, 108]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0083_user_0007_scene_0010_cfg_0005_104422070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'swat with a flyswatter', and now you are performing subtask 'swat with a flyswatter.'\nYour starting position is [0.58, 0.37]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.37], [0.6, 0.47], [0.6, 0.55], [0.58, 0.45], [0.6, 0.67], [0.64, 0.51], [0.68, 0.47], [0.71, 0.48], [0.66, 0.48], [0.63, 0.51]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0083_user_0007_scene_0010_cfg_0005_104422070044#0.jpg", + "gt": "[[374, 132], [385, 170], [382, 199], [370, 161], [381, 240], [412, 182], [434, 169], [454, 172], [422, 172], [403, 182]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0071_user_0020_scene_0007_cfg_0006_104122061602#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close a sliding window', and now you are performing subtask 'push the handle of the sliding window to close the lower window.'\nWith start location [0.78, 0.62], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.78, 0.62], [0.76, 0.64], [0.75, 0.62], [0.73, 0.62], [0.76, 0.6], [0.78, 0.62], [0.75, 0.63], [0.76, 0.57], [0.75, 0.51], [0.7, 0.44]]}" + } + ], + "images": "RH20T_cfg6_task_0071_user_0020_scene_0007_cfg_0006_104122061602#320", + "gt": "[[502, 225], [487, 232], [480, 224], [465, 223], [486, 215], [497, 225], [479, 227], [487, 205], [479, 182], [451, 157]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_039422060546#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the third button on the desk.'\nYou start from [0.46, 0.31]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.31], [0.53, 0.14], [0.54, 0.15], [0.53, 0.12], [0.55, 0.17], [0.53, 0.08], [0.51, 0.05], [0.52, 0.15], [0.46, 0.18], [0.43, 0.3]]}" + } + ], + "images": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_039422060546#280", + "gt": "[[295, 113], [337, 50], [347, 54], [340, 44], [350, 61], [341, 28], [325, 19], [332, 55], [296, 66], [272, 107]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0061_user_0007_scene_0008_cfg_0005_105422061350#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the first block out of the box and place it on the table, then take the second block and place it on top of the first block', and now you are performing subtask 'place the first block on the table, pick up the second block in the box, transfer the second block from the box to above the first block, place the second block on the first block.'\nWith start location [0.34, 0.47], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.34, 0.47], [0.34, 0.42], [0.36, 0.3], [0.39, 0.22], [0.4, 0.29], [0.4, 0.27], [0.38, 0.27], [0.37, 0.35], [0.34, 0.43], [0.4, 0.16]]}" + } + ], + "images": "RH20T_cfg5_task_0061_user_0007_scene_0008_cfg_0005_105422061350#345", + "gt": "[[220, 171], [216, 152], [228, 107], [247, 81], [255, 104], [256, 96], [246, 96], [238, 125], [220, 155], [253, 57]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0083_user_0007_scene_0010_cfg_0005_105422061350#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'swat with a flyswatter', and now you are performing subtask 'swat with a flyswatter.'\nYou begin at [0.43, 0.13]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.13], [0.49, 0.29], [0.5, 0.18], [0.53, 0.37], [0.4, 0.1], [0.38, 0.11], [0.36, 0.16], [0.43, 0.24], [0.46, 0.23], [0.48, 0.17]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0083_user_0007_scene_0010_cfg_0005_105422061350#0.jpg", + "gt": "[[276, 47], [315, 104], [319, 65], [339, 134], [259, 36], [242, 38], [231, 57], [277, 86], [296, 82], [306, 61]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "30469_exterior_image_1_left#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer, then take out a plush toy from the drawer, then close the drawer', and now you are performing subtask 'push the drawer back to the closed position.'\nYour initial coordinate is [0.48, 0.34]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.34], [0.51, 0.32], [0.62, 0.46], [0.66, 0.47], [0.68, 0.5], [0.67, 0.56], [0.56, 0.57], [0.61, 0.52], [0.62, 0.48], [0.62, 0.48]]}" + } + ], + "images": "30469_exterior_image_1_left#753", + "gt": "[[155, 62], [164, 57], [198, 82], [211, 85], [218, 90], [214, 101], [180, 102], [195, 93], [198, 86], [198, 86]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0029_user_0014_scene_0003_cfg_0006_104122061330#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp the handle and open the drawer', and now you are performing subtask 'grasp the handle and open the drawer.'\nBegin at [0.42, 0.27] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.27], [0.44, 0.25], [0.49, 0.14], [0.53, 0.12], [0.54, 0.09], [0.56, 0.18], [0.47, 0.28], [0.42, 0.37], [0.38, 0.35], [0.38, 0.35]]}" + } + ], + "images": "RH20T_cfg6_task_0029_user_0014_scene_0003_cfg_0006_104122061330#0", + "gt": "[[266, 97], [281, 89], [314, 49], [338, 42], [343, 34], [356, 64], [301, 102], [269, 133], [245, 126], [245, 126]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "91405_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the can and place it on the lower shelf', and now you are performing subtask 'pick up the can and place it on the lower shelf.'\nFrom start point [0.16, 0.56], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.16, 0.56], [0.23, 0.62], [0.26, 0.77], [0.31, 0.73], [0.25, 0.49], [0.28, 0.19], [0.33, 0.09], [0.4, 0.24], [0.32, 0.24], [0.27, 0.38]]}" + } + ], + "images": "91405_exterior_image_1_left#0", + "gt": "[[50, 100], [75, 112], [82, 138], [99, 131], [81, 88], [88, 34], [105, 17], [128, 44], [101, 44], [85, 68]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "5583_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the blue marker into the bowl', and now you are performing subtask 'put the blue marker into the bowl.'\nBegin at [0.7, 0.54] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.54], [0.65, 0.59], [0.54, 0.68], [0.52, 0.72], [0.52, 0.72], [0.52, 0.57], [0.53, 0.49], [0.55, 0.48], [0.56, 0.51], [0.58, 0.51]]}" + } + ], + "images": "5583_exterior_image_2_left#0", + "gt": "[[223, 97], [207, 107], [173, 123], [165, 130], [166, 129], [165, 103], [170, 89], [175, 87], [179, 92], [185, 91]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "55839_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the marker from the white cup', and now you are performing subtask 'remove the marker from the white cup.'\nYour starting position is [0.42, 0.01]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.01], [0.48, 0.12], [0.51, 0.26], [0.51, 0.45], [0.5, 0.27], [0.55, 0.26], [0.62, 0.36], [0.64, 0.55], [0.6, 0.6], [0.54, 0.27]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/55839_exterior_image_1_left#0.jpg", + "gt": "[[135.0, 2.5], [153.0, 22.5], [163.0, 46.5], [164.0, 80.5], [159.0, 48.5], [176.0, 47.5], [199.0, 64.5], [204.0, 99.5], [192.0, 107.5], [174.0, 48.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0033_user_0020_scene_0001_cfg_0007_104122061018#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pushing another small block to the designated position while grasping one small block', and now you are performing subtask 'pushing another small block to the designated position while grasping one small block.'\nThe trajectory starts at [0.57, 0.09]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.09], [0.57, 0.19], [0.55, 0.27], [0.59, 0.44], [0.59, 0.6], [0.51, 0.71], [0.46, 0.6], [0.42, 0.36], [0.44, 0.39], [0.51, 0.22]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0033_user_0020_scene_0001_cfg_0007_104122061018#0.jpg", + "gt": "[[363, 31], [368, 69], [354, 98], [378, 158], [376, 216], [327, 257], [295, 217], [271, 128], [279, 142], [329, 79]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0014_user_0014_scene_0001_cfg_0001_750612070853#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the block from the scale', and now you are performing subtask 'remove the block from the scale.'\nStarting at [0.56, 0.23], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.56, 0.22], [0.56, 0.29], [0.57, 0.27], [0.57, 0.28], [0.61, 0.31], [0.7, 0.46], [0.7, 0.67], [0.69, 0.59], [0.59, 0.39], [0.56, 0.37]]}" + } + ], + "images": "RH20T_cfg1_task_0014_user_0014_scene_0001_cfg_0001_750612070853#0", + "gt": "[[358, 81], [361, 103], [366, 97], [362, 102], [392, 110], [445, 165], [448, 240], [443, 212], [379, 140], [357, 134]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0028_user_0020_scene_0003_cfg_0007_104122061018#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp the handle and close the drawer', and now you are performing subtask 'grasp the handle and close the drawer.'\nYou begin at [0.57, 0.09]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.09], [0.57, 0.12], [0.46, 0.47], [0.45, 0.54], [0.58, 0.68], [0.56, 0.63], [0.54, 0.53], [0.55, 0.49], [0.57, 0.49], [0.55, 0.18]]}" + } + ], + "images": "RH20T_cfg7_task_0028_user_0020_scene_0003_cfg_0007_104122061018#0", + "gt": "[[363, 31], [364, 45], [293, 171], [290, 195], [374, 246], [358, 228], [344, 191], [352, 176], [367, 175], [350, 66]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "57393_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the plush watermelon and put it in the box', and now you are performing subtask 'pick up the plush watermelon and put it in the box.'\nYou begin at [0.6, 0.33]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.33], [0.64, 0.53], [0.71, 0.62], [0.67, 0.6], [0.68, 0.67], [0.64, 0.46], [0.67, 0.36], [0.7, 0.33], [0.74, 0.34], [0.71, 0.33]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/57393_exterior_image_1_left#0.jpg", + "gt": "[[191, 60], [206, 95], [227, 112], [215, 108], [219, 121], [204, 83], [213, 64], [223, 60], [237, 62], [227, 59]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "67778_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the staple remover to the left', and now you are performing subtask 'move the staple remover to the left.'\nYour starting position is [0.71, 0.01]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.71, 0.01], [0.84, 0.05], [0.85, 0.12], [0.79, 0.34], [0.7, 0.43], [0.74, 0.43], [0.62, 0.34], [0.59, 0.33], [0.57, 0.39], [0.52, 0.31]]}" + } + ], + "images": "67778_exterior_image_1_left#0", + "gt": "[[226.0, 1.5], [268.0, 9.5], [273.0, 21.5], [254.0, 60.5], [223.0, 76.5], [236.0, 77.5], [199.0, 60.5], [188.0, 59.5], [183.0, 69.5], [167.0, 55.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "22632_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the towel then fold it again', and now you are performing subtask 'fold the towel.'\nYour starting position is [0.72, 0.6]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.72, 0.6], [0.7, 0.73], [0.57, 0.69], [0.59, 0.72], [0.57, 0.74], [0.55, 0.57], [0.68, 0.38], [0.76, 0.46], [0.85, 0.28], [0.85, 0.28]]}" + } + ], + "images": "22632_exterior_image_1_left#367", + "gt": "[[230, 108], [225, 132], [182, 124], [190, 130], [181, 133], [177, 102], [216, 69], [242, 83], [273, 50], [273, 50]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "37209_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the green object from the box', and now you are performing subtask 'remove the green object from the box.'\nBegin at [0.78, 0.0] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.78, 0.0], [0.8, 0.03], [0.75, 0.11], [0.77, 0.18], [0.75, 0.17], [0.74, 0.23], [0.74, 0.05], [0.7, 0.08], [0.73, 0.18], [0.78, 0.05]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/37209_exterior_image_2_left#0.jpg", + "gt": "[[251, 0], [257, 5], [241, 20], [247, 33], [239, 31], [238, 41], [238, 9], [223, 15], [233, 32], [250, 9]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "48910_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the towel twice', and now you are performing subtask 'fold the towel in half.'\nThe trajectory starts at [0.29, 0.02]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.29, 0.02], [0.45, 0.02], [0.56, 0.19], [0.6, 0.61], [0.55, 0.73], [0.58, 0.59], [0.42, 0.62], [0.5, 0.47], [0.41, 0.52], [0.41, 0.52]]}" + } + ], + "images": "48910_exterior_image_1_left#0", + "gt": "[[93, 3], [144, 4], [180, 35], [191, 110], [177, 132], [185, 107], [133, 112], [161, 84], [132, 93], [132, 93]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "21996_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the bottle from the top part of the segment tray and put it on the transparent box', and now you are performing subtask 'take the bottle from the top part of the segment tray and put it on the transparent box.'\nStarting at [0.27, 0.87], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.27, 0.87], [0.39, 0.67], [0.49, 0.57], [0.5, 0.61], [0.57, 0.49], [0.66, 0.55], [0.6, 0.59], [0.6, 0.64], [0.63, 0.69], [0.57, 0.63]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/21996_exterior_image_2_left#0.jpg", + "gt": "[[85, 157], [124, 120], [158, 103], [159, 110], [184, 88], [212, 99], [192, 107], [191, 116], [202, 125], [182, 114]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0034_user_0014_scene_0007_cfg_0007_104122064161#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'place the fourth block on top of the right block, pick up the fifth block from the desktop, transfer the fifth block from the desktop to the top of the right block, place the fifth block on top of the right block.'\nYour starting position is [0.53, 0.77]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.77], [0.56, 0.62], [0.53, 0.55], [0.56, 0.47], [0.57, 0.49], [0.56, 0.57], [0.54, 0.58], [0.59, 0.59], [0.61, 0.7], [0.59, 0.76]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0034_user_0014_scene_0007_cfg_0007_104122064161#561.jpg", + "gt": "[[339, 276], [357, 222], [339, 199], [358, 169], [366, 177], [358, 207], [343, 210], [379, 212], [388, 252], [378, 273]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "3439_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the pen from the mug cup and place it on the table', and now you are performing subtask 'remove the pen from the mug cup and place it on the table.'\nStarting at [0.52, 0.01], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.01], [0.51, 0.07], [0.54, 0.26], [0.53, 0.29], [0.55, 0.36], [0.56, 0.31], [0.54, 0.23], [0.5, 0.33], [0.5, 0.39], [0.49, 0.36]]}" + } + ], + "images": "3439_exterior_image_1_left#0", + "gt": "[[165.5, 2.0], [163.5, 13.0], [172.5, 47.0], [169.5, 52.0], [175.5, 64.0], [180.5, 55.0], [172.5, 41.0], [158.5, 59.0], [158.5, 70.0], [157.5, 64.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0127_user_0010_scene_0001_cfg_0004_104122062823#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'plug in the power cord of the desk lamp, turn on the socket, and light up the desk lamp', and now you are performing subtask 'insert the power cable into the socket on the table, press the button on the socket to turn it on.'\nThe trajectory starts at [0.47, 0.21]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.21], [0.45, 0.19], [0.46, 0.2], [0.5, 0.27], [0.49, 0.31], [0.5, 0.29], [0.5, 0.31], [0.49, 0.32], [0.45, 0.25], [0.41, 0.21]]}" + } + ], + "images": "RH20T_cfg4_task_0127_user_0010_scene_0001_cfg_0004_104122062823#264", + "gt": "[[300, 74], [285, 70], [294, 73], [318, 98], [314, 111], [317, 105], [320, 111], [313, 117], [291, 90], [262, 77]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0096_user_0007_scene_0008_cfg_0005_104122063678#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take something out of a drawer', and now you are performing subtask 'pull the drawer handle to the outside of the table, pick up the object from the drawer, transfer the object from the drawer to the top of the table, place the object on the table.'\nYour initial coordinate is [0.62, 0.55]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.55], [0.62, 0.64], [0.58, 0.72], [0.59, 0.59], [0.61, 0.56], [0.57, 0.41], [0.57, 0.42], [0.55, 0.43], [0.54, 0.52], [0.56, 0.44]]}" + } + ], + "images": "RH20T_cfg5_task_0096_user_0007_scene_0008_cfg_0005_104122063678#401", + "gt": "[[399, 197], [400, 232], [371, 260], [377, 214], [393, 203], [366, 146], [364, 152], [352, 153], [348, 187], [357, 157]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0038_user_0010_scene_0003_cfg_0003_104122063550#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the cup to the top of the cup rack and then place it down', and now you are performing subtask 'move the cup to the top of the cup rack and then place it down.'\nYour starting position is [0.43, 0.53]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.53], [0.51, 0.47], [0.55, 0.52], [0.53, 0.56], [0.42, 0.6], [0.4, 0.65], [0.41, 0.61], [0.35, 0.64], [0.39, 0.62], [0.37, 0.58]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg3_task_0038_user_0010_scene_0003_cfg_0003_104122063550#0.jpg", + "gt": "[[277, 192], [326, 170], [349, 186], [337, 202], [271, 215], [256, 233], [261, 221], [225, 229], [248, 222], [234, 208]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "13565_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the towel', and now you are performing subtask 'move the towel.'\nStart at [0.6, 0.01]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.01], [0.68, 0.19], [0.74, 0.45], [0.72, 0.43], [0.73, 0.46], [0.72, 0.56], [0.69, 0.61], [0.66, 0.62], [0.53, 0.53], [0.54, 0.44]]}" + } + ], + "images": "13565_exterior_image_1_left#0", + "gt": "[[192, 2], [217, 34], [237, 81], [229, 78], [234, 83], [230, 100], [222, 110], [212, 111], [170, 95], [172, 80]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0003_user_0014_scene_0010_cfg_0006_104122060811#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the red button at the bottom.'\nBegin at [0.6, 0.35] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.35], [0.6, 0.34], [0.61, 0.34], [0.61, 0.32], [0.62, 0.32], [0.61, 0.31], [0.61, 0.31], [0.63, 0.31], [0.61, 0.3], [0.61, 0.3]]}" + } + ], + "images": "RH20T_cfg6_task_0003_user_0014_scene_0010_cfg_0006_104122060811#0", + "gt": "[[385, 127], [386, 123], [392, 121], [391, 117], [396, 115], [392, 113], [393, 111], [401, 113], [390, 108], [390, 108]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_819112070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the first block from the table, transfer the first block from the table to the position above the second block, place the first block on top of the second block.'\nYou start from [0.62, 0.46]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.46], [0.6, 0.47], [0.58, 0.51], [0.56, 0.51], [0.55, 0.57], [0.57, 0.49], [0.56, 0.5], [0.52, 0.56], [0.52, 0.58], [0.51, 0.61]]}" + } + ], + "images": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_819112070044#0", + "gt": "[[400, 167], [383, 170], [373, 182], [357, 184], [352, 205], [365, 178], [359, 180], [331, 203], [333, 208], [328, 218]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "14166_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the towel from the bar and put it on the seat', and now you are performing subtask 'pick up the towel from the bar.'\nYou begin at [0.5, 0.46]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.46], [0.44, 0.5], [0.41, 0.42], [0.4, 0.43], [0.41, 0.45], [0.42, 0.34], [0.47, 0.26], [0.45, 0.41], [0.47, 0.44], [0.34, 0.45]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/14166_exterior_image_2_left#0.jpg", + "gt": "[[161, 82], [140, 90], [131, 76], [127, 78], [131, 81], [133, 62], [149, 47], [145, 74], [152, 79], [109, 81]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "76853_exterior_image_2_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the towel and put it on the black rack', and now you are performing subtask 'place the towel on the black rack.'\nFrom start point [0.78, 0.63], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.78, 0.63], [0.73, 0.62], [0.71, 0.7], [0.68, 0.37], [0.7, 0.31], [0.77, 0.31], [0.81, 0.42], [0.87, 0.43], [0.9, 0.56], [0.86, 0.58]]}" + } + ], + "images": "76853_exterior_image_2_left#318", + "gt": "[[249, 113], [233, 112], [228, 126], [216, 66], [224, 56], [246, 55], [260, 76], [279, 78], [288, 100], [274, 104]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0104_user_0020_scene_0006_cfg_0006_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'drag the silver plate under the left white object to the left side of the white plate', and now you are performing subtask 'drag the silver plate under the left white object to the left side of the white plate.'\nWith start location [0.53, 0.31], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.31], [0.52, 0.26], [0.55, 0.26], [0.58, 0.31], [0.59, 0.45], [0.58, 0.54], [0.59, 0.49], [0.56, 0.41], [0.56, 0.29], [0.55, 0.34]]}" + } + ], + "images": "RH20T_cfg6_task_0104_user_0020_scene_0006_cfg_0006_104122064161#0", + "gt": "[[338, 111], [335, 93], [354, 93], [370, 113], [379, 163], [374, 196], [378, 178], [361, 148], [360, 105], [352, 124]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "54529_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the shirt', and now you are performing subtask 'fold the shirt.'\nYou begin at [0.24, 0.5]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.24, 0.5], [0.39, 0.53], [0.62, 0.74], [0.57, 0.72], [0.53, 0.86], [0.55, 0.78], [0.52, 0.67], [0.46, 0.68], [0.52, 0.71], [0.42, 0.77]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/54529_exterior_image_2_left#0.jpg", + "gt": "[[78, 90], [126, 95], [197, 134], [182, 129], [171, 154], [176, 141], [166, 121], [148, 122], [165, 128], [134, 139]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_038522063145#TrajectoryPrediction_Task#12", + "task": "TrajectoryPrediction_Task", + "p_index": 12, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with small Lego blocks for the nth time with n blocks', and now you are performing subtask 'place the blue small Lego block on the white small Lego block, press the blue small Lego block, pick up the red small Lego block from the desktop.'\nYou start from [0.7, 0.44]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.44], [0.7, 0.44], [0.7, 0.42], [0.7, 0.44], [0.7, 0.43], [0.71, 0.43], [0.7, 0.38], [0.7, 0.36], [0.7, 0.37], [0.69, 0.37]]}" + } + ], + "images": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_038522063145#1411", + "gt": "[[450, 157], [446, 159], [448, 151], [450, 157], [448, 155], [452, 155], [447, 138], [449, 131], [448, 134], [442, 134]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "14083_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'spoon out some contents from the bowl on the left using the silver spoon and place them in the bowl on the right', and now you are performing subtask 'pick up the silver spoon from the bowl.'\nYou begin at [0.2, 0.38]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.2, 0.38], [0.24, 0.41], [0.25, 0.43], [0.24, 0.45], [0.27, 0.45], [0.28, 0.48], [0.3, 0.49], [0.31, 0.52], [0.31, 0.49], [0.3, 0.48]]}" + } + ], + "images": "14083_exterior_image_2_left#0", + "gt": "[[63, 69], [77, 74], [80, 77], [78, 81], [85, 81], [89, 86], [97, 89], [99, 93], [99, 89], [96, 86]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0013_user_0020_scene_0010_cfg_0007_104122061602#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the block on the scale', and now you are performing subtask 'pick up the big block from the table, transfer the big block from the table to the scale, place the big block on the scale.'\nThe trajectory starts at [0.7, 0.32]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.32], [0.71, 0.36], [0.7, 0.37], [0.7, 0.39], [0.76, 0.46], [0.83, 0.58], [0.82, 0.6], [0.8, 0.52], [0.73, 0.49], [0.74, 0.47]]}" + } + ], + "images": "RH20T_cfg7_task_0013_user_0020_scene_0010_cfg_0007_104122061602#0", + "gt": "[[448, 116], [455, 130], [447, 134], [449, 142], [486, 165], [531, 208], [528, 217], [515, 189], [470, 177], [473, 171]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "78597_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the towel', and now you are performing subtask 'fold the towel on the countertop.'\nBegin at [0.57, 0.51] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.51], [0.58, 0.54], [0.62, 0.61], [0.61, 0.61], [0.6, 0.62], [0.61, 0.63], [0.59, 0.65], [0.59, 0.58], [0.6, 0.56], [0.61, 0.56]]}" + } + ], + "images": "78597_exterior_image_1_left#304", + "gt": "[[182, 92], [185, 98], [199, 110], [196, 109], [191, 112], [195, 113], [189, 117], [189, 105], [191, 101], [194, 101]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_104122063633#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'place the object on the tabletop, pick up the object in the box, transfer the object from the box to the tabletop, place the object on the tabletop, pick up the cube in the box.'\nYou start from [0.28, 0.44]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.28, 0.44], [0.27, 0.49], [0.27, 0.52], [0.28, 0.38], [0.3, 0.34], [0.32, 0.36], [0.32, 0.4], [0.32, 0.37], [0.3, 0.38], [0.32, 0.53]]}" + } + ], + "images": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_104122063633#389", + "gt": "[[177, 157], [170, 175], [172, 189], [178, 137], [192, 123], [204, 130], [206, 143], [202, 132], [193, 137], [202, 191]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "78664_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the red cup to the left side of the pot lid', and now you are performing subtask 'move the red cup to the left side of the pot lid.'\nWith start location [0.38, 0.26], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.26], [0.42, 0.3], [0.35, 0.51], [0.42, 0.56], [0.44, 0.66], [0.43, 0.22], [0.32, 0.23], [0.35, 0.23], [0.32, 0.4], [0.31, 0.31]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/78664_exterior_image_1_left#0.jpg", + "gt": "[[122.0, 47.0], [135.0, 54.0], [111.0, 92.0], [133.0, 101.0], [142.0, 119.0], [139.0, 40.0], [102.0, 42.0], [113.0, 42.0], [103.0, 72.0], [100.0, 56.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0096_user_0007_scene_0008_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take something out of a drawer', and now you are performing subtask 'pull the drawer handle to the outside of the table.'\nWith start location [0.5, 0.37], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.52, 0.33], [0.59, 0.36], [0.62, 0.54], [0.62, 0.47], [0.64, 0.58], [0.6, 0.71], [0.59, 0.54], [0.6, 0.49], [0.62, 0.55]]}" + } + ], + "images": "RH20T_cfg5_task_0096_user_0007_scene_0008_cfg_0005_104122063678#0", + "gt": "[[323, 132], [336, 120], [379, 129], [398, 196], [394, 170], [412, 210], [384, 254], [377, 193], [386, 176], [399, 197]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "15001_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the glue stick inside the cup then take it out', and now you are performing subtask 'pick up the glue stick in the cup.'\nYou begin at [0.84, 0.29]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.84, 0.29], [0.83, 0.46], [0.74, 0.63], [0.76, 0.69], [0.77, 0.59], [0.71, 0.49], [0.61, 0.41], [0.54, 0.41], [0.53, 0.47], [0.54, 0.43]]}" + } + ], + "images": "15001_exterior_image_1_left#0", + "gt": "[[270, 52], [265, 83], [237, 113], [243, 125], [246, 107], [227, 88], [196, 73], [174, 73], [171, 85], [172, 77]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0005_user_0008_scene_0001_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'approach and touch the side of the small block', and now you are performing subtask 'approach and touch the side of the small block.'\nYou start from [0.63, 0.38]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.38], [0.64, 0.37], [0.63, 0.37], [0.64, 0.4], [0.63, 0.54], [0.65, 0.62], [0.65, 0.69], [0.65, 0.64], [0.64, 0.53], [0.64, 0.53]]}" + } + ], + "images": "RH20T_cfg1_task_0005_user_0008_scene_0001_cfg_0001_038522062547#0", + "gt": "[[405, 135], [407, 134], [404, 134], [408, 144], [403, 195], [418, 223], [418, 247], [419, 231], [408, 192], [411, 191]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "75812_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the clothes in the bag', and now you are performing subtask 'place the clothes in the bag.'\nFrom start point [0.49, 0.75], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.75], [0.47, 0.98], [0.45, 0.64], [0.57, 0.15], [0.58, 0.69], [0.5, 0.73], [0.54, 0.73], [0.52, 0.63], [0.57, 0.56], [0.57, 0.56]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/75812_exterior_image_1_left#141.jpg", + "gt": "[[156.5, 135.0], [149.5, 177.0], [142.5, 115.0], [183.5, 27.0], [185.5, 125.0], [161.5, 131.0], [173.5, 131.0], [166.5, 113.0], [181.5, 101.0], [181.5, 101.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "79883_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the gray towel from the chair and place it on the top rail', and now you are performing subtask 'take the gray towel from the chair and place it on the top rail.'\nYou begin at [0.6, 0.3]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.3], [0.61, 0.44], [0.58, 0.43], [0.65, 0.56], [0.57, 0.63], [0.57, 0.71], [0.55, 0.18], [0.51, 0.13], [0.51, 0.4], [0.49, 0.3]]}" + } + ], + "images": "79883_exterior_image_1_left#0", + "gt": "[[192, 54], [195, 80], [187, 77], [208, 100], [183, 113], [182, 128], [175, 33], [163, 24], [162, 72], [156, 54]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "18559_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the lid on the table', and now you are performing subtask 'pick up the lid from the bowl.'\nBegin at [0.02, 0.49] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.02, 0.49], [0.08, 0.52], [0.15, 0.58], [0.26, 0.63], [0.29, 0.68], [0.29, 0.71], [0.29, 0.74], [0.3, 0.73], [0.3, 0.76], [0.32, 0.73]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/18559_exterior_image_1_left#0.jpg", + "gt": "[[6.0, 88.0], [25.0, 93.0], [49.0, 105.0], [82.0, 114.0], [92.0, 122.0], [94.0, 128.0], [92.0, 134.0], [95.0, 132.0], [97.0, 136.0], [104.0, 132.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "77557_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the trash bin', and now you are performing subtask 'open the trash bin.'\nYou begin at [0.54, 0.42]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.42], [0.57, 0.43], [0.58, 0.46], [0.57, 0.46], [0.57, 0.58], [0.55, 0.64], [0.55, 0.68], [0.56, 0.54], [0.54, 0.51], [0.54, 0.51]]}" + } + ], + "images": "77557_exterior_image_2_left#0", + "gt": "[[172, 76], [183, 77], [187, 83], [184, 83], [184, 105], [176, 115], [176, 122], [179, 98], [173, 92], [174, 91]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0129_user_0020_scene_0009_cfg_0006_104122063633#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'plug in the bulb holder with a bulb to the socket', and now you are performing subtask 'plug in the bulb holder with a bulb to the socket.'\nStarting at [0.36, 0.33], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.33], [0.34, 0.42], [0.34, 0.48], [0.28, 0.62], [0.22, 0.61], [0.24, 0.68], [0.22, 0.57], [0.24, 0.37], [0.28, 0.36], [0.31, 0.42]]}" + } + ], + "images": "RH20T_cfg6_task_0129_user_0020_scene_0009_cfg_0006_104122063633#0", + "gt": "[[232, 119], [216, 151], [220, 172], [182, 225], [139, 219], [154, 244], [143, 206], [151, 133], [182, 131], [200, 152]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0011_user_0007_scene_0004_cfg_0005_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'water the plant', and now you are performing subtask 'water the plant.'\nYou start from [0.49, 0.14]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.14], [0.53, 0.0], [0.57, 0.0], [0.61, 0.08], [0.62, 0.21], [0.6, 0.0], [0.45, 0.0], [0.6, 0.0], [0.6, 0.04], [0.56, 0.0]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0011_user_0007_scene_0004_cfg_0005_104122061850#0.jpg", + "gt": "[[311, 50], [339, 0], [366, 1], [393, 27], [399, 74], [385, 0], [288, 0], [382, 0], [383, 14], [359, 0]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0071_user_0007_scene_0006_cfg_0002_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close a sliding window', and now you are performing subtask 'close a sliding window.'\nBegin at [0.62, 0.54] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.54], [0.59, 0.52], [0.59, 0.57], [0.6, 0.59], [0.63, 0.61], [0.59, 0.58], [0.61, 0.59], [0.6, 0.53], [0.62, 0.54], [0.64, 0.62]]}" + } + ], + "images": "RH20T_cfg2_task_0071_user_0007_scene_0006_cfg_0002_104122063678#0", + "gt": "[[400, 196], [377, 189], [378, 204], [381, 213], [405, 218], [377, 210], [388, 211], [383, 192], [396, 196], [411, 224]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "64105_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the cloth on the countertop into a triangle twice', and now you are performing subtask 'fold the cloth on the countertop into a triangle twice.'\nYou start from [0.76, 0.42]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.76, 0.42], [0.65, 0.68], [0.62, 0.61], [0.5, 0.73], [0.65, 0.78], [0.69, 0.85], [0.66, 0.94], [0.73, 0.97], [0.54, 0.69], [0.6, 0.44]]}" + } + ], + "images": "64105_exterior_image_1_left#0", + "gt": "[[244, 75], [207, 122], [198, 109], [160, 132], [209, 140], [222, 153], [212, 170], [235, 175], [173, 125], [192, 79]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "27569_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the green cloth', and now you are performing subtask 'fold the green cloth.'\nFrom start point [0.78, 0.04], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.78, 0.04], [0.65, 0.37], [0.63, 0.52], [0.6, 0.51], [0.62, 0.52], [0.61, 0.43], [0.64, 0.39], [0.73, 0.45], [0.73, 0.5], [0.73, 0.36]]}" + } + ], + "images": "27569_exterior_image_2_left#0", + "gt": "[[251, 8], [208, 67], [203, 94], [191, 92], [198, 93], [195, 77], [206, 71], [233, 81], [234, 90], [234, 65]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "43138_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the silver pot to the left', and now you are performing subtask 'move the silver pot to the left.'\nYour initial coordinate is [0.72, 0.01]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.73, 0.01], [0.72, 0.21], [0.71, 0.31], [0.71, 0.38], [0.7, 0.33], [0.68, 0.29], [0.62, 0.27], [0.58, 0.27], [0.59, 0.22], [0.6, 0.21]]}" + } + ], + "images": "43138_exterior_image_1_left#0", + "gt": "[[232, 2], [231, 37], [226, 56], [227, 68], [225, 60], [216, 53], [200, 48], [186, 49], [188, 39], [192, 37]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_819112070044#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'putting a portion of the Lego blocks on the Lego baseplate', and now you are performing subtask 'insert the orange Lego brick into the Lego base, pick up the second red Lego brick from the table.'\nStarting at [0.6, 0.64], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.64], [0.6, 0.57], [0.61, 0.51], [0.59, 0.48], [0.6, 0.46], [0.6, 0.47], [0.59, 0.45], [0.58, 0.45], [0.57, 0.51], [0.57, 0.45]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_819112070044#477.jpg", + "gt": "[[386, 230], [385, 204], [393, 185], [379, 173], [386, 166], [382, 170], [377, 161], [371, 162], [367, 183], [368, 161]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "15364_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the marker from the glass cup and put it on the table', and now you are performing subtask 'pick up the marker from the glass cup and put it on the table.'\nStarting at [0.92, 0.32], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.92, 0.32], [0.61, 0.34], [0.48, 0.48], [0.39, 0.67], [0.4, 0.72], [0.39, 0.63], [0.49, 0.58], [0.51, 0.6], [0.49, 0.65], [0.47, 0.6]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/15364_exterior_image_2_left#0.jpg", + "gt": "[[293, 57], [194, 61], [154, 87], [125, 120], [129, 129], [126, 113], [157, 105], [163, 108], [157, 117], [149, 108]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_037522061512#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'place the block on the other blocks, pick up the block on the flat surface, transfer the block from the flat surface to the other blocks, place the block on the other blocks.'\nBegin at [0.38, 0.33] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.33], [0.3, 0.38], [0.29, 0.43], [0.29, 0.39], [0.35, 0.33], [0.36, 0.32], [0.38, 0.35], [0.4, 0.4], [0.38, 0.38], [0.36, 0.34]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_037522061512#606.jpg", + "gt": "[[241, 119], [194, 136], [184, 154], [186, 139], [221, 118], [233, 117], [246, 125], [256, 143], [244, 138], [231, 122]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "91740_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push the plastic wrap off the container, then pick up the container and move it slightly up onto the table', and now you are performing subtask 'place the container on the plastic wrap.'\nWith start location [0.47, 0.56], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.56], [0.49, 0.61], [0.51, 0.63], [0.51, 0.64], [0.46, 0.55], [0.44, 0.55], [0.43, 0.53], [0.43, 0.54], [0.47, 0.58], [0.5, 0.57]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/91740_exterior_image_1_left#302.jpg", + "gt": "[[151.0, 101.0], [157.0, 110.0], [164.0, 113.0], [162.0, 115.0], [146.0, 99.0], [142.0, 99.0], [139.0, 95.0], [138.0, 97.0], [150.0, 104.0], [161.0, 102.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "31545_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the top drawer on the left', and now you are performing subtask 'close the top drawer on the left.'\nYour starting position is [0.33, 0.67]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.33, 0.67], [0.36, 0.64], [0.42, 0.51], [0.43, 0.47], [0.43, 0.44], [0.42, 0.42], [0.42, 0.39], [0.43, 0.39], [0.44, 0.43], [0.46, 0.44]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/31545_exterior_image_2_left#0.jpg", + "gt": "[[105, 120], [114, 115], [134, 92], [138, 85], [138, 79], [135, 76], [134, 71], [137, 71], [141, 77], [147, 79]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0034_user_0003_scene_0006_cfg_0002_104122063678#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'transfer the third square from the large square to the right side of the first square, place the third square to the right of the first square, pick up the fourth square from the countertop, transfer the fourth square from the countertop to the first and third squares, place the fourth square on the first and third squares.'\nThe trajectory starts at [0.55, 0.95]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.95], [0.55, 0.91], [0.58, 0.86], [0.64, 0.82], [0.64, 0.79], [0.64, 0.82], [0.63, 0.81], [0.57, 0.85], [0.56, 0.9], [0.6, 0.87]]}" + } + ], + "images": "RH20T_cfg2_task_0034_user_0003_scene_0006_cfg_0002_104122063678#548", + "gt": "[[355, 341], [355, 327], [374, 308], [409, 294], [412, 284], [412, 297], [402, 292], [367, 306], [358, 325], [384, 314]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "59486_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the black masking tape on the cream plate', and now you are performing subtask 'put the black masking tape on the cream plate.'\nYour starting position is [0.26, 0.05]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.26, 0.05], [0.25, 0.18], [0.24, 0.22], [0.28, 0.33], [0.28, 0.41], [0.3, 0.44], [0.33, 0.34], [0.35, 0.33], [0.43, 0.5], [0.41, 0.43]]}" + } + ], + "images": "59486_exterior_image_1_left#0", + "gt": "[[82.5, 9.0], [80.5, 32.0], [75.5, 39.0], [90.5, 60.0], [89.5, 74.0], [96.5, 80.0], [104.5, 61.0], [113.5, 59.0], [137.5, 90.0], [132.5, 77.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_038522062547#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'pick up the fourth square on the table, pick up the fourth square from the right side of the table, transfer the fourth square from the right side of the table to the top of the first square, place the fourth square on top of the first square, pick up the fifth square from the right side of the table.'\nYour initial coordinate is [0.71, 0.47]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.71, 0.47], [0.72, 0.48], [0.72, 0.51], [0.71, 0.49], [0.72, 0.5], [0.68, 0.52], [0.65, 0.51], [0.62, 0.52], [0.62, 0.46], [0.67, 0.44]]}" + } + ], + "images": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_038522062547#524", + "gt": "[[454, 170], [462, 174], [460, 185], [453, 177], [461, 181], [432, 189], [417, 184], [396, 188], [397, 165], [428, 159]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "66497_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the metal can out of the cupboard', and now you are performing subtask 'take the metal can out of the cupboard.'\nYou start from [0.16, 0.41]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.16, 0.41], [0.23, 0.21], [0.33, 0.15], [0.39, 0.18], [0.36, 0.12], [0.34, 0.12], [0.31, 0.21], [0.37, 0.74], [0.33, 0.73], [0.32, 0.66]]}" + } + ], + "images": "66497_exterior_image_1_left#0", + "gt": "[[51, 74], [74, 38], [105, 27], [124, 33], [115, 22], [108, 22], [100, 37], [119, 133], [107, 131], [103, 119]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0008_user_0007_scene_0001_cfg_0005_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grab the block and place it at the designated location', and now you are performing subtask 'grab the block and place it at the designated location.'\nYou begin at [0.47, 0.4]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.4], [0.43, 0.41], [0.33, 0.3], [0.26, 0.16], [0.27, 0.34], [0.27, 0.29], [0.29, 0.35], [0.33, 0.6], [0.36, 0.46], [0.35, 0.4]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0008_user_0007_scene_0001_cfg_0005_037522062165#0.jpg", + "gt": "[[298, 144], [273, 146], [210, 109], [164, 57], [172, 124], [171, 104], [187, 126], [211, 215], [233, 167], [225, 143]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "76696_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the blue cup to the right', and now you are performing subtask 'move the blue cup to the right.'\nYou begin at [0.69, 0.0]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.0], [0.64, 0.22], [0.55, 0.41], [0.53, 0.54], [0.54, 0.66], [0.47, 0.48], [0.47, 0.44], [0.49, 0.43], [0.5, 0.49], [0.5, 0.39]]}" + } + ], + "images": "76696_exterior_image_2_left#0", + "gt": "[[220, 0], [206, 39], [175, 74], [169, 97], [173, 118], [150, 86], [150, 79], [157, 78], [160, 88], [160, 71]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "13748_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the sock on the plate and put it near the window', and now you are performing subtask 'take the sock on the plate and put it near the window.'\nYour initial coordinate is [0.25, 0.31]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.25, 0.31], [0.24, 0.4], [0.31, 0.38], [0.33, 0.48], [0.35, 0.36], [0.51, 0.43], [0.54, 0.49], [0.5, 0.51], [0.51, 0.54], [0.51, 0.42]]}" + } + ], + "images": "13748_exterior_image_2_left#0", + "gt": "[[80, 55], [77, 72], [98, 69], [105, 87], [112, 65], [164, 78], [172, 88], [161, 92], [164, 98], [162, 76]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "5631_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the longer rectangular block on top of the cube', and now you are performing subtask 'put the longer rectangular block on top of the cube.'\nFrom start point [0.83, 0.0], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.83, 0.0], [0.8, 0.2], [0.73, 0.26], [0.73, 0.31], [0.7, 0.23], [0.65, 0.26], [0.68, 0.28], [0.67, 0.33], [0.68, 0.14], [0.74, 0.01]]}" + } + ], + "images": "5631_exterior_image_2_left#0", + "gt": "[[265, 0], [255, 36], [235, 47], [232, 56], [223, 42], [209, 47], [217, 50], [215, 60], [219, 25], [237, 2]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "57652_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'hang up the cloth on the glass barrier', and now you are performing subtask 'place the cloth on the glass barrier.'\nWith start location [0.63, 0.68], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.68], [0.63, 0.71], [0.6, 0.64], [0.55, 0.74], [0.52, 0.42], [0.52, 0.22], [0.49, 0.05], [0.48, 0.1], [0.51, 0.22], [0.52, 0.14]]}" + } + ], + "images": "57652_exterior_image_1_left#186", + "gt": "[[203, 122], [201, 128], [191, 116], [177, 134], [166, 76], [166, 40], [158, 9], [154, 18], [162, 39], [165, 26]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "47675_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn off the switch on the extension cord then close the clear container', and now you are performing subtask 'press the switch on the power strip to turn it off.'\nYou begin at [0.88, 0.0]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.88, 0.0], [0.77, 0.12], [0.72, 0.24], [0.68, 0.28], [0.7, 0.29], [0.67, 0.33], [0.63, 0.4], [0.59, 0.44], [0.61, 0.53], [0.57, 0.38]]}" + } + ], + "images": "47675_exterior_image_1_left#0", + "gt": "[[280, 0], [245, 22], [229, 44], [219, 51], [224, 52], [213, 59], [203, 72], [190, 80], [194, 95], [182, 69]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0039_user_0014_scene_0008_cfg_0007_104122063633#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pour the water from one cup into another empty cup', and now you are performing subtask 'pour the water from one cup into another empty cup.'\nYou start from [0.37, 0.34]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.34], [0.33, 0.32], [0.33, 0.25], [0.32, 0.34], [0.34, 0.24], [0.33, 0.36], [0.34, 0.28], [0.37, 0.45], [0.36, 0.62], [0.36, 0.29]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0039_user_0014_scene_0008_cfg_0007_104122063633#0.jpg", + "gt": "[[234, 121], [213, 114], [211, 90], [204, 122], [215, 87], [212, 130], [219, 102], [239, 161], [229, 223], [230, 103]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0034_user_0014_scene_0001_cfg_0007_104122064161#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'pick up the square, place the square on another square.'\nStart at [0.59, 0.65]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.65], [0.59, 0.67], [0.57, 0.7], [0.56, 0.75], [0.58, 0.66], [0.59, 0.55], [0.58, 0.49], [0.56, 0.47], [0.55, 0.48], [0.58, 0.49]]}" + } + ], + "images": "RH20T_cfg7_task_0034_user_0014_scene_0001_cfg_0007_104122064161#339", + "gt": "[[375, 234], [375, 242], [362, 253], [356, 269], [370, 236], [376, 199], [370, 176], [360, 171], [355, 174], [369, 175]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "71190_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the orange cloth', and now you are performing subtask 'fold the orange cloth on the table.'\nThe trajectory starts at [0.37, 0.51]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.51], [0.44, 0.63], [0.41, 0.77], [0.38, 0.74], [0.4, 0.76], [0.4, 0.82], [0.42, 0.76], [0.45, 0.96], [0.48, 1.0], [0.48, 0.84]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/71190_exterior_image_2_left#0.jpg", + "gt": "[[117, 92], [140, 114], [131, 139], [120, 133], [128, 137], [128, 148], [135, 136], [145, 173], [154, 180], [154, 152]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "8853_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the plush toy from the blue basket, put it in the open drawer, and then close the drawer', and now you are performing subtask 'push the drawer to close it.'\nYour initial coordinate is [0.75, 0.56]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.75, 0.56], [0.7, 0.42], [0.68, 0.51], [0.76, 0.61], [0.68, 0.62], [0.71, 0.49], [0.74, 0.63], [0.57, 0.68], [0.68, 0.51], [0.66, 0.46]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/8853_exterior_image_1_left#189.jpg", + "gt": "[[239.5, 100.0], [223.5, 76.0], [218.5, 92.0], [243.5, 110.0], [218.5, 112.0], [228.5, 89.0], [236.5, 113.0], [181.5, 122.0], [218.5, 91.0], [211.5, 83.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg3_task_0058_user_0010_scene_0010_cfg_0003_104122063550#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn off the power strips', and now you are performing subtask 'press the button on the power strip above, press the button on the power strip above.'\nStarting at [0.44, 0.53], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.53], [0.5, 0.47], [0.42, 0.49], [0.42, 0.44], [0.45, 0.38], [0.53, 0.29], [0.54, 0.32], [0.53, 0.39], [0.55, 0.37], [0.55, 0.37]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg3_task_0058_user_0010_scene_0010_cfg_0003_104122063550#0.jpg", + "gt": "[[280, 191], [317, 168], [266, 175], [271, 157], [286, 138], [339, 105], [347, 117], [339, 139], [353, 134], [353, 134]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "49270_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the marker from the cup, move the cup to the left and put the marker in the cup', and now you are performing subtask 'pick up the marker from the table.'\nYour initial coordinate is [0.53, 0.26]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.26], [0.63, 0.39], [0.64, 0.4], [0.64, 0.37], [0.64, 0.37], [0.64, 0.38], [0.63, 0.38], [0.64, 0.37], [0.62, 0.38], [0.63, 0.4]]}" + } + ], + "images": "49270_exterior_image_1_left#372", + "gt": "[[171, 46], [203, 71], [205, 72], [204, 67], [205, 66], [204, 68], [203, 68], [204, 67], [200, 68], [201, 72]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "72089_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the green bowl behind the wine glass', and now you are performing subtask 'put the green bowl behind the wine glass.'\nStarting at [0.8, 0.52], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.8, 0.52], [0.77, 0.36], [0.81, 0.34], [0.84, 0.44], [0.78, 0.57], [0.78, 0.38], [0.7, 0.39], [0.68, 0.56], [0.69, 0.61], [0.71, 0.5]]}" + } + ], + "images": "72089_exterior_image_1_left#0", + "gt": "[[255.0, 93.5], [248.0, 65.5], [259.0, 60.5], [270.0, 78.5], [250.0, 103.5], [249.0, 67.5], [225.0, 69.5], [218.0, 100.5], [220.0, 110.5], [227.0, 90.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "69829_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the open drawer on the right', and now you are performing subtask 'close the open drawer on the right.'\nYour starting position is [0.72, 0.01]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.72, 0.01], [0.8, 0.05], [0.78, 0.06], [0.8, 0.1], [0.9, 0.19], [0.9, 0.25], [0.99, 0.27], [0.95, 0.37], [0.84, 0.39], [0.84, 0.39]]}" + } + ], + "images": "69829_exterior_image_1_left#0", + "gt": "[[231, 1], [255, 9], [250, 11], [257, 18], [289, 35], [288, 45], [316, 49], [305, 67], [270, 70], [270, 70]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "15117_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick the black lid and put it on the bottle', and now you are performing subtask 'pick the black lid and put it on the bottle.'\nYou begin at [0.76, 0.04]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.76, 0.04], [0.72, 0.28], [0.75, 0.44], [0.71, 0.49], [0.69, 0.22], [0.56, 0.15], [0.59, 0.09], [0.55, 0.18], [0.56, 0.1], [0.56, 0.1]]}" + } + ], + "images": "15117_exterior_image_2_left#0", + "gt": "[[242, 8], [229, 51], [241, 79], [227, 88], [222, 39], [180, 27], [188, 16], [177, 32], [178, 18], [178, 18]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0060_user_0014_scene_0006_cfg_0006_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn the knob to decrease the volume of the speaker', and now you are performing subtask 'turn the knob to decrease the volume of the speaker.'\nYou start from [0.53, 0.33]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.32], [0.56, 0.41], [0.57, 0.46], [0.55, 0.44], [0.56, 0.46], [0.55, 0.5], [0.56, 0.47], [0.56, 0.44], [0.54, 0.38], [0.5, 0.29]]}" + } + ], + "images": "RH20T_cfg6_task_0060_user_0014_scene_0006_cfg_0006_104122064161#0", + "gt": "[[337, 117], [358, 149], [364, 164], [354, 159], [359, 164], [355, 180], [361, 171], [359, 159], [346, 136], [318, 103]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "69026_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the lid from the pot', and now you are performing subtask 'remove the lid from the pot.'\nYou start from [0.94, 0.64]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.94, 0.64], [0.88, 0.5], [0.91, 0.52], [0.88, 0.51], [0.85, 0.62], [0.77, 0.57], [0.79, 0.68], [0.77, 0.58], [0.63, 0.79], [0.61, 0.64]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/69026_exterior_image_1_left#0.jpg", + "gt": "[[301, 116], [283, 90], [290, 93], [280, 92], [272, 111], [247, 103], [253, 123], [248, 105], [203, 143], [194, 115]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "37507_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the second drawer from the right, pick up the clear lid from the counter and place it in the open drawer', and now you are performing subtask 'pull the second drawer from the right to open the drawer.'\nWith start location [0.46, 0.4], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.4], [0.38, 0.53], [0.37, 0.58], [0.41, 0.67], [0.45, 0.64], [0.46, 0.64], [0.45, 0.63], [0.37, 0.74], [0.33, 0.66], [0.33, 0.6]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/37507_exterior_image_1_left#0.jpg", + "gt": "[[146.5, 72.0], [122.5, 96.0], [119.5, 105.0], [132.5, 121.0], [142.5, 115.0], [147.5, 116.0], [144.5, 114.0], [118.5, 134.0], [104.5, 118.0], [104.5, 108.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0008_user_0014_scene_0007_cfg_0006_104122061330#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grab the block and place it at the designated location', and now you are performing subtask 'grab the block and place it at the designated location.'\nWith start location [0.42, 0.27], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.27], [0.42, 0.37], [0.42, 0.41], [0.45, 0.46], [0.46, 0.55], [0.44, 0.32], [0.48, 0.22], [0.51, 0.21], [0.51, 0.25], [0.49, 0.22]]}" + } + ], + "images": "RH20T_cfg6_task_0008_user_0014_scene_0007_cfg_0006_104122061330#0", + "gt": "[[266, 97], [267, 133], [271, 146], [287, 166], [292, 198], [280, 115], [307, 78], [326, 76], [325, 89], [315, 79]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "13970_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the silver object on the sink drain', and now you are performing subtask 'put the silver object on the sink drain.'\nStarting at [0.86, 0.41], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.86, 0.41], [0.83, 0.37], [0.84, 0.39], [0.77, 0.4], [0.69, 0.47], [0.69, 0.51], [0.6, 0.55], [0.61, 0.59], [0.62, 0.56], [0.66, 0.54]]}" + } + ], + "images": "13970_exterior_image_2_left#0", + "gt": "[[276, 74], [267, 66], [268, 70], [246, 72], [222, 84], [221, 91], [193, 99], [195, 106], [197, 100], [210, 98]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "66111_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the small bottle backward and then place the big bottle beside the small bottle', and now you are performing subtask 'pick up the big bottle on the table.'\nYour starting position is [0.55, 0.49]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.49], [0.44, 0.44], [0.38, 0.55], [0.33, 0.53], [0.31, 0.65], [0.32, 0.59], [0.39, 0.59], [0.46, 0.73], [0.41, 0.6], [0.42, 0.54]]}" + } + ], + "images": "66111_exterior_image_1_left#0", + "gt": "[[175.0, 87.5], [141.0, 78.5], [123.0, 99.5], [106.0, 95.5], [100.0, 116.5], [104.0, 105.5], [126.0, 105.5], [148.0, 131.5], [131.0, 107.5], [135.0, 97.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "87301_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pour the fries out of the light green cup and onto the white plate', and now you are performing subtask 'pour the fries out of the light green cup and onto the white plate.'\nStarting at [0.43, 0.28], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.28], [0.27, 0.47], [0.3, 0.53], [0.29, 0.38], [0.2, 0.41], [0.22, 0.51], [0.18, 0.48], [0.2, 0.25], [0.33, 0.57], [0.33, 0.39]]}" + } + ], + "images": "87301_exterior_image_1_left#0", + "gt": "[[138.5, 50.0], [87.5, 85.0], [94.5, 96.0], [92.5, 69.0], [62.5, 74.0], [69.5, 91.0], [56.5, 87.0], [63.5, 45.0], [105.5, 102.0], [105.5, 70.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "62473_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the cup to the left side of the table', and now you are performing subtask 'move the cup to the left side of the table.'\nFrom start point [0.64, 0.29], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.64, 0.29], [0.6, 0.33], [0.57, 0.31], [0.57, 0.35], [0.51, 0.41], [0.51, 0.49], [0.53, 0.47], [0.55, 0.42], [0.54, 0.47], [0.54, 0.47]]}" + } + ], + "images": "62473_exterior_image_1_left#0", + "gt": "[[206, 53], [191, 60], [182, 56], [182, 63], [162, 73], [163, 89], [171, 85], [175, 76], [173, 85], [173, 85]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_819112070044#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'putting a portion of the Lego blocks on the Lego baseplate', and now you are performing subtask 'insert the purple Lego brick into the Lego base, pick up the yellow Lego brick from the table, transfer the yellow Lego brick from the table to the Lego base, insert the yellow Lego brick into the Lego base.'\nYou start from [0.66, 0.54]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.66, 0.54], [0.66, 0.44], [0.65, 0.46], [0.63, 0.45], [0.63, 0.48], [0.64, 0.44], [0.65, 0.45], [0.63, 0.44], [0.63, 0.55], [0.62, 0.57]]}" + } + ], + "images": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_819112070044#245", + "gt": "[[423, 196], [424, 158], [414, 165], [402, 161], [401, 173], [409, 160], [413, 162], [406, 157], [402, 197], [396, 205]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0066_user_0014_scene_0007_cfg_0007_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'pick up the building blocks on the table, transfer the building blocks from the desktop to the building board, place the building blocks on the building board.'\nWith start location [0.53, 0.32], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.32], [0.64, 0.54], [0.66, 0.63], [0.66, 0.72], [0.64, 0.75], [0.65, 0.72], [0.6, 0.57], [0.56, 0.54], [0.55, 0.59], [0.55, 0.59]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0066_user_0014_scene_0007_cfg_0007_104122064161#0.jpg", + "gt": "[[337, 115], [407, 196], [424, 228], [422, 258], [411, 271], [415, 259], [384, 205], [361, 194], [354, 212], [354, 212]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0001_user_0020_scene_0001_cfg_0005_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the green button on the yellow base', and now you are performing subtask 'press the green button on the yellow base.'\nYou begin at [0.47, 0.4]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.4], [0.44, 0.43], [0.42, 0.49], [0.36, 0.56], [0.36, 0.59], [0.36, 0.6], [0.37, 0.62], [0.38, 0.61], [0.38, 0.63], [0.38, 0.54]]}" + } + ], + "images": "RH20T_cfg5_task_0001_user_0020_scene_0001_cfg_0005_037522062165#0", + "gt": "[[298, 144], [283, 156], [268, 178], [229, 203], [233, 211], [229, 217], [236, 222], [242, 219], [240, 226], [246, 194]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "63266_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the sauce bottle from the sink and put it on the stove', and now you are performing subtask 'pick up the sauce bottle from the sink and put it on the stove.'\nFrom start point [0.99, 0.11], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.99, 0.11], [0.95, 0.12], [0.78, 0.28], [0.72, 0.39], [0.7, 0.47], [0.7, 0.32], [0.7, 0.43], [0.67, 0.48], [0.69, 0.42], [0.77, 0.27]]}" + } + ], + "images": "63266_exterior_image_2_left#0", + "gt": "[[317, 19], [304, 22], [249, 50], [229, 70], [225, 84], [224, 57], [225, 77], [215, 86], [220, 75], [248, 49]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "69741_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the grey salt cellar then set it down again', and now you are performing subtask 'pick up the grey salt cellar then set it down again.'\nThe trajectory starts at [0.93, 0.14]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.93, 0.14], [0.73, 0.3], [0.65, 0.27], [0.56, 0.38], [0.49, 0.38], [0.51, 0.47], [0.48, 0.59], [0.5, 0.44], [0.47, 0.54], [0.49, 0.52]]}" + } + ], + "images": "69741_exterior_image_2_left#0", + "gt": "[[298, 25], [233, 54], [207, 48], [178, 68], [157, 68], [163, 84], [153, 106], [161, 80], [149, 97], [158, 93]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "59898_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the spoon from the bowl and put it in the box', and now you are performing subtask 'pick up the spoon from the bowl and put it in the box.'\nFrom start point [0.57, 0.45], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.45], [0.53, 0.45], [0.48, 0.52], [0.48, 0.6], [0.51, 0.66], [0.52, 0.5], [0.55, 0.51], [0.56, 0.59], [0.56, 0.52], [0.53, 0.5]]}" + } + ], + "images": "59898_exterior_image_1_left#0", + "gt": "[[183.0, 81.5], [169.0, 80.5], [154.0, 94.5], [154.0, 108.5], [162.0, 119.5], [165.0, 90.5], [175.0, 92.5], [180.0, 105.5], [178.0, 93.5], [171.0, 89.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "23666_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the red can in the cabinet', and now you are performing subtask 'put the red can in the cabinet.'\nYou start from [0.65, 0.54]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.54], [0.61, 0.66], [0.57, 0.71], [0.59, 0.6], [0.59, 0.32], [0.57, 0.0], [0.56, 0.08], [0.57, 0.07], [0.61, 0.13], [0.62, 0.39]]}" + } + ], + "images": "23666_exterior_image_2_left#0", + "gt": "[[207, 97], [195, 119], [182, 127], [188, 108], [188, 58], [181, 0], [180, 15], [184, 13], [195, 24], [199, 70]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "57351_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the top drawer of the white file cabinet', and now you are performing subtask 'close the top drawer of the white file cabinet.'\nStarting at [0.47, 0.8], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.8], [0.48, 0.66], [0.47, 0.57], [0.45, 0.51], [0.42, 0.46], [0.42, 0.44], [0.43, 0.41], [0.43, 0.4], [0.43, 0.38], [0.45, 0.38]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/57351_exterior_image_2_left#0.jpg", + "gt": "[[152, 144], [153, 119], [150, 102], [144, 92], [135, 83], [135, 79], [139, 74], [137, 72], [137, 69], [145, 69]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "72446_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press down the leftmost switch of the power strip', and now you are performing subtask 'press down the leftmost switch of the power strip.'\nThe trajectory starts at [0.78, 0.0]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.78, 0.0], [0.75, 0.13], [0.75, 0.32], [0.73, 0.32], [0.75, 0.38], [0.73, 0.43], [0.76, 0.43], [0.76, 0.46], [0.73, 0.44], [0.71, 0.25]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/72446_exterior_image_1_left#0.jpg", + "gt": "[[250, 0], [241, 24], [239, 58], [234, 58], [239, 68], [234, 78], [243, 78], [242, 82], [234, 80], [226, 45]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "66062_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn the two knobs on the lid clockwise', and now you are performing subtask 'twist another knob on the lid of the pot clockwise.'\nYou begin at [0.31, 0.68]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.31, 0.68], [0.4, 0.81], [0.44, 0.84], [0.45, 0.83], [0.43, 0.89], [0.43, 0.81], [0.46, 0.66], [0.52, 0.61], [0.63, 0.76], [0.59, 0.82]]}" + } + ], + "images": "66062_exterior_image_1_left#0", + "gt": "[[100, 123], [127, 146], [141, 152], [145, 149], [138, 161], [137, 145], [148, 118], [166, 109], [201, 137], [190, 147]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "80402_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the two bottles and put them on the counter', and now you are performing subtask 'place the second bottle on the counter.'\nWith start location [0.39, 0.65], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.65], [0.23, 0.53], [0.21, 0.48], [0.22, 0.43], [0.23, 0.47], [0.22, 0.42], [0.26, 0.39], [0.33, 0.44], [0.38, 0.51], [0.42, 0.61]]}" + } + ], + "images": "80402_exterior_image_1_left#231", + "gt": "[[126, 117], [75, 95], [68, 87], [70, 78], [75, 85], [71, 75], [83, 71], [106, 80], [123, 92], [135, 109]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "54044_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up and put down the green block, then put the salt cellar into the pot and the green block into the sink', and now you are performing subtask 'place the salt cellar inside the pot.'\nYour starting position is [0.44, 0.34]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.34], [0.39, 0.29], [0.41, 0.27], [0.44, 0.31], [0.46, 0.26], [0.51, 0.26], [0.54, 0.34], [0.5, 0.26], [0.61, 0.22], [0.56, 0.35]]}" + } + ], + "images": "54044_exterior_image_2_left#275", + "gt": "[[142, 62], [125, 53], [130, 48], [141, 55], [148, 46], [164, 46], [173, 61], [161, 47], [195, 39], [179, 63]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0066_user_0014_scene_0007_cfg_0007_104122064161#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'place the building blocks on the pile of building blocks on the building board, pick up the building blocks on the table, transfer the building block from the building board to the stack of building blocks, place the building blocks on the pile of building blocks on the building board.'\nYou begin at [0.51, 0.49]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.49], [0.55, 0.47], [0.58, 0.49], [0.61, 0.46], [0.67, 0.49], [0.68, 0.54], [0.63, 0.44], [0.57, 0.47], [0.56, 0.49], [0.57, 0.49]]}" + } + ], + "images": "RH20T_cfg7_task_0066_user_0014_scene_0007_cfg_0007_104122064161#449", + "gt": "[[329, 178], [352, 168], [373, 176], [392, 164], [428, 178], [432, 194], [406, 159], [365, 169], [358, 177], [366, 176]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0101_user_0007_scene_0010_cfg_0005_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'clean the table with a cloth', and now you are performing subtask 'clean the table with a cloth.'\nBegin at [0.48, 0.14] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.14], [0.5, 0.26], [0.5, 0.47], [0.52, 0.56], [0.53, 0.48], [0.57, 0.45], [0.52, 0.6], [0.59, 0.41], [0.51, 0.26], [0.47, 0.26]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0101_user_0007_scene_0010_cfg_0005_104122061850#0.jpg", + "gt": "[[305, 50], [318, 93], [318, 171], [331, 203], [342, 173], [368, 163], [335, 216], [375, 149], [326, 94], [298, 95]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_038522062547#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the first button on the desk, press the second button on the desk.'\nYou start from [0.66, 0.78]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.66, 0.78], [0.68, 0.78], [0.71, 0.55], [0.71, 0.41], [0.65, 0.43], [0.66, 0.39], [0.64, 0.41], [0.64, 0.39], [0.65, 0.41], [0.66, 0.4]]}" + } + ], + "images": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_038522062547#133", + "gt": "[[425, 282], [435, 282], [455, 198], [454, 147], [419, 156], [420, 142], [410, 146], [409, 140], [415, 147], [425, 144]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0003_user_0014_scene_0006_cfg_0005_104122061850#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the green button on the gray and white base, press the red button on the black and orange base.'\nFrom start point [0.54, 0.36], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.36], [0.55, 0.32], [0.58, 0.28], [0.58, 0.28], [0.6, 0.32], [0.6, 0.32], [0.6, 0.33], [0.6, 0.33], [0.57, 0.28], [0.55, 0.27]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0003_user_0014_scene_0006_cfg_0005_104122061850#390.jpg", + "gt": "[[344, 129], [352, 114], [369, 101], [374, 100], [385, 116], [383, 114], [383, 119], [387, 120], [363, 99], [350, 96]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "88471_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the grey towel from the right to the left', and now you are performing subtask 'fold the grey towel from the right to the left.'\nStart at [0.93, 0.29]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.93, 0.29], [0.86, 0.25], [0.73, 0.51], [0.74, 0.28], [0.68, 0.3], [0.67, 0.36], [0.71, 0.45], [0.64, 0.56], [0.65, 0.36], [0.65, 0.36]]}" + } + ], + "images": "88471_exterior_image_1_left#0", + "gt": "[[299, 53], [275, 45], [235, 92], [238, 50], [219, 54], [213, 64], [227, 81], [206, 101], [207, 65], [207, 65]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "17001_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fix the lid on the pot', and now you are performing subtask 'fix the lid on the pot.'\nYour starting position is [0.56, 0.42]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.56, 0.42], [0.53, 0.37], [0.42, 0.38], [0.39, 0.47], [0.38, 0.44], [0.4, 0.41], [0.4, 0.49], [0.41, 0.34], [0.5, 0.5], [0.5, 0.5]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/17001_exterior_image_2_left#0.jpg", + "gt": "[[180, 75], [169, 66], [135, 69], [126, 85], [121, 79], [127, 73], [127, 89], [130, 61], [160, 90], [160, 90]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "51601_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the pen into the bowl, then take it out and put it back on the table', and now you are performing subtask 'put the pen back on the table.'\nWith start location [0.55, 0.18], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.18], [0.57, 0.24], [0.58, 0.24], [0.58, 0.28], [0.58, 0.22], [0.57, 0.17], [0.56, 0.2], [0.56, 0.34], [0.58, 0.41], [0.61, 0.45]]}" + } + ], + "images": "51601_exterior_image_2_left#107", + "gt": "[[177, 33], [182, 43], [187, 44], [185, 51], [187, 40], [184, 31], [179, 36], [179, 62], [186, 74], [195, 81]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_035622060973#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the fourth block from the surface, transfer the fourth block from the desktop to the second position above the third block, place the fourth block on top of the third block.'\nStarting at [0.29, 0.51], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.29, 0.51], [0.29, 0.57], [0.31, 0.65], [0.3, 0.58], [0.35, 0.48], [0.35, 0.37], [0.34, 0.33], [0.33, 0.34], [0.34, 0.3], [0.34, 0.3]]}" + } + ], + "images": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_035622060973#584", + "gt": "[[186, 184], [185, 205], [199, 235], [191, 210], [225, 172], [223, 133], [218, 120], [211, 123], [215, 107], [215, 107]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_750612070853#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn them on', and now you are performing subtask 'press the fourth button on the right side of the power strip.'\nWith start location [0.4, 0.37], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.37], [0.41, 0.38], [0.41, 0.39], [0.41, 0.33], [0.4, 0.32], [0.41, 0.4], [0.41, 0.44], [0.43, 0.46], [0.43, 0.46], [0.43, 0.46]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_750612070853#277.jpg", + "gt": "[[257, 134], [263, 135], [264, 139], [262, 120], [258, 114], [263, 144], [262, 159], [273, 167], [275, 164], [275, 167]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "5673_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the cup and pour its contents into the white bowl', and now you are performing subtask 'pick up the cup and pour its contents into the white bowl.'\nThe trajectory starts at [0.62, 0.56]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.56], [0.54, 0.57], [0.52, 0.59], [0.52, 0.63], [0.52, 0.58], [0.47, 0.65], [0.46, 0.64], [0.47, 0.61], [0.48, 0.59], [0.48, 0.61]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/5673_exterior_image_1_left#0.jpg", + "gt": "[[198, 100], [173, 103], [166, 106], [165, 113], [167, 104], [151, 117], [148, 116], [152, 109], [154, 107], [155, 109]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "44324_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unstack the cans and put them beside each other', and now you are performing subtask 'place the second can next to the other cans.'\nYour initial coordinate is [0.5, 0.51]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.51], [0.51, 0.53], [0.48, 0.52], [0.46, 0.59], [0.43, 0.58], [0.43, 0.61], [0.45, 0.6], [0.45, 0.63], [0.45, 0.57], [0.45, 0.6]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/44324_exterior_image_2_left#207.jpg", + "gt": "[[160, 92], [163, 95], [153, 94], [147, 106], [136, 104], [138, 109], [145, 108], [143, 113], [144, 102], [143, 108]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "20007_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer on the right, take out three mugs and place them on the countertop', and now you are performing subtask 'pick up the first cup from the right drawer.'\nBegin at [0.75, 0.58] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.75, 0.58], [0.62, 0.62], [0.64, 0.73], [0.73, 0.79], [0.75, 0.8], [0.72, 0.68], [0.68, 0.64], [0.71, 0.71], [0.68, 0.67], [0.67, 0.79]]}" + } + ], + "images": "20007_exterior_image_1_left#0", + "gt": "[[239, 105], [198, 112], [204, 132], [232, 143], [241, 144], [230, 123], [219, 115], [226, 127], [216, 120], [214, 142]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "10380_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the marker from the mug, place it on the table then put it back in the mug', and now you are performing subtask 'place the marker back in the cup.'\nYou start from [0.88, 0.58]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.88, 0.58], [0.88, 0.6], [0.88, 0.59], [0.88, 0.56], [0.77, 0.41], [0.73, 0.42], [0.76, 0.51], [0.76, 0.44], [0.79, 0.44], [0.81, 0.48]]}" + } + ], + "images": "10380_exterior_image_1_left#225", + "gt": "[[283.0, 104.0], [280.0, 108.0], [283.0, 106.0], [280.0, 101.0], [246.0, 74.0], [233.0, 76.0], [242.0, 92.0], [243.0, 80.0], [253.0, 80.0], [259.0, 87.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0015_user_0014_scene_0005_cfg_0006_104122060811#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'play the drum', and now you are performing subtask 'play the drum.'\nYou start from [0.6, 0.35]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.35], [0.59, 0.3], [0.58, 0.32], [0.58, 0.35], [0.57, 0.33], [0.58, 0.34], [0.56, 0.35], [0.55, 0.38], [0.54, 0.38], [0.56, 0.37]]}" + } + ], + "images": "RH20T_cfg6_task_0015_user_0014_scene_0005_cfg_0006_104122060811#0", + "gt": "[[385, 127], [379, 107], [371, 117], [371, 127], [368, 119], [371, 123], [360, 126], [351, 136], [347, 136], [357, 132]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "42064_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push the faucet to the left and then lift the handle', and now you are performing subtask 'push the handle of the faucet upwards.'\nFrom start point [0.59, 0.46], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.46], [0.55, 0.46], [0.52, 0.43], [0.52, 0.41], [0.52, 0.37], [0.58, 0.33], [0.64, 0.26], [0.71, 0.22], [0.75, 0.14], [0.73, 0.16]]}" + } + ], + "images": "42064_exterior_image_1_left#174", + "gt": "[[189, 83], [177, 82], [167, 77], [165, 74], [165, 66], [185, 60], [205, 47], [226, 39], [241, 25], [235, 28]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "7985_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the towel', and now you are performing subtask 'fold the towel.'\nThe trajectory starts at [0.58, 0.42]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.42], [0.55, 0.43], [0.52, 0.49], [0.53, 0.49], [0.51, 0.52], [0.51, 0.57], [0.51, 0.55], [0.57, 0.48], [0.65, 0.5], [0.65, 0.5]]}" + } + ], + "images": "7985_exterior_image_1_left#192", + "gt": "[[186, 76], [177, 77], [167, 88], [171, 88], [162, 93], [162, 103], [162, 99], [182, 86], [209, 90], [209, 90]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0130_user_0020_scene_0004_cfg_0006_104122061018#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb', and now you are performing subtask 'pick up the bulb on the table, transfer the bulb from the table to the socket, insert the bulb into the socket, press the switch on the socket to turn on the bulb.'\nYour starting position is [0.57, 0.08]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.08], [0.58, 0.16], [0.62, 0.28], [0.46, 0.5], [0.43, 0.62], [0.42, 0.57], [0.49, 0.34], [0.48, 0.28], [0.46, 0.35], [0.46, 0.24]]}" + } + ], + "images": "RH20T_cfg6_task_0130_user_0020_scene_0004_cfg_0006_104122061018#0", + "gt": "[[367, 28], [371, 58], [400, 100], [297, 179], [272, 224], [269, 207], [311, 122], [305, 101], [293, 126], [296, 88]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "14075_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the object into the pot, then open the drawer and take the object from the pot and put it into the drawer', and now you are performing subtask 'pull the handle of the drawer to open the drawer.'\nBegin at [0.72, 0.46] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.73, 0.46], [0.72, 0.66], [0.52, 0.82], [0.51, 0.86], [0.57, 0.83], [0.5, 0.86], [0.56, 0.97], [0.57, 0.88], [0.63, 0.85], [0.63, 0.85]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/14075_exterior_image_1_left#0.jpg", + "gt": "[[232.0, 82.0], [230.0, 119.0], [167.0, 147.0], [164.0, 154.0], [183.0, 150.0], [161.0, 154.0], [178.0, 174.0], [184.0, 159.0], [201.0, 153.0], [201.0, 153.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0123_user_0010_scene_0004_cfg_0005_037522062165#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the knife on the knife rack', and now you are performing subtask 'place the knife on the knife rack, pick up the knife on the knife rack, put the knife on the knife rack.'\nFrom start point [0.34, 0.57], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.34, 0.57], [0.36, 0.58], [0.37, 0.65], [0.36, 0.67], [0.38, 0.69], [0.3, 0.66], [0.38, 0.65], [0.44, 0.59], [0.46, 0.49], [0.46, 0.49]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0123_user_0010_scene_0004_cfg_0005_037522062165#458.jpg", + "gt": "[[218, 204], [232, 210], [236, 235], [229, 242], [241, 249], [192, 239], [241, 233], [281, 214], [294, 176], [294, 176]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "18441_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'wipe the countertop with a towel', and now you are performing subtask 'wipe the countertop with a towel.'\nFrom start point [0.83, 0.0], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.83, 0.0], [0.78, 0.06], [0.8, 0.11], [0.76, 0.16], [0.74, 0.23], [0.68, 0.16], [0.73, 0.23], [0.69, 0.21], [0.71, 0.22], [0.7, 0.05]]}" + } + ], + "images": "18441_exterior_image_1_left#0", + "gt": "[[265, 0], [251, 11], [257, 20], [243, 28], [238, 41], [216, 29], [234, 42], [221, 38], [226, 39], [224, 9]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "1784_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the bowl in front of the table', and now you are performing subtask 'place the bowl in front of the table.'\nBegin at [0.38, 0.12] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.12], [0.44, 0.18], [0.46, 0.27], [0.46, 0.39], [0.47, 0.5], [0.5, 0.52], [0.52, 0.6], [0.54, 0.62], [0.52, 0.59], [0.52, 0.34]]}" + } + ], + "images": "1784_exterior_image_1_left#0", + "gt": "[[123.0, 22.0], [140.0, 33.0], [146.0, 48.0], [147.0, 71.0], [151.0, 90.0], [161.0, 94.0], [166.0, 108.0], [174.0, 112.0], [168.0, 107.0], [165.0, 62.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "32876_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pour the contents from the left bowl into the right bowl', and now you are performing subtask 'place the left bowl on the table.'\nStart at [0.57, 0.79]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.79], [0.57, 0.74], [0.6, 0.7], [0.61, 0.56], [0.64, 0.54], [0.64, 0.48], [0.69, 0.4], [0.61, 0.63], [0.62, 0.66], [0.62, 0.54]]}" + } + ], + "images": "32876_exterior_image_1_left#294", + "gt": "[[182, 142], [181, 133], [191, 126], [194, 101], [205, 98], [206, 86], [221, 72], [195, 113], [199, 119], [198, 98]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0062_user_0014_scene_0003_cfg_0006_104122060811#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the toilet paper on its holder', and now you are performing subtask 'put the toilet paper on its holder.'\nYou start from [0.6, 0.36]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.36], [0.6, 0.36], [0.62, 0.3], [0.63, 0.31], [0.63, 0.33], [0.63, 0.32], [0.6, 0.35], [0.57, 0.45], [0.57, 0.42], [0.6, 0.36]]}" + } + ], + "images": "RH20T_cfg6_task_0062_user_0014_scene_0003_cfg_0006_104122060811#0", + "gt": "[[387, 129], [383, 130], [394, 109], [403, 113], [404, 120], [402, 115], [386, 125], [365, 161], [368, 152], [387, 129]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0016_user_0014_scene_0008_cfg_0001_750612070851#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'hit the pool ball', and now you are performing subtask 'hit the pool ball.'\nWith start location [0.5, 0.43], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.43], [0.5, 0.43], [0.48, 0.64], [0.5, 0.79], [0.49, 0.81], [0.49, 0.76], [0.48, 0.79], [0.49, 0.76], [0.53, 0.76], [0.53, 0.74]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0016_user_0014_scene_0008_cfg_0001_750612070851#0.jpg", + "gt": "[[317, 154], [321, 156], [308, 230], [320, 283], [316, 292], [311, 275], [310, 283], [312, 273], [339, 274], [341, 267]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "73834_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the cloth', and now you are performing subtask 'move the cloth.'\nYour initial coordinate is [0.46, 0.01]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.01], [0.47, 0.11], [0.49, 0.19], [0.53, 0.28], [0.54, 0.38], [0.57, 0.47], [0.7, 0.62], [0.71, 0.62], [0.7, 0.53], [0.72, 0.54]]}" + } + ], + "images": "73834_exterior_image_1_left#0", + "gt": "[[146, 1], [150, 20], [157, 34], [170, 50], [172, 68], [182, 84], [224, 112], [226, 112], [224, 96], [229, 98]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "50112_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the cup slightly down on the table', and now you are performing subtask 'move the cup slightly down on the table.'\nFrom start point [0.68, 0.01], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.01], [0.62, 0.09], [0.62, 0.1], [0.64, 0.09], [0.58, 0.19], [0.5, 0.39], [0.49, 0.49], [0.5, 0.47], [0.53, 0.49], [0.52, 0.37]]}" + } + ], + "images": "50112_exterior_image_1_left#0", + "gt": "[[217.0, 1.5], [200.0, 15.5], [198.0, 18.5], [204.0, 15.5], [186.0, 33.5], [161.0, 69.5], [157.0, 87.5], [161.0, 84.5], [169.0, 88.5], [168.0, 66.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0094_user_0007_scene_0009_cfg_0004_104422070011#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'push the box to open the box.'\nFrom start point [0.52, 0.28], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.28], [0.52, 0.31], [0.51, 0.28], [0.52, 0.31], [0.58, 0.27], [0.61, 0.28], [0.6, 0.37], [0.6, 0.34], [0.6, 0.33], [0.56, 0.34]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0094_user_0007_scene_0009_cfg_0004_104422070011#0.jpg", + "gt": "[[332, 101], [336, 112], [325, 100], [332, 111], [374, 98], [391, 101], [386, 134], [387, 123], [383, 119], [359, 123]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "58558_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the clothes', and now you are performing subtask 'fold the left sleeve of the garment towards the middle.'\nYour initial coordinate is [0.49, 0.32]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.32], [0.49, 0.41], [0.43, 0.56], [0.44, 0.63], [0.47, 0.61], [0.45, 0.65], [0.53, 0.53], [0.67, 0.52], [0.73, 0.59], [0.69, 0.55]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/58558_exterior_image_2_left#0.jpg", + "gt": "[[156, 57], [156, 74], [137, 100], [140, 113], [149, 110], [144, 117], [169, 96], [215, 94], [233, 106], [220, 99]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "24423_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker in the mug on the table', and now you are performing subtask 'put the marker in the mug on the table.'\nStarting at [0.97, 0.07], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.97, 0.07], [0.77, 0.2], [0.79, 0.21], [0.66, 0.36], [0.67, 0.43], [0.63, 0.46], [0.67, 0.27], [0.63, 0.4], [0.63, 0.59], [0.62, 0.58]]}" + } + ], + "images": "24423_exterior_image_2_left#0", + "gt": "[[311, 13], [247, 36], [253, 38], [211, 64], [213, 77], [202, 83], [215, 48], [201, 72], [202, 107], [198, 105]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "38198_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the clothing from the backrest of the chair and place them on the table', and now you are performing subtask 'place the clothing on the tabletop.'\nYour initial coordinate is [0.6, 0.53]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.52], [0.56, 0.44], [0.52, 0.43], [0.52, 0.6], [0.47, 0.75], [0.51, 0.49], [0.57, 0.35], [0.6, 0.4], [0.6, 0.56], [0.56, 0.42]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/38198_exterior_image_1_left#253.jpg", + "gt": "[[192.5, 94.5], [179.5, 78.5], [165.5, 77.5], [165.5, 108.5], [151.5, 135.5], [162.5, 88.5], [182.5, 62.5], [191.5, 72.5], [190.5, 101.5], [178.5, 75.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0065_user_0010_scene_0003_cfg_0004_104122062823#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with small Lego blocks for the nth time with n blocks', and now you are performing subtask 'place the small Lego block on the gray Lego base plate, pick up the red small Lego block from the surface, transfer the red small Lego block from the surface to the gray Lego base plate, place the red small Lego block on the gray Lego base plate.'\nFrom start point [0.42, 0.31], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.31], [0.41, 0.27], [0.42, 0.31], [0.45, 0.15], [0.43, 0.2], [0.41, 0.14], [0.4, 0.3], [0.38, 0.29], [0.39, 0.18], [0.45, 0.12]]}" + } + ], + "images": "RH20T_cfg4_task_0065_user_0010_scene_0003_cfg_0004_104122062823#444", + "gt": "[[267, 112], [264, 97], [270, 110], [286, 54], [278, 71], [265, 51], [253, 109], [246, 103], [251, 64], [290, 44]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0104_user_0007_scene_0007_cfg_0005_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'drag the plate back after holding it down', and now you are performing subtask 'drag the plate back after holding it down.'\nYour initial coordinate is [0.45, 0.42]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.42], [0.38, 0.48], [0.38, 0.52], [0.31, 0.64], [0.3, 0.69], [0.31, 0.72], [0.33, 0.73], [0.31, 0.68], [0.36, 0.78], [0.31, 0.64]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0104_user_0007_scene_0007_cfg_0005_037522062165#0.jpg", + "gt": "[[289, 152], [244, 174], [242, 189], [198, 229], [191, 247], [196, 259], [210, 261], [200, 246], [228, 280], [201, 230]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "53874_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the bin', and now you are performing subtask 'close the bin.'\nStart at [0.47, 0.39]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.39], [0.47, 0.42], [0.51, 0.48], [0.52, 0.46], [0.52, 0.41], [0.54, 0.54], [0.52, 0.64], [0.52, 0.66], [0.53, 0.6], [0.52, 0.57]]}" + } + ], + "images": "53874_exterior_image_2_left#0", + "gt": "[[152, 70], [150, 75], [162, 86], [167, 82], [166, 74], [172, 98], [168, 115], [165, 118], [170, 108], [168, 102]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0003_user_0010_scene_0010_cfg_0004_038522062288#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the three buttons in sequence', and now you are performing subtask 'press the green button on the right, press the red button in the middle.'\nYour initial coordinate is [0.36, 0.33]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.33], [0.29, 0.43], [0.28, 0.41], [0.32, 0.41], [0.3, 0.39], [0.18, 0.34], [0.18, 0.4], [0.17, 0.37], [0.18, 0.36], [0.25, 0.38]]}" + } + ], + "images": "RH20T_cfg4_task_0003_user_0010_scene_0010_cfg_0004_038522062288#0", + "gt": "[[231, 120], [188, 154], [181, 147], [202, 148], [195, 142], [113, 124], [116, 144], [111, 133], [116, 128], [157, 136]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "31401_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the purple packet out of the sink and place it on the countertop', and now you are performing subtask 'take the purple packet out of the sink and place it on the countertop.'\nFrom start point [0.72, 0.33], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.72, 0.33], [0.68, 0.78], [0.7, 0.82], [0.69, 0.83], [0.68, 0.78], [0.69, 0.74], [0.71, 0.74], [0.62, 0.88], [0.62, 0.79], [0.64, 0.78]]}" + } + ], + "images": "31401_exterior_image_2_left#0", + "gt": "[[230, 60], [216, 140], [224, 147], [221, 150], [216, 140], [221, 133], [226, 134], [200, 158], [200, 143], [205, 140]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "68303_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the carrot plush toy backwards, move the blue towel backwards', and now you are performing subtask 'pick up the blue object from the front side of the table.'\nYour starting position is [0.35, 0.03]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.03], [0.37, 0.28], [0.37, 0.36], [0.36, 0.47], [0.34, 0.54], [0.34, 0.6], [0.35, 0.61], [0.34, 0.57], [0.35, 0.56], [0.35, 0.56]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/68303_exterior_image_1_left#0.jpg", + "gt": "[[112.5, 5.5], [119.5, 50.5], [119.5, 64.5], [114.5, 85.5], [107.5, 97.5], [109.5, 108.5], [111.5, 109.5], [109.5, 103.5], [110.5, 100.5], [110.5, 100.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0034_user_0014_scene_0001_cfg_0007_104122064161#TrajectoryPrediction_Task#10", + "task": "TrajectoryPrediction_Task", + "p_index": 10, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'pick up the square, place the square on another square, press the square.'\nBegin at [0.6, 0.6] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.6], [0.67, 0.72], [0.66, 0.76], [0.64, 0.66], [0.6, 0.49], [0.57, 0.47], [0.57, 0.45], [0.59, 0.46], [0.6, 0.38], [0.58, 0.23]]}" + } + ], + "images": "RH20T_cfg7_task_0034_user_0014_scene_0001_cfg_0007_104122064161#536", + "gt": "[[385, 217], [427, 258], [424, 272], [407, 239], [387, 178], [365, 171], [368, 163], [375, 166], [385, 136], [370, 83]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "64004_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the left cabinet door', and now you are performing subtask 'open the left cabinet door.'\nStarting at [0.57, 0.31], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.31], [0.59, 0.32], [0.57, 0.31], [0.58, 0.34], [0.64, 0.33], [0.79, 0.42], [0.87, 0.48], [0.72, 0.44], [0.65, 0.36], [0.65, 0.36]]}" + } + ], + "images": "64004_exterior_image_2_left#0", + "gt": "[[181, 56], [189, 58], [183, 56], [187, 61], [205, 60], [253, 75], [277, 87], [231, 79], [208, 65], [208, 65]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "82756_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the white cup from the countertop and move it to the right', and now you are performing subtask 'pick up the white cup from the countertop and move it to the right.'\nYour initial coordinate is [0.17, 0.02]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.17, 0.02], [0.3, 0.24], [0.29, 0.39], [0.36, 0.56], [0.37, 0.53], [0.43, 0.52], [0.48, 0.61], [0.44, 0.5], [0.46, 0.52], [0.45, 0.51]]}" + } + ], + "images": "82756_exterior_image_1_left#0", + "gt": "[[55.5, 4.0], [94.5, 43.0], [93.5, 70.0], [115.5, 100.0], [118.5, 96.0], [137.5, 94.0], [152.5, 110.0], [141.5, 90.0], [148.5, 94.0], [143.5, 91.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0002_user_0014_scene_0006_cfg_0006_104122061602#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull out a napkin', and now you are performing subtask 'pick up a napkin from the surface.'\nYou begin at [0.75, 0.43]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.75, 0.43], [0.74, 0.44], [0.74, 0.45], [0.74, 0.45], [0.74, 0.44], [0.74, 0.43], [0.74, 0.43], [0.75, 0.43], [0.75, 0.44], [0.75, 0.44]]}" + } + ], + "images": "RH20T_cfg6_task_0002_user_0014_scene_0006_cfg_0006_104122061602#168", + "gt": "[[477, 155], [475, 158], [475, 162], [473, 162], [476, 160], [475, 153], [474, 155], [479, 155], [478, 160], [478, 160]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "90036_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the white cloth from the glass panel onto the table', and now you are performing subtask 'place the white cloth from the glass panel onto the table.'\nStart at [0.34, 0.17]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.34, 0.17], [0.33, 0.27], [0.35, 0.31], [0.3, 0.36], [0.31, 0.43], [0.3, 0.5], [0.31, 0.51], [0.32, 0.49], [0.32, 0.49], [0.31, 0.5]]}" + } + ], + "images": "90036_exterior_image_2_left#0", + "gt": "[[109, 30], [106, 48], [112, 56], [96, 65], [98, 78], [95, 90], [99, 91], [101, 88], [103, 89], [100, 90]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "91638_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the shirt', and now you are performing subtask 'fold the shirt in half on the table.'\nYou begin at [0.69, 0.11]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.11], [0.8, 0.43], [0.79, 0.49], [0.72, 0.46], [0.62, 0.34], [0.57, 0.23], [0.58, 0.14], [0.64, 0.12], [0.68, 0.07], [0.68, 0.07]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/91638_exterior_image_1_left#265.jpg", + "gt": "[[222, 19], [256, 77], [254, 89], [230, 83], [200, 61], [184, 42], [186, 26], [205, 22], [218, 12], [218, 12]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "53930_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn the faucet to the right', and now you are performing subtask 'turn the faucet to the right.'\nThe trajectory starts at [0.34, 0.01]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.34, 0.01], [0.44, 0.29], [0.46, 0.38], [0.46, 0.46], [0.44, 0.5], [0.47, 0.47], [0.52, 0.48], [0.55, 0.46], [0.52, 0.41], [0.52, 0.41]]}" + } + ], + "images": "53930_exterior_image_1_left#0", + "gt": "[[110, 2], [140, 53], [146, 68], [148, 83], [140, 90], [151, 85], [167, 86], [176, 82], [167, 73], [167, 73]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "72158_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the yellow cylinder inside the orange pot', and now you are performing subtask 'put the yellow cylinder inside the orange pot.'\nStarting at [0.56, 0.01], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.56, 0.01], [0.6, 0.14], [0.62, 0.24], [0.62, 0.36], [0.64, 0.46], [0.66, 0.49], [0.67, 0.4], [0.67, 0.26], [0.64, 0.39], [0.64, 0.28]]}" + } + ], + "images": "72158_exterior_image_1_left#0", + "gt": "[[179, 2], [193, 26], [197, 44], [198, 65], [204, 82], [210, 88], [214, 72], [213, 47], [205, 70], [206, 50]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "8914_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the yellow thing in the orange cup then pour its contents on the plate', and now you are performing subtask 'place the yellow object in the orange cup.'\nYou begin at [0.48, 0.34]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.34], [0.73, 0.4], [0.71, 0.41], [0.86, 0.55], [0.76, 0.72], [0.83, 0.8], [0.79, 0.83], [0.75, 0.71], [0.73, 0.68], [0.74, 0.81]]}" + } + ], + "images": "8914_exterior_image_2_left#0", + "gt": "[[155, 61], [234, 72], [227, 74], [275, 99], [244, 129], [267, 144], [254, 150], [241, 127], [233, 122], [238, 145]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0009_user_0014_scene_0007_cfg_0001_038522062547#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take out a Hanoi block and throw it aside', and now you are performing subtask 'pick up the ring from the wooden rack, transfer the ring from the wooden rack to the tabletop, place the ring on the tabletop.'\nYou start from [0.62, 0.2]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.2], [0.62, 0.17], [0.62, 0.17], [0.66, 0.26], [0.67, 0.31], [0.67, 0.36], [0.66, 0.38], [0.66, 0.38], [0.66, 0.37], [0.66, 0.37]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0009_user_0014_scene_0007_cfg_0001_038522062547#334.jpg", + "gt": "[[399, 73], [397, 61], [399, 60], [420, 94], [427, 113], [426, 128], [423, 137], [421, 137], [423, 134], [422, 132]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "5143_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the object from the box and put it on the table', and now you are performing subtask 'place the packaging on the desktop.'\nYour initial coordinate is [0.48, 0.58]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.58], [0.56, 0.78], [0.53, 0.8], [0.53, 0.83], [0.5, 0.73], [0.43, 0.55], [0.46, 0.62], [0.52, 0.7], [0.56, 0.69], [0.54, 0.7]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/5143_exterior_image_1_left#198.jpg", + "gt": "[[153, 105], [180, 140], [171, 144], [170, 150], [161, 132], [136, 99], [147, 112], [168, 126], [180, 124], [173, 126]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0101_user_0007_scene_0002_cfg_0002_105422061350#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'cleaning the table with a cloth', and now you are performing subtask 'cleaning the table with a cloth.'\nWith start location [0.5, 0.3], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.3], [0.44, 0.23], [0.44, 0.24], [0.41, 0.24], [0.39, 0.32], [0.38, 0.45], [0.43, 0.36], [0.46, 0.34], [0.47, 0.34], [0.47, 0.34]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg2_task_0101_user_0007_scene_0002_cfg_0002_105422061350#0.jpg", + "gt": "[[319, 108], [284, 82], [283, 87], [260, 86], [247, 115], [241, 162], [276, 130], [293, 121], [298, 124], [298, 124]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0037_user_0007_scene_0006_cfg_0002_036422060215#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the second block on the table, transfer the second block from the table to the top of the third block, place the second block on the third block.'\nYou start from [0.5, 0.68]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.68], [0.5, 0.69], [0.49, 0.69], [0.5, 0.65], [0.5, 0.64], [0.5, 0.64], [0.5, 0.66], [0.49, 0.67], [0.49, 0.67], [0.49, 0.68]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg2_task_0037_user_0007_scene_0006_cfg_0002_036422060215#379.jpg", + "gt": "[[317, 244], [317, 250], [316, 247], [322, 233], [321, 230], [318, 232], [318, 237], [311, 240], [313, 241], [312, 243]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0070_user_0010_scene_0008_cfg_0003_104122062823#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a sliding window', and now you are performing subtask 'push the handle of the sliding window to outside the window frame, push the handle of the sliding window to outside the window frame.'\nFrom start point [0.39, 0.29], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.29], [0.45, 0.32], [0.44, 0.11], [0.41, 0.04], [0.39, 0.05], [0.39, 0.17], [0.36, 0.03], [0.47, 0.03], [0.47, 0.08], [0.47, 0.08]]}" + } + ], + "images": "RH20T_cfg3_task_0070_user_0010_scene_0008_cfg_0003_104122062823#192", + "gt": "[[250, 103], [291, 116], [279, 38], [265, 16], [251, 18], [247, 63], [231, 11], [298, 11], [300, 28], [300, 28]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0004_user_0006_scene_0003_cfg_0002_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the small block on the left and move it to the right', and now you are performing subtask 'pick up the small block on the left and move it to the right.'\nYou start from [0.55, 0.71]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.71], [0.63, 0.61], [0.66, 0.61], [0.69, 0.71], [0.66, 0.81], [0.68, 0.75], [0.64, 0.76], [0.61, 0.82], [0.64, 0.76], [0.63, 0.74]]}" + } + ], + "images": "RH20T_cfg2_task_0004_user_0006_scene_0003_cfg_0002_036422060215#0", + "gt": "[[349, 254], [402, 219], [424, 221], [444, 256], [420, 291], [435, 269], [407, 274], [391, 297], [409, 272], [403, 267]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "11875_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the fourth switch from the left', and now you are performing subtask 'turn on the fourth switch from the left.'\nStart at [0.75, 0.16]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.75, 0.16], [0.66, 0.24], [0.59, 0.33], [0.57, 0.36], [0.56, 0.38], [0.56, 0.43], [0.54, 0.46], [0.53, 0.46], [0.52, 0.48], [0.52, 0.51]]}" + } + ], + "images": "11875_exterior_image_1_left#0", + "gt": "[[240, 29], [212, 43], [189, 60], [181, 64], [178, 68], [178, 77], [174, 82], [169, 83], [165, 86], [166, 92]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0002_user_0014_scene_0006_cfg_0006_104122061018#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull out a napkin', and now you are performing subtask 'pull out a napkin.'\nStart at [0.57, 0.08]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.08], [0.55, 0.11], [0.49, 0.16], [0.45, 0.23], [0.45, 0.3], [0.43, 0.32], [0.43, 0.37], [0.47, 0.13], [0.44, 0.23], [0.44, 0.23]]}" + } + ], + "images": "RH20T_cfg6_task_0002_user_0014_scene_0006_cfg_0006_104122061018#0", + "gt": "[[367, 28], [355, 39], [313, 57], [287, 82], [286, 108], [275, 116], [274, 133], [301, 47], [283, 82], [283, 82]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0037_user_0020_scene_0006_cfg_0007_037522061512#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'transfer the fourth block from the table to above the third block, pick up the fifth block from the table, transfer the fifth block from the table to above the fourth block.'\nYour initial coordinate is [0.32, 0.27]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.27], [0.29, 0.39], [0.29, 0.48], [0.3, 0.56], [0.28, 0.36], [0.28, 0.3], [0.31, 0.2], [0.32, 0.16], [0.35, 0.14], [0.36, 0.17]]}" + } + ], + "images": "RH20T_cfg7_task_0037_user_0020_scene_0006_cfg_0007_037522061512#450", + "gt": "[[206, 96], [183, 140], [183, 174], [191, 200], [178, 129], [179, 108], [201, 73], [203, 58], [226, 50], [231, 60]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "42632_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the lid', and now you are performing subtask 'press the lid.'\nStart at [0.54, 0.63]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.63], [0.52, 0.67], [0.52, 0.68], [0.52, 0.68], [0.53, 0.71], [0.52, 0.66], [0.55, 0.54], [0.56, 0.54], [0.54, 0.56], [0.55, 0.54]]}" + } + ], + "images": "42632_exterior_image_1_left#156", + "gt": "[[173, 113], [165, 120], [165, 123], [167, 123], [169, 128], [166, 119], [176, 98], [178, 97], [174, 100], [175, 98]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "25550_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the black spoon from the black cutlery holder', and now you are performing subtask 'remove the black spoon from the black cutlery holder.'\nStart at [0.34, 0.38]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.34, 0.38], [0.46, 0.3], [0.52, 0.38], [0.62, 0.35], [0.71, 0.54], [0.64, 0.36], [0.54, 0.44], [0.54, 0.59], [0.61, 0.73], [0.52, 0.4]]}" + } + ], + "images": "25550_exterior_image_1_left#0", + "gt": "[[109.5, 68.5], [148.5, 54.5], [167.5, 68.5], [197.5, 63.5], [226.5, 96.5], [205.5, 64.5], [171.5, 79.5], [172.5, 105.5], [194.5, 130.5], [166.5, 71.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0022_user_0014_scene_0007_cfg_0006_037522061512#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'wave the flag', and now you are performing subtask 'pick up the flag on the table.'\nYour starting position is [0.4, 0.23]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.23], [0.36, 0.3], [0.38, 0.44], [0.37, 0.46], [0.36, 0.46], [0.35, 0.52], [0.34, 0.53], [0.34, 0.51], [0.33, 0.52], [0.34, 0.51]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0022_user_0014_scene_0007_cfg_0006_037522061512#0.jpg", + "gt": "[[257, 83], [229, 108], [240, 157], [235, 167], [233, 166], [223, 188], [217, 191], [216, 182], [214, 188], [215, 184]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "63647_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer and then put the three plush toys inside', and now you are performing subtask 'transfer a plush toy from the top of the table to the inside of the drawer.'\nYour initial coordinate is [0.79, 0.5]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.79, 0.5], [0.81, 0.62], [0.79, 0.73], [0.72, 0.81], [0.7, 0.89], [0.71, 0.91], [0.75, 0.9], [0.76, 0.88], [0.78, 0.88], [0.75, 0.86]]}" + } + ], + "images": "63647_exterior_image_1_left#0", + "gt": "[[251.5, 89.5], [259.5, 111.5], [251.5, 130.5], [230.5, 146.5], [224.5, 159.5], [227.5, 163.5], [239.5, 161.5], [242.5, 158.5], [250.5, 158.5], [241.5, 155.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_035622060973#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn them on', and now you are performing subtask 'press the second button on the right side of the power strip.'\nBegin at [0.35, 0.23] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.23], [0.36, 0.24], [0.37, 0.24], [0.37, 0.25], [0.36, 0.25], [0.36, 0.24], [0.36, 0.23], [0.36, 0.23], [0.35, 0.25], [0.35, 0.22]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_035622060973#136.jpg", + "gt": "[[227, 84], [229, 86], [235, 86], [235, 91], [233, 91], [231, 87], [233, 82], [228, 82], [227, 89], [225, 79]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0076_user_0010_scene_0001_cfg_0001_038522063145#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stacking blocks', and now you are performing subtask 'stacking blocks.'\nBegin at [0.69, 0.28] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.28], [0.77, 0.38], [0.78, 0.33], [0.79, 0.39], [0.76, 0.38], [0.79, 0.44], [0.75, 0.36], [0.69, 0.42], [0.69, 0.36], [0.73, 0.38]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0076_user_0010_scene_0001_cfg_0001_038522063145#0.jpg", + "gt": "[[440, 99], [494, 135], [500, 118], [503, 140], [488, 136], [504, 160], [477, 130], [441, 150], [444, 130], [465, 138]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0035_user_0017_scene_0005_cfg_0004_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up a small block', and now you are performing subtask 'pick up a small block.'\nYour initial coordinate is [0.45, 0.0]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.0], [0.41, 0.0], [0.44, 0.0], [0.43, 0.0], [0.4, 0.21], [0.42, 0.32], [0.42, 0.33], [0.41, 0.32], [0.45, 0.05], [0.46, 0.05]]}" + } + ], + "images": "RH20T_cfg4_task_0035_user_0017_scene_0005_cfg_0004_104122062823#0", + "gt": "[[289, 0], [262, 0], [282, 0], [276, 0], [253, 76], [270, 114], [270, 120], [263, 117], [291, 17], [296, 18]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0061_user_0014_scene_0008_cfg_0001_039422060546#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'place the small cube on the lid of the gift box, pick up the button from inside the gift box, transfer the button from the gift box to the desktop, place the button on the desktop.'\nYour initial coordinate is [0.4, 0.47]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.47], [0.42, 0.34], [0.46, 0.3], [0.46, 0.33], [0.43, 0.25], [0.41, 0.4], [0.36, 0.38], [0.27, 0.49], [0.29, 0.5], [0.3, 0.43]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0061_user_0014_scene_0008_cfg_0001_039422060546#88.jpg", + "gt": "[[258, 171], [271, 122], [293, 109], [296, 119], [277, 91], [262, 144], [232, 138], [170, 178], [188, 180], [193, 153]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "19949_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the orange marker in the clear cup', and now you are performing subtask 'put the orange marker in the clear cup.'\nStarting at [0.7, 0.0], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.0], [0.56, 0.37], [0.59, 0.24], [0.52, 0.28], [0.55, 0.2], [0.41, 0.22], [0.36, 0.29], [0.42, 0.26], [0.46, 0.11], [0.53, 0.09]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/19949_exterior_image_2_left#0.jpg", + "gt": "[[225, 0], [179, 67], [190, 43], [167, 50], [175, 36], [130, 39], [114, 52], [134, 47], [148, 20], [170, 17]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "47994_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the bottle from the left side of the tray and put it on the table on the left side of the tray', and now you are performing subtask 'take the bottle from the left side of the tray and put it on the table on the left side of the tray.'\nYour starting position is [0.68, 0.53]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.53], [0.65, 0.56], [0.68, 0.56], [0.63, 0.62], [0.62, 0.71], [0.65, 0.76], [0.71, 0.69], [0.74, 0.73], [0.7, 0.76], [0.69, 0.73]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/47994_exterior_image_1_left#0.jpg", + "gt": "[[217.5, 96.0], [206.5, 101.0], [217.5, 101.0], [202.5, 111.0], [197.5, 127.0], [207.5, 136.0], [227.5, 124.0], [235.5, 131.0], [224.5, 137.0], [221.5, 131.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0037_user_0014_scene_0010_cfg_0001_750612070851#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the block from the table onto the two blocks standing upright, pick up the block from the table, transfer the block from the table to the three blocks standing upright, place the block from the table onto the three blocks standing upright.'\nYou begin at [0.68, 0.84]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.84], [0.71, 0.83], [0.73, 0.83], [0.73, 0.83], [0.75, 0.84], [0.74, 0.81], [0.71, 0.79], [0.67, 0.78], [0.65, 0.79], [0.65, 0.82]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0037_user_0014_scene_0010_cfg_0001_750612070851#236.jpg", + "gt": "[[433, 304], [454, 298], [470, 298], [465, 300], [477, 303], [471, 291], [456, 284], [428, 280], [418, 285], [416, 294]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0073_user_0010_scene_0010_cfg_0003_038522062288#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the things in the drawer', and now you are performing subtask 'pick up the small square from the table, transfer the small square from the table to the open drawer, place the small square in the drawer.'\nBegin at [0.37, 0.35] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.35], [0.27, 0.43], [0.16, 0.46], [0.2, 0.5], [0.21, 0.38], [0.28, 0.26], [0.3, 0.27], [0.29, 0.34], [0.33, 0.31], [0.35, 0.7]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg3_task_0073_user_0010_scene_0010_cfg_0003_038522062288#0.jpg", + "gt": "[[236, 126], [173, 155], [105, 165], [125, 181], [136, 135], [178, 92], [192, 96], [183, 122], [213, 110], [221, 251]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "79987_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the marker out of the cup and put it on the table', and now you are performing subtask 'take the marker out of the cup and put it on the table.'\nYou begin at [0.67, 0.01]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.67, 0.01], [0.64, 0.22], [0.57, 0.31], [0.54, 0.54], [0.51, 0.55], [0.52, 0.47], [0.6, 0.18], [0.5, 0.46], [0.52, 0.5], [0.58, 0.47]]}" + } + ], + "images": "79987_exterior_image_1_left#0", + "gt": "[[214.0, 2.0], [205.0, 40.0], [182.0, 56.0], [172.0, 98.0], [163.0, 99.0], [165.0, 84.0], [192.0, 32.0], [161.0, 83.0], [168.0, 90.0], [187.0, 85.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "72353_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the towel', and now you are performing subtask 'unfold the towel on the countertop through the operation.'\nBegin at [0.67, 0.73] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.67, 0.73], [0.64, 0.74], [0.63, 0.9], [0.61, 0.84], [0.61, 0.82], [0.62, 0.82], [0.64, 0.7], [0.68, 0.62], [0.72, 0.61], [0.66, 0.65]]}" + } + ], + "images": "72353_exterior_image_1_left#185", + "gt": "[[213, 131], [205, 133], [202, 162], [195, 152], [196, 147], [200, 147], [206, 126], [217, 112], [231, 109], [210, 117]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0051_user_0016_scene_0003_cfg_0002_105422061350#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the dish off the dish rack', and now you are performing subtask 'take the dish off the dish rack.'\nYou begin at [0.48, 0.27]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.27], [0.48, 0.32], [0.44, 0.33], [0.44, 0.39], [0.39, 0.43], [0.35, 0.52], [0.32, 0.5], [0.31, 0.56], [0.35, 0.42], [0.38, 0.37]]}" + } + ], + "images": "RH20T_cfg2_task_0051_user_0016_scene_0003_cfg_0002_105422061350#0", + "gt": "[[305, 98], [307, 117], [282, 119], [280, 140], [252, 154], [224, 186], [204, 181], [198, 202], [221, 152], [241, 133]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0037_user_0014_scene_0010_cfg_0001_750612070851#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the block on another block, pick up the block from the table, transfer the block from the table to the two blocks standing upright, place the block from the table onto the two blocks standing upright.'\nThe trajectory starts at [0.64, 0.88]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.64, 0.88], [0.65, 0.83], [0.64, 0.77], [0.65, 0.76], [0.65, 0.78], [0.65, 0.74], [0.64, 0.81], [0.65, 0.86], [0.66, 0.84], [0.68, 0.84]]}" + } + ], + "images": "RH20T_cfg1_task_0037_user_0014_scene_0010_cfg_0001_750612070851#158", + "gt": "[[410, 317], [413, 300], [408, 278], [418, 275], [416, 281], [417, 267], [411, 290], [413, 310], [424, 301], [433, 304]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0129_user_0020_scene_0009_cfg_0006_104122061330#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'plug in the bulb holder with a bulb to the socket', and now you are performing subtask 'plug in the bulb holder with a bulb to the socket.'\nYour initial coordinate is [0.42, 0.27]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.27], [0.45, 0.26], [0.5, 0.18], [0.6, 0.21], [0.6, 0.31], [0.6, 0.19], [0.56, 0.38], [0.58, 0.51], [0.58, 0.36], [0.49, 0.32]]}" + } + ], + "images": "RH20T_cfg6_task_0129_user_0020_scene_0009_cfg_0006_104122061330#0", + "gt": "[[266, 97], [287, 95], [321, 65], [387, 77], [383, 111], [383, 69], [359, 135], [371, 184], [370, 130], [313, 114]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "59590_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pour the contents from the orange cup onto the gray plate', and now you are performing subtask 'pour the contents from the orange cup onto the gray plate.'\nStarting at [0.54, 0.28], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.28], [0.28, 0.5], [0.16, 0.31], [0.32, 0.32], [0.26, 0.36], [0.31, 0.28], [0.35, 0.48], [0.32, 0.43], [0.28, 0.5], [0.28, 0.5]]}" + } + ], + "images": "59590_exterior_image_1_left#0", + "gt": "[[173, 51], [91, 90], [50, 55], [102, 57], [83, 64], [99, 50], [113, 86], [101, 78], [89, 90], [89, 90]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "83961_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the pen in the mug', and now you are performing subtask 'place the pen in the mug.'\nBegin at [0.76, 0.02] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.76, 0.02], [0.69, 0.11], [0.73, 0.09], [0.68, 0.16], [0.57, 0.54], [0.59, 0.24], [0.63, 0.13], [0.64, 0.17], [0.55, 0.29], [0.53, 0.2]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/83961_exterior_image_1_left#0.jpg", + "gt": "[[242, 3], [221, 19], [234, 16], [218, 29], [182, 98], [189, 44], [202, 24], [206, 30], [175, 52], [170, 36]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "72392_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place three sweets in the cup', and now you are performing subtask 'transfer the second candy from the table to the cup.'\nYou start from [0.45, 0.46]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.46], [0.47, 0.54], [0.47, 0.56], [0.45, 0.57], [0.45, 0.51], [0.4, 0.43], [0.37, 0.41], [0.28, 0.44], [0.27, 0.47], [0.27, 0.51]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/72392_exterior_image_2_left#114.jpg", + "gt": "[[143, 83], [151, 97], [151, 101], [145, 103], [143, 92], [128, 78], [118, 74], [88, 79], [85, 84], [87, 91]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0046_user_0020_scene_0010_cfg_0006_104122061602#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the cup on the cup rack', and now you are performing subtask 'put the cup on the cup rack.'\nStarting at [0.7, 0.31], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.31], [0.74, 0.38], [0.78, 0.34], [0.77, 0.44], [0.77, 0.31], [0.73, 0.29], [0.72, 0.21], [0.69, 0.2], [0.7, 0.23], [0.7, 0.23]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0046_user_0020_scene_0010_cfg_0006_104122061602#0.jpg", + "gt": "[[446, 112], [474, 135], [498, 123], [490, 158], [491, 110], [469, 104], [463, 75], [442, 71], [449, 84], [449, 84]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0029_user_0016_scene_0009_cfg_0004_036422060909#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp the handle and open the drawer', and now you are performing subtask 'grasp the handle and open the drawer.'\nYou begin at [0.53, 0.22]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.22], [0.52, 0.19], [0.57, 0.16], [0.6, 0.18], [0.59, 0.2], [0.61, 0.19], [0.65, 0.24], [0.66, 0.28], [0.55, 0.33], [0.5, 0.23]]}" + } + ], + "images": "RH20T_cfg4_task_0029_user_0016_scene_0009_cfg_0004_036422060909#0", + "gt": "[[337, 78], [334, 67], [368, 56], [383, 66], [379, 71], [393, 70], [419, 88], [424, 102], [355, 118], [322, 84]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "29854_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the black drawer', and now you are performing subtask 'open the black drawer.'\nWith start location [0.43, 0.83], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.83], [0.51, 0.67], [0.6, 0.54], [0.61, 0.47], [0.64, 0.52], [0.63, 0.46], [0.66, 0.53], [0.56, 0.61], [0.56, 0.48], [0.54, 0.52]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/29854_exterior_image_2_left#0.jpg", + "gt": "[[138, 150], [162, 120], [191, 97], [195, 84], [205, 93], [201, 83], [210, 96], [178, 110], [180, 87], [172, 93]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0043_user_0014_scene_0008_cfg_0006_104122064161#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unscrew the lid from the jar', and now you are performing subtask 'twist the lid of the jar counterclockwise.'\nYou start from [0.51, 0.49]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.49], [0.5, 0.51], [0.5, 0.51], [0.5, 0.52], [0.5, 0.51], [0.5, 0.49], [0.5, 0.49], [0.5, 0.49], [0.5, 0.49], [0.5, 0.49]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0043_user_0014_scene_0008_cfg_0006_104122064161#296.jpg", + "gt": "[[324, 177], [323, 182], [318, 185], [319, 187], [317, 184], [319, 175], [319, 178], [319, 175], [321, 175], [320, 177]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "90989_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the green marker out of the white mug', and now you are performing subtask 'take the green marker out of the white mug.'\nYour starting position is [0.55, 0.49]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.49], [0.53, 0.64], [0.55, 0.61], [0.55, 0.67], [0.56, 0.5], [0.55, 0.47], [0.53, 0.49], [0.54, 0.53], [0.57, 0.47], [0.57, 0.47]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/90989_exterior_image_2_left#0.jpg", + "gt": "[[177, 89], [169, 115], [176, 110], [177, 120], [179, 90], [175, 85], [170, 89], [174, 95], [182, 85], [182, 85]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "60918_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pour the candy out of the clear cup and onto the table', and now you are performing subtask 'pour the candy out of the clear cup and onto the table.'\nYou begin at [0.48, 0.79]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.79], [0.54, 0.79], [0.56, 0.87], [0.58, 0.81], [0.61, 0.87], [0.68, 0.75], [0.76, 0.89], [0.7, 0.79], [0.63, 0.87], [0.61, 0.84]]}" + } + ], + "images": "60918_exterior_image_1_left#0", + "gt": "[[154, 142], [173, 142], [178, 157], [187, 146], [194, 157], [216, 135], [242, 160], [224, 143], [203, 157], [195, 151]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0018_user_0008_scene_0007_cfg_0001_750612070851#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'play jenga', and now you are performing subtask 'play jenga.'\nYou begin at [0.52, 0.43]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.43], [0.54, 0.52], [0.61, 0.67], [0.66, 0.71], [0.65, 0.69], [0.68, 0.72], [0.68, 0.7], [0.61, 0.71], [0.56, 0.61], [0.56, 0.53]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0018_user_0008_scene_0007_cfg_0001_750612070851#0.jpg", + "gt": "[[330, 155], [346, 187], [393, 240], [422, 256], [419, 248], [437, 259], [438, 253], [389, 254], [360, 220], [360, 191]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "38044_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the light red object on the right side of the table', and now you are performing subtask 'put the light red object on the right side of the table.'\nYour initial coordinate is [0.77, 0.24]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.77, 0.24], [0.68, 0.26], [0.71, 0.37], [0.67, 0.68], [0.73, 0.64], [0.75, 0.6], [0.73, 0.59], [0.73, 0.63], [0.84, 0.62], [0.79, 0.64]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/38044_exterior_image_1_left#0.jpg", + "gt": "[[247, 43], [218, 47], [227, 67], [214, 122], [234, 115], [240, 108], [233, 107], [234, 113], [268, 111], [254, 116]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_039422060546#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'fold the third lid on the box on the table.'\nBegin at [0.32, 0.53] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.53], [0.29, 0.49], [0.27, 0.37], [0.24, 0.32], [0.25, 0.34], [0.25, 0.37], [0.25, 0.34], [0.23, 0.32], [0.26, 0.43], [0.3, 0.52]]}" + } + ], + "images": "RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_039422060546#124", + "gt": "[[202, 191], [187, 175], [170, 134], [152, 115], [160, 122], [158, 134], [157, 122], [148, 117], [167, 153], [194, 188]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0037_user_0007_scene_0004_cfg_0005_105422061350#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the third block on top of the second block, pick up the fourth block from the surface, place the fourth block on top of the third block.'\nStarting at [0.38, 0.13], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.13], [0.38, 0.08], [0.35, 0.09], [0.26, 0.27], [0.23, 0.41], [0.26, 0.56], [0.28, 0.18], [0.35, 0.0], [0.38, 0.03], [0.38, 0.0]]}" + } + ], + "images": "RH20T_cfg5_task_0037_user_0007_scene_0004_cfg_0005_105422061350#927", + "gt": "[[246, 47], [243, 29], [225, 34], [165, 96], [150, 146], [167, 200], [181, 66], [221, 0], [246, 10], [245, 0]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0029_user_0016_scene_0009_cfg_0004_104422070011#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp the handle and open the drawer', and now you are performing subtask 'grasp the handle and open the drawer.'\nBegin at [0.52, 0.29] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.29], [0.53, 0.34], [0.51, 0.25], [0.46, 0.23], [0.46, 0.25], [0.42, 0.18], [0.38, 0.19], [0.37, 0.21], [0.38, 0.22], [0.56, 0.24]]}" + } + ], + "images": "RH20T_cfg4_task_0029_user_0016_scene_0009_cfg_0004_104422070011#0", + "gt": "[[334, 103], [339, 123], [326, 91], [292, 83], [293, 90], [269, 65], [241, 68], [234, 76], [240, 79], [358, 85]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "70471_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the white and black mugs from the drawer, push the drawer closed, close the open bottom cabinet', and now you are performing subtask 'push the drawer closed.'\nStarting at [0.54, 0.56], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.56], [0.56, 0.6], [0.58, 0.56], [0.61, 0.57], [0.7, 0.71], [0.71, 0.76], [0.72, 0.74], [0.65, 0.77], [0.6, 0.71], [0.56, 0.71]]}" + } + ], + "images": "70471_exterior_image_1_left#595", + "gt": "[[173, 101], [180, 108], [187, 100], [195, 102], [225, 127], [226, 136], [229, 134], [209, 139], [191, 128], [180, 128]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0013_user_0014_scene_0007_cfg_0006_104122061330#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the block on the scale', and now you are performing subtask 'place the block on the scale.'\nYou start from [0.42, 0.27]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.27], [0.39, 0.34], [0.4, 0.42], [0.4, 0.39], [0.42, 0.3], [0.46, 0.31], [0.45, 0.29], [0.36, 0.29], [0.35, 0.28], [0.36, 0.27]]}" + } + ], + "images": "RH20T_cfg6_task_0013_user_0014_scene_0007_cfg_0006_104122061330#0", + "gt": "[[266, 97], [251, 124], [255, 151], [254, 140], [270, 108], [293, 110], [285, 103], [230, 106], [224, 101], [230, 97]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0061_user_0007_scene_0010_cfg_0005_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'pick up the object in the container, transfer the object from the container to the table, place the object on the table.'\nBegin at [0.45, 0.3] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.3], [0.44, 0.3], [0.42, 0.36], [0.43, 0.46], [0.43, 0.42], [0.42, 0.41], [0.4, 0.42], [0.36, 0.46], [0.34, 0.5], [0.34, 0.57]]}" + } + ], + "images": "RH20T_cfg5_task_0061_user_0007_scene_0010_cfg_0005_036422060215#0", + "gt": "[[286, 108], [281, 108], [267, 130], [273, 164], [273, 152], [268, 148], [256, 150], [229, 165], [216, 180], [216, 205]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0057_user_0010_scene_0006_cfg_0003_104122062823#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn on the power strips', and now you are performing subtask 'press the button on the middle right power strip, press the button on the middle left power strip, press the button on the left power strip.'\nBegin at [0.21, 0.14] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.21, 0.14], [0.23, 0.15], [0.22, 0.17], [0.22, 0.17], [0.21, 0.15], [0.2, 0.14], [0.19, 0.13], [0.18, 0.13], [0.18, 0.14], [0.19, 0.14]]}" + } + ], + "images": "RH20T_cfg3_task_0057_user_0010_scene_0006_cfg_0003_104122062823#106", + "gt": "[[136, 52], [145, 55], [140, 60], [141, 63], [137, 54], [130, 49], [120, 46], [117, 47], [117, 50], [120, 52]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0002_user_0010_scene_0001_cfg_0005_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull out a napkin', and now you are performing subtask 'pull out a napkin.'\nYou begin at [0.53, 0.25]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.25], [0.53, 0.28], [0.54, 0.32], [0.54, 0.28], [0.51, 0.21], [0.5, 0.21], [0.5, 0.22], [0.48, 0.3], [0.45, 0.32], [0.44, 0.35]]}" + } + ], + "images": "RH20T_cfg5_task_0002_user_0010_scene_0001_cfg_0005_104122061850#0", + "gt": "[[336, 89], [337, 99], [348, 116], [346, 100], [329, 77], [322, 74], [323, 80], [309, 108], [285, 115], [283, 125]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0037_user_0014_scene_0009_cfg_0007_104122061018#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the third block from the desktop, transfer the block from the desktop to the first block, place the block on the first block.'\nYou begin at [0.6, 0.58]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.58], [0.6, 0.61], [0.59, 0.61], [0.6, 0.56], [0.6, 0.54], [0.59, 0.53], [0.55, 0.52], [0.54, 0.54], [0.55, 0.51], [0.57, 0.48]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0037_user_0014_scene_0009_cfg_0007_104122061018#175.jpg", + "gt": "[[382, 209], [381, 220], [379, 220], [385, 201], [383, 194], [376, 191], [353, 189], [344, 194], [354, 182], [368, 173]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "58283_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'no action', and now you are performing subtask 'no action.'\nYour initial coordinate is [0.23, 0.31]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.23, 0.31], [0.42, 0.36], [0.43, 0.52], [0.53, 0.58], [0.53, 0.66], [0.56, 0.71], [0.53, 0.7], [0.52, 0.76], [0.55, 0.78], [0.55, 0.73]]}" + } + ], + "images": "58283_exterior_image_2_left#0", + "gt": "[[75, 55], [133, 65], [136, 94], [171, 104], [171, 118], [180, 127], [169, 126], [168, 137], [175, 140], [176, 131]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "73732_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open and close the microwave door and then close the top fridge door', and now you are performing subtask 'push the microwave door to the closed position.'\nYour starting position is [0.59, 0.21]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.21], [0.62, 0.27], [0.69, 0.25], [0.71, 0.31], [0.76, 0.33], [0.73, 0.34], [0.81, 0.68], [0.59, 0.61], [0.46, 0.44], [0.69, 0.47]]}" + } + ], + "images": "73732_exterior_image_1_left#421", + "gt": "[[189.5, 38.0], [199.5, 48.0], [220.5, 45.0], [226.5, 56.0], [243.5, 60.0], [234.5, 62.0], [258.5, 122.0], [188.5, 109.0], [148.5, 79.0], [219.5, 84.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0096_user_0007_scene_0003_cfg_0005_104122063678#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer and place the cube on the table', and now you are performing subtask 'pull the handle of the drawer to pull open the drawer, pick up the block from the drawer, transfer the block from the drawer to the table, place the block on the table.'\nYour starting position is [0.58, 0.47]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.47], [0.58, 0.4], [0.52, 0.4], [0.48, 0.52], [0.5, 0.45], [0.55, 0.4], [0.63, 0.43], [0.62, 0.5], [0.59, 0.54], [0.61, 0.42]]}" + } + ], + "images": "RH20T_cfg5_task_0096_user_0007_scene_0003_cfg_0005_104122063678#372", + "gt": "[[372, 170], [369, 143], [330, 143], [308, 188], [319, 161], [350, 143], [401, 155], [394, 180], [377, 194], [391, 152]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "83136_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'folding the towel', and now you are performing subtask 'fold the towel.'\nWith start location [0.3, 0.51], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.3, 0.51], [0.51, 0.31], [0.52, 0.47], [0.45, 0.57], [0.62, 0.51], [0.51, 0.39], [0.43, 0.52], [0.52, 0.48], [0.49, 0.29], [0.41, 0.29]]}" + } + ], + "images": "83136_exterior_image_1_left#162", + "gt": "[[95, 91], [164, 55], [165, 85], [145, 103], [198, 92], [163, 71], [137, 93], [168, 87], [156, 53], [132, 52]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "52264_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pour the contents of the cup into the pot', and now you are performing subtask 'pour the contents of the cup into the pot.'\nFrom start point [0.7, 0.18], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.18], [0.6, 0.33], [0.51, 0.65], [0.5, 0.51], [0.62, 0.46], [0.65, 0.31], [0.58, 0.41], [0.51, 0.63], [0.55, 0.61], [0.55, 0.61]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/52264_exterior_image_2_left#0.jpg", + "gt": "[[225, 33], [193, 60], [162, 117], [161, 91], [198, 83], [209, 55], [185, 74], [163, 114], [175, 110], [175, 110]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "26333_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the button atop the lid then pick up the lid and put it on the table', and now you are performing subtask 'place the lid on the table.'\nYour starting position is [0.58, 0.25]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.25], [0.58, 0.17], [0.57, 0.31], [0.55, 0.32], [0.57, 0.31], [0.56, 0.21], [0.57, 0.14], [0.62, 0.13], [0.64, 0.41], [0.64, 0.23]]}" + } + ], + "images": "26333_exterior_image_1_left#147", + "gt": "[[186, 45], [185, 31], [184, 56], [177, 58], [184, 56], [178, 38], [182, 26], [200, 23], [206, 73], [204, 41]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "779_exterior_image_2_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer then take the plush toy from the drawer and put it in the box then close the drawer', and now you are performing subtask 'push the drawer handle to close the drawer.'\nThe trajectory starts at [0.57, 0.27]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.27], [0.43, 0.39], [0.42, 0.57], [0.46, 0.66], [0.57, 0.61], [0.6, 0.56], [0.57, 0.47], [0.43, 0.33], [0.43, 0.27], [0.45, 0.26]]}" + } + ], + "images": "779_exterior_image_2_left#415", + "gt": "[[181, 49], [137, 71], [133, 103], [148, 119], [183, 110], [193, 101], [184, 85], [139, 60], [137, 48], [145, 47]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0073_user_0014_scene_0003_cfg_0002_036422060215#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the block in the drawer, close the drawer', and now you are performing subtask 'place the square in the drawer, push the drawer to close it.'\nStarting at [0.54, 0.44], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.44], [0.48, 0.44], [0.49, 0.48], [0.52, 0.48], [0.53, 0.49], [0.59, 0.49], [0.55, 0.46], [0.53, 0.46], [0.5, 0.47], [0.5, 0.49]]}" + } + ], + "images": "RH20T_cfg2_task_0073_user_0014_scene_0003_cfg_0002_036422060215#178", + "gt": "[[344, 158], [310, 160], [314, 173], [335, 173], [339, 178], [375, 176], [355, 167], [339, 166], [322, 171], [322, 177]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "21996_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the bottle from the top part of the segment tray and put it on the transparent box', and now you are performing subtask 'take the bottle from the top part of the segment tray and put it on the transparent box.'\nBegin at [0.67, 0.31] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.67, 0.31], [0.59, 0.5], [0.57, 0.71], [0.6, 0.77], [0.58, 0.79], [0.57, 0.74], [0.56, 0.78], [0.59, 0.82], [0.58, 0.71], [0.58, 0.71]]}" + } + ], + "images": "21996_exterior_image_1_left#0", + "gt": "[[214.5, 56.0], [188.5, 90.0], [182.5, 127.0], [190.5, 138.0], [186.5, 142.0], [181.5, 134.0], [179.5, 141.0], [188.5, 148.0], [185.5, 127.0], [185.5, 127.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "17711_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the white paper in the black bin then put the orange object in the open drawer then close the drawer', and now you are performing subtask 'transfer the white paper from the table to the black bin.'\nYou start from [0.39, 0.26]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.26], [0.38, 0.26], [0.4, 0.3], [0.4, 0.31], [0.42, 0.3], [0.44, 0.32], [0.43, 0.29], [0.48, 0.2], [0.53, 0.14], [0.55, 0.13]]}" + } + ], + "images": "17711_exterior_image_1_left#0", + "gt": "[[123.5, 46.5], [121.5, 47.5], [128.5, 53.5], [127.5, 55.5], [133.5, 53.5], [139.5, 57.5], [138.5, 51.5], [152.5, 35.5], [168.5, 24.5], [176.5, 23.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122061602#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a sliding window', and now you are performing subtask 'slide the left window.'\nYour starting position is [0.7, 0.29]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.29], [0.7, 0.48], [0.71, 0.47], [0.72, 0.49], [0.71, 0.5], [0.72, 0.52], [0.72, 0.49], [0.72, 0.51], [0.74, 0.51], [0.73, 0.49]]}" + } + ], + "images": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122061602#0", + "gt": "[[445, 106], [451, 173], [455, 171], [460, 178], [455, 180], [458, 189], [462, 176], [460, 184], [475, 183], [468, 177]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "17981_exterior_image_1_left#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the orange cloth', and now you are performing subtask 'fold the cloth on the table for the last time.'\nFrom start point [0.54, 0.28], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.28], [0.52, 0.44], [0.53, 0.49], [0.58, 0.46], [0.53, 0.51], [0.59, 0.47], [0.52, 0.56], [0.54, 0.38], [0.62, 0.4], [0.57, 0.27]]}" + } + ], + "images": "17981_exterior_image_1_left#320", + "gt": "[[173, 51], [167, 80], [171, 89], [185, 83], [171, 92], [189, 84], [166, 101], [172, 68], [197, 72], [181, 49]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0015_user_0006_scene_0003_cfg_0002_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'play the drum', and now you are performing subtask 'play the drum.'\nThe trajectory starts at [0.45, 0.55]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.55], [0.48, 0.64], [0.48, 0.71], [0.45, 0.81], [0.43, 0.82], [0.47, 0.89], [0.43, 0.88], [0.44, 0.96], [0.38, 0.98], [0.34, 0.74]]}" + } + ], + "images": "RH20T_cfg2_task_0015_user_0006_scene_0003_cfg_0002_104122063678#0", + "gt": "[[289, 199], [308, 230], [308, 255], [288, 290], [272, 295], [301, 319], [273, 318], [280, 344], [246, 353], [220, 265]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "52032_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the clear lid from the top of the black object to the top of the cup and press on the lid', and now you are performing subtask 'move the clear lid from the top of the black object to the top of the cup and press on the lid.'\nWith start location [0.32, 0.43], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.43], [0.44, 0.51], [0.44, 0.64], [0.47, 0.58], [0.53, 0.63], [0.53, 0.58], [0.53, 0.65], [0.55, 0.62], [0.56, 0.67], [0.46, 0.56]]}" + } + ], + "images": "52032_exterior_image_2_left#0", + "gt": "[[102, 78], [140, 91], [140, 115], [150, 104], [170, 113], [170, 105], [169, 117], [175, 111], [178, 121], [148, 100]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "29724_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the salt cellar to the left of the table', and now you are performing subtask 'move the salt cellar to the left of the table.'\nYour starting position is [0.34, 0.21]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.34, 0.21], [0.38, 0.32], [0.4, 0.46], [0.44, 0.56], [0.43, 0.61], [0.44, 0.5], [0.46, 0.5], [0.46, 0.52], [0.47, 0.47], [0.48, 0.48]]}" + } + ], + "images": "29724_exterior_image_1_left#0", + "gt": "[[108.0, 38.0], [121.0, 57.0], [128.0, 83.0], [140.0, 101.0], [138.0, 110.0], [142.0, 90.0], [147.0, 90.0], [147.0, 94.0], [149.0, 85.0], [153.0, 86.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "70471_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the white and black mugs from the drawer, push the drawer closed, close the open bottom cabinet', and now you are performing subtask 'place the black cup on the countertop.'\nYour starting position is [0.46, 0.53]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.53], [0.42, 0.54], [0.39, 0.61], [0.43, 0.62], [0.44, 0.66], [0.42, 0.67], [0.43, 0.68], [0.4, 0.67], [0.43, 0.54], [0.49, 0.56]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/70471_exterior_image_2_left#283.jpg", + "gt": "[[147, 95], [135, 97], [125, 109], [139, 111], [140, 119], [134, 120], [139, 123], [128, 121], [138, 97], [158, 101]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "47132_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker on the table', and now you are performing subtask 'put the marker on the table.'\nYou begin at [0.5, 0.01]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.01], [0.52, 0.42], [0.55, 0.44], [0.55, 0.33], [0.58, 0.28], [0.67, 0.52], [0.67, 0.5], [0.62, 0.38], [0.63, 0.38], [0.63, 0.38]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/47132_exterior_image_1_left#0.jpg", + "gt": "[[160, 1], [167, 75], [176, 79], [177, 60], [186, 51], [213, 94], [214, 90], [197, 68], [201, 69], [201, 69]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "54874_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker in the cup', and now you are performing subtask 'put the marker in the cup.'\nYou begin at [0.49, 0.0]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.0], [0.47, 0.02], [0.47, 0.1], [0.43, 0.17], [0.46, 0.31], [0.42, 0.25], [0.47, 0.01], [0.52, 0.0], [0.54, 0.08], [0.52, 0.08]]}" + } + ], + "images": "54874_exterior_image_2_left#0", + "gt": "[[158, 0], [149, 4], [149, 18], [136, 31], [146, 55], [134, 45], [151, 1], [168, 0], [172, 15], [165, 14]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "51719_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the object from the bowl and put it on the table', and now you are performing subtask 'remove the object from the bowl and put it on the table.'\nThe trajectory starts at [0.38, 0.47]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.47], [0.46, 0.47], [0.53, 0.5], [0.56, 0.61], [0.62, 0.68], [0.6, 0.52], [0.57, 0.54], [0.55, 0.63], [0.55, 0.84], [0.54, 0.69]]}" + } + ], + "images": "51719_exterior_image_2_left#0", + "gt": "[[120, 85], [146, 84], [171, 90], [180, 109], [200, 122], [193, 93], [184, 97], [177, 114], [177, 151], [173, 125]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_038522062547#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the third block on top of the second block, pick up the fourth block from the surface, transfer the fourth block from the desktop to the second position above the third block, place the fourth block on top of the third block.'\nYour initial coordinate is [0.72, 0.27]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.72, 0.27], [0.75, 0.19], [0.74, 0.16], [0.79, 0.15], [0.79, 0.23], [0.75, 0.32], [0.79, 0.17], [0.73, 0.09], [0.73, 0.07], [0.73, 0.19]]}" + } + ], + "images": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_038522062547#528", + "gt": "[[461, 96], [482, 67], [476, 58], [508, 53], [506, 82], [482, 115], [504, 61], [466, 34], [469, 24], [466, 67]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "25286_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the boxes inside the case', and now you are performing subtask 'pick up the box on the table.'\nStart at [0.39, 0.4]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.4], [0.6, 0.56], [0.64, 0.7], [0.62, 0.77], [0.62, 0.58], [0.58, 0.56], [0.6, 0.57], [0.55, 0.64], [0.57, 0.58], [0.68, 0.79]]}" + } + ], + "images": "25286_exterior_image_1_left#0", + "gt": "[[124, 72], [193, 101], [206, 126], [199, 139], [199, 104], [186, 101], [193, 102], [175, 116], [184, 105], [217, 143]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "84537_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unwrap the charger cable', and now you are performing subtask 'shake the charger cable to unwrap it.'\nBegin at [0.78, 0.0] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.77, 0.0], [0.73, 0.22], [0.62, 0.19], [0.59, 0.38], [0.63, 0.41], [0.5, 0.32], [0.41, 0.35], [0.41, 0.39], [0.48, 0.36], [0.52, 0.43]]}" + } + ], + "images": "84537_exterior_image_1_left#0", + "gt": "[[248, 0], [232, 39], [200, 34], [189, 69], [201, 74], [161, 57], [130, 63], [131, 70], [153, 65], [166, 77]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0029_user_0014_scene_0005_cfg_0006_104122061330#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer', and now you are performing subtask 'open the drawer.'\nYou start from [0.42, 0.27]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.27], [0.36, 0.34], [0.35, 0.39], [0.35, 0.41], [0.36, 0.41], [0.4, 0.3], [0.47, 0.2], [0.46, 0.16], [0.43, 0.16], [0.41, 0.17]]}" + } + ], + "images": "RH20T_cfg6_task_0029_user_0014_scene_0005_cfg_0006_104122061330#0", + "gt": "[[266, 97], [229, 124], [225, 139], [227, 148], [231, 146], [254, 107], [301, 71], [293, 59], [276, 57], [260, 63]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_038522062547#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with small Lego blocks for the nth time with n blocks', and now you are performing subtask 'place the blue small Lego block on the white small Lego block, press the blue small Lego block, pick up the red small Lego block from the desktop, transfer the red small Lego block from the desktop to the blue small Lego block, place the red small Lego block on the blue small Lego block.'\nYour starting position is [0.65, 0.52]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.52], [0.64, 0.5], [0.65, 0.52], [0.66, 0.46], [0.69, 0.45], [0.7, 0.37], [0.71, 0.39], [0.66, 0.47], [0.65, 0.54], [0.65, 0.49]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_038522062547#1454.jpg", + "gt": "[[416, 186], [409, 181], [419, 187], [422, 167], [440, 162], [446, 133], [453, 141], [420, 169], [415, 193], [415, 178]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0066_user_0014_scene_0007_cfg_0007_104122061018#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'place the building blocks on the pile of building blocks on the building board, pick up the building blocks on the table, transfer the building blocks from the desktop to the pile of building blocks on the building board, place the building blocks on the pile of building blocks on the building board, pick up the building blocks on the table.'\nYour starting position is [0.42, 0.31]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.31], [0.35, 0.44], [0.35, 0.49], [0.36, 0.5], [0.34, 0.4], [0.38, 0.24], [0.43, 0.11], [0.48, 0.19], [0.46, 0.16], [0.45, 0.27]]}" + } + ], + "images": "RH20T_cfg7_task_0066_user_0014_scene_0007_cfg_0007_104122061018#671", + "gt": "[[267, 111], [223, 159], [223, 177], [232, 181], [220, 144], [243, 86], [277, 39], [306, 69], [295, 58], [285, 96]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0057_user_0016_scene_0001_cfg_0002_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn on the power strips', and now you are performing subtask 'press the button of the first row socket, press the button of the power strip.'\nYou start from [0.44, 0.49]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.49], [0.46, 0.48], [0.46, 0.46], [0.4, 0.43], [0.34, 0.47], [0.3, 0.46], [0.26, 0.5], [0.26, 0.55], [0.23, 0.52], [0.24, 0.56]]}" + } + ], + "images": "RH20T_cfg2_task_0057_user_0016_scene_0001_cfg_0002_037522062165#0", + "gt": "[[279, 175], [295, 172], [293, 164], [253, 155], [220, 168], [194, 165], [166, 180], [166, 199], [150, 188], [152, 200]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "26685_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push the can to the right', and now you are performing subtask 'push the can to the right.'\nFrom start point [0.32, 0.12], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.12], [0.38, 0.11], [0.47, 0.13], [0.56, 0.19], [0.57, 0.23], [0.57, 0.33], [0.59, 0.36], [0.62, 0.34], [0.67, 0.29], [0.65, 0.23]]}" + } + ], + "images": "26685_exterior_image_1_left#0", + "gt": "[[102.5, 21.0], [122.5, 20.0], [150.5, 23.0], [179.5, 35.0], [183.5, 41.0], [181.5, 59.0], [188.5, 64.0], [197.5, 62.0], [214.5, 52.0], [208.5, 42.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0041_user_0007_scene_0009_cfg_0002_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'clean the tabletop with a sponge', and now you are performing subtask 'clean the tabletop with a sponge.'\nYou start from [0.55, 0.71]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.71], [0.57, 0.73], [0.58, 0.63], [0.62, 0.59], [0.68, 0.67], [0.68, 0.7], [0.66, 0.68], [0.66, 0.71], [0.64, 0.69], [0.56, 0.74]]}" + } + ], + "images": "RH20T_cfg2_task_0041_user_0007_scene_0009_cfg_0002_036422060215#0", + "gt": "[[349, 255], [368, 263], [374, 226], [396, 214], [437, 241], [437, 253], [423, 246], [424, 257], [411, 250], [358, 266]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "77287_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the two brown bags on the table into the blue basket', and now you are performing subtask 'place the second brown bag in the blue basket.'\nStart at [0.63, 0.61]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.61], [0.62, 0.61], [0.62, 0.66], [0.62, 0.67], [0.63, 0.53], [0.64, 0.51], [0.64, 0.6], [0.67, 0.57], [0.68, 0.58], [0.69, 0.61]]}" + } + ], + "images": "77287_exterior_image_1_left#173", + "gt": "[[203, 109], [200, 110], [197, 118], [199, 121], [201, 96], [206, 91], [204, 108], [215, 102], [219, 104], [221, 109]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "22788_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull the second rod out', and now you are performing subtask 'pull the second rod out.'\nYou begin at [0.71, 0.42]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.71, 0.42], [0.65, 0.35], [0.66, 0.38], [0.65, 0.38], [0.65, 0.54], [0.62, 0.61], [0.66, 0.58], [0.72, 0.58], [0.73, 0.5], [0.72, 0.47]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/22788_exterior_image_2_left#0.jpg", + "gt": "[[227, 75], [209, 63], [212, 68], [207, 68], [207, 97], [200, 110], [212, 104], [229, 104], [235, 90], [230, 84]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "22632_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the towel then fold it again', and now you are performing subtask 'unfold the towel on the table.'\nYou start from [0.99, 0.3]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.99, 0.3], [0.88, 0.42], [0.96, 0.51], [0.87, 0.48], [0.77, 0.59], [0.77, 0.5], [0.73, 0.49], [0.72, 0.72], [0.62, 0.97], [0.65, 0.72]]}" + } + ], + "images": "22632_exterior_image_1_left#0", + "gt": "[[317, 54], [283, 76], [307, 92], [278, 87], [247, 106], [246, 90], [232, 89], [230, 130], [200, 174], [208, 129]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "29602_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the brush from the table and sweep the paper pieces', and now you are performing subtask 'pick up the brush from the table and sweep the paper pieces.'\nThe trajectory starts at [0.41, 0.56]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.56], [0.49, 0.62], [0.48, 0.89], [0.45, 0.76], [0.56, 0.8], [0.64, 0.71], [0.52, 0.72], [0.48, 0.87], [0.5, 0.77], [0.5, 0.77]]}" + } + ], + "images": "29602_exterior_image_1_left#0", + "gt": "[[132, 101], [157, 112], [154, 161], [145, 136], [179, 144], [204, 127], [165, 130], [155, 156], [160, 139], [160, 139]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0046_user_0020_scene_0010_cfg_0006_104122061330#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the cup on the cup rack', and now you are performing subtask 'put the cup on the cup rack.'\nWith start location [0.42, 0.29], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.29], [0.48, 0.4], [0.53, 0.39], [0.54, 0.5], [0.56, 0.3], [0.55, 0.2], [0.55, 0.19], [0.54, 0.23], [0.52, 0.26], [0.51, 0.23]]}" + } + ], + "images": "RH20T_cfg6_task_0046_user_0020_scene_0010_cfg_0006_104122061330#0", + "gt": "[[266, 103], [309, 144], [337, 141], [346, 181], [359, 108], [354, 73], [349, 67], [344, 84], [330, 92], [325, 83]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "70557_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put a corner of the green towel on the brown teddy bear toy', and now you are performing subtask 'put a corner of the green towel on the brown teddy bear toy.'\nYour initial coordinate is [0.5, 0.03]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.03], [0.59, 0.35], [0.59, 0.44], [0.62, 0.49], [0.59, 0.41], [0.56, 0.39], [0.52, 0.41], [0.49, 0.47], [0.48, 0.38], [0.51, 0.33]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/70557_exterior_image_1_left#0.jpg", + "gt": "[[159, 5], [188, 63], [188, 79], [198, 88], [188, 74], [180, 71], [166, 74], [156, 85], [155, 69], [162, 59]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "29279_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker in the mug', and now you are performing subtask 'put the marker in the mug.'\nYour starting position is [0.43, 0.49]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.49], [0.56, 0.56], [0.57, 0.69], [0.59, 0.51], [0.62, 0.57], [0.64, 0.57], [0.63, 0.64], [0.66, 0.58], [0.65, 0.54], [0.67, 0.54]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/29279_exterior_image_2_left#0.jpg", + "gt": "[[138, 88], [178, 101], [182, 124], [189, 92], [197, 103], [204, 102], [203, 116], [211, 104], [207, 98], [214, 97]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0065_user_0007_scene_0010_cfg_0002_105422061350#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with small Lego blocks for the nth time with n blocks', and now you are performing subtask 'pick up the white Lego block from the table, transfer the white Lego block from the table to the building area, place the white Lego block in the building area, press the white Lego block in the building area.'\nBegin at [0.32, 0.24] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.24], [0.4, 0.31], [0.43, 0.34], [0.43, 0.36], [0.44, 0.37], [0.43, 0.38], [0.44, 0.36], [0.44, 0.38], [0.44, 0.36], [0.34, 0.27]]}" + } + ], + "images": "RH20T_cfg2_task_0065_user_0007_scene_0010_cfg_0002_105422061350#739", + "gt": "[[202, 85], [259, 111], [276, 123], [274, 130], [283, 133], [278, 138], [282, 130], [280, 135], [283, 131], [218, 96]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0047_user_0014_scene_0002_cfg_0006_104122061018#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the cup off the cup rack', and now you are performing subtask 'take the cup off the cup rack.'\nYour initial coordinate is [0.57, 0.08]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.08], [0.58, 0.04], [0.56, 0.03], [0.56, 0.0], [0.37, 0.0], [0.43, 0.0], [0.45, 0.13], [0.44, 0.18], [0.45, 0.0], [0.45, 0.0]]}" + } + ], + "images": "RH20T_cfg6_task_0047_user_0014_scene_0002_cfg_0006_104122061018#0", + "gt": "[[367, 28], [373, 15], [361, 9], [361, 0], [238, 0], [273, 0], [287, 48], [282, 65], [291, 0], [291, 0]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "13360_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the lid on the black pot', and now you are performing subtask 'put the lid on the black pot.'\nThe trajectory starts at [0.36, 0.51]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.51], [0.3, 0.57], [0.31, 0.66], [0.25, 0.59], [0.27, 0.52], [0.31, 0.52], [0.37, 0.56], [0.39, 0.6], [0.38, 0.64], [0.37, 0.59]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/13360_exterior_image_1_left#0.jpg", + "gt": "[[114.0, 91.5], [95.0, 103.5], [99.0, 119.5], [80.0, 105.5], [87.0, 94.5], [99.0, 94.5], [118.0, 100.5], [124.0, 108.5], [120.0, 114.5], [118.0, 106.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122061018#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'place the block on the block board, pick up the block on the flat surface, transfer the block from the flat surface to the other blocks, place the block on the other blocks.'\nYou begin at [0.56, 0.42]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.56, 0.42], [0.53, 0.37], [0.47, 0.36], [0.41, 0.41], [0.37, 0.47], [0.41, 0.32], [0.47, 0.21], [0.47, 0.19], [0.5, 0.18], [0.5, 0.23]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122061018#438.jpg", + "gt": "[[356, 151], [338, 132], [304, 129], [263, 148], [239, 168], [264, 115], [304, 75], [303, 67], [321, 64], [319, 82]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "81948_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the white bottle to the right', and now you are performing subtask 'move the white bottle to the right.'\nStarting at [0.35, 0.08], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.08], [0.42, 0.12], [0.49, 0.27], [0.5, 0.38], [0.49, 0.29], [0.49, 0.41], [0.58, 0.62], [0.58, 0.79], [0.6, 0.77], [0.49, 0.44]]}" + } + ], + "images": "81948_exterior_image_1_left#0", + "gt": "[[113.0, 14.0], [134.0, 21.0], [157.0, 49.0], [161.0, 68.0], [156.0, 53.0], [156.0, 73.0], [186.0, 111.0], [185.0, 142.0], [191.0, 138.0], [156.0, 79.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg3_task_0003_user_0016_scene_0002_cfg_0003_104122062295#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the three buttons in sequence', and now you are performing subtask 'press the green button on the right.'\nWith start location [0.41, 0.24], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.24], [0.38, 0.23], [0.42, 0.26], [0.33, 0.16], [0.28, 0.17], [0.24, 0.23], [0.24, 0.26], [0.27, 0.25], [0.23, 0.28], [0.25, 0.24]]}" + } + ], + "images": "RH20T_cfg3_task_0003_user_0016_scene_0002_cfg_0003_104122062295#0", + "gt": "[[265, 86], [246, 84], [266, 94], [209, 59], [182, 60], [152, 84], [152, 92], [171, 89], [147, 101], [162, 86]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0071_user_0014_scene_0003_cfg_0002_105422061350#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close a sliding window', and now you are performing subtask 'close a sliding window.'\nYour initial coordinate is [0.48, 0.28]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.28], [0.5, 0.34], [0.46, 0.28], [0.45, 0.21], [0.43, 0.19], [0.44, 0.16], [0.38, 0.28], [0.39, 0.21], [0.41, 0.19], [0.47, 0.28]]}" + } + ], + "images": "RH20T_cfg2_task_0071_user_0014_scene_0003_cfg_0002_105422061350#0", + "gt": "[[306, 100], [322, 124], [293, 101], [288, 77], [278, 68], [283, 57], [243, 101], [251, 77], [265, 69], [299, 100]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0044_user_0020_scene_0001_cfg_0007_104122061018#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up a bag of things', and now you are performing subtask 'pick up a bag of things.'\nYou start from [0.57, 0.09]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.09], [0.54, 0.12], [0.51, 0.18], [0.5, 0.17], [0.5, 0.19], [0.46, 0.22], [0.46, 0.24], [0.46, 0.28], [0.51, 0.16], [0.53, 0.12]]}" + } + ], + "images": "RH20T_cfg7_task_0044_user_0020_scene_0001_cfg_0007_104122061018#0", + "gt": "[[363, 31], [344, 43], [327, 64], [321, 63], [319, 68], [296, 79], [292, 87], [294, 102], [326, 58], [342, 44]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "81483_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the chocolate bar and put it on the bottom shelf', and now you are performing subtask 'pick up the chocolate bar and put it on the bottom shelf.'\nBegin at [0.32, 0.48] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.48], [0.37, 0.69], [0.38, 0.82], [0.33, 0.66], [0.32, 0.47], [0.35, 0.27], [0.37, 0.25], [0.4, 0.29], [0.34, 0.27], [0.29, 0.51]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/81483_exterior_image_1_left#0.jpg", + "gt": "[[103, 87], [119, 124], [122, 148], [105, 119], [101, 85], [111, 48], [119, 45], [129, 52], [109, 49], [94, 92]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0039_user_0014_scene_0002_cfg_0006_104122063633#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pour the water from one cup into another empty cup', and now you are performing subtask 'pour the water from one cup into another empty cup.'\nYou start from [0.36, 0.33]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.33], [0.36, 0.41], [0.36, 0.34], [0.34, 0.32], [0.32, 0.56], [0.32, 0.32], [0.32, 0.29], [0.34, 0.49], [0.34, 0.41], [0.34, 0.51]]}" + } + ], + "images": "RH20T_cfg6_task_0039_user_0014_scene_0002_cfg_0006_104122063633#0", + "gt": "[[233, 118], [230, 148], [230, 124], [216, 117], [208, 200], [202, 115], [208, 103], [215, 176], [218, 148], [217, 185]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "44908_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the pot's lid to the table', and now you are performing subtask 'move the pot's lid to the table.'\nYour initial coordinate is [0.28, 0.02]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.28, 0.02], [0.43, 0.24], [0.52, 0.49], [0.51, 0.57], [0.47, 0.49], [0.47, 0.22], [0.46, 0.2], [0.43, 0.34], [0.43, 0.51], [0.41, 0.46]]}" + } + ], + "images": "44908_exterior_image_1_left#0", + "gt": "[[90, 4], [136, 44], [166, 89], [164, 103], [152, 88], [152, 40], [148, 36], [137, 62], [137, 91], [132, 82]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0017_user_0014_scene_0002_cfg_0006_104122063633#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the pen into the pen holder', and now you are performing subtask 'pick up the pen from the table, transfer the pen from the table to the position of the pen holder, place the pen in the pen holder.'\nBegin at [0.41, 0.56] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.56], [0.4, 0.56], [0.41, 0.61], [0.43, 0.51], [0.45, 0.59], [0.47, 0.57], [0.47, 0.66], [0.46, 0.6], [0.44, 0.56], [0.43, 0.57]]}" + } + ], + "images": "RH20T_cfg6_task_0017_user_0014_scene_0002_cfg_0006_104122063633#253", + "gt": "[[263, 202], [254, 201], [265, 218], [276, 182], [287, 214], [300, 207], [300, 237], [293, 215], [282, 202], [278, 207]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "19382_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn off the light and then turn it back on', and now you are performing subtask 'press the switch on the wall to turn on the light.'\nYour starting position is [0.3, 0.44]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.3, 0.44], [0.28, 0.43], [0.27, 0.41], [0.3, 0.39], [0.29, 0.39], [0.28, 0.43], [0.27, 0.43], [0.28, 0.39], [0.29, 0.39], [0.29, 0.39]]}" + } + ], + "images": "19382_exterior_image_1_left#142", + "gt": "[[96, 80], [88, 77], [86, 73], [96, 70], [92, 70], [90, 78], [87, 78], [88, 71], [94, 70], [94, 70]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "12219_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the marker from the table then put it into the red cup then take it out of the red cup and put it back on the table', and now you are performing subtask 'take the marker from the table then put it into the red cup then take it out of the red cup and put it back on the table.'\nYour starting position is [0.65, 0.21]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.21], [0.49, 0.39], [0.47, 0.64], [0.56, 0.47], [0.53, 0.42], [0.58, 0.44], [0.56, 0.54], [0.6, 0.48], [0.55, 0.48], [0.52, 0.59]]}" + } + ], + "images": "12219_exterior_image_2_left#0", + "gt": "[[209, 37], [157, 71], [151, 116], [179, 84], [171, 75], [186, 80], [178, 97], [191, 86], [175, 86], [165, 106]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0094_user_0007_scene_0006_cfg_0005_104422070044#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'press the part of the box.'\nYou begin at [0.66, 0.39]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.66, 0.39], [0.66, 0.39], [0.69, 0.42], [0.69, 0.45], [0.69, 0.44], [0.68, 0.43], [0.67, 0.42], [0.67, 0.41], [0.66, 0.4], [0.66, 0.41]]}" + } + ], + "images": "RH20T_cfg5_task_0094_user_0007_scene_0006_cfg_0005_104422070044#97", + "gt": "[[422, 142], [424, 141], [443, 151], [442, 161], [440, 157], [436, 153], [430, 150], [427, 147], [422, 145], [420, 149]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0017_user_0014_scene_0002_cfg_0006_104122061330#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the pen into the pen holder', and now you are performing subtask 'put the pen into the pen holder.'\nStarting at [0.42, 0.27], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.27], [0.44, 0.42], [0.4, 0.12], [0.43, 0.1], [0.43, 0.38], [0.41, 0.17], [0.42, 0.11], [0.4, 0.12], [0.42, 0.12], [0.42, 0.16]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0017_user_0014_scene_0002_cfg_0006_104122061330#0.jpg", + "gt": "[[266, 97], [280, 152], [259, 44], [275, 35], [275, 136], [260, 62], [267, 41], [254, 43], [266, 42], [269, 59]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0003_user_0010_scene_0001_cfg_0005_104122061850#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the red button in the middle, press the green button on the left.'\nYour initial coordinate is [0.53, 0.32]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.32], [0.54, 0.24], [0.58, 0.14], [0.59, 0.15], [0.6, 0.19], [0.59, 0.12], [0.52, 0.04], [0.52, 0.09], [0.49, 0.16], [0.46, 0.17]]}" + } + ], + "images": "RH20T_cfg5_task_0003_user_0010_scene_0001_cfg_0005_104122061850#299", + "gt": "[[341, 116], [346, 87], [370, 52], [379, 55], [381, 69], [376, 44], [335, 15], [330, 33], [313, 58], [295, 60]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0051_user_0015_scene_0002_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the dish off the dish rack', and now you are performing subtask 'take the dish off the dish rack.'\nStart at [0.63, 0.37]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.71, 0.51], [0.68, 0.52], [0.68, 0.47], [0.66, 0.44], [0.66, 0.15], [0.65, 0.05], [0.67, 0.06], [0.65, 0.04], [0.7, 0.21]]}" + } + ], + "images": "RH20T_cfg1_task_0051_user_0015_scene_0002_cfg_0001_038522062547#0", + "gt": "[[406, 134], [455, 185], [434, 186], [437, 171], [420, 160], [424, 55], [415, 18], [429, 21], [416, 13], [447, 77]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "21213_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the two paper bags out of the wooden crate', and now you are performing subtask 'place the bag in the box.'\nYou start from [0.68, 0.73]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.73], [0.65, 0.75], [0.66, 0.78], [0.68, 0.57], [0.7, 0.59], [0.74, 0.56], [0.74, 0.61], [0.76, 0.58], [0.7, 0.65], [0.71, 0.67]]}" + } + ], + "images": "21213_exterior_image_1_left#0", + "gt": "[[217, 131], [209, 135], [212, 140], [216, 102], [224, 107], [236, 101], [236, 109], [242, 105], [224, 117], [228, 120]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "65487_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the white mug to the left', and now you are performing subtask 'move the white mug to the left.'\nWith start location [0.22, 0.08], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.22, 0.08], [0.33, 0.11], [0.38, 0.17], [0.33, 0.16], [0.37, 0.21], [0.53, 0.29], [0.51, 0.28], [0.47, 0.39], [0.47, 0.51], [0.41, 0.44]]}" + } + ], + "images": "65487_exterior_image_2_left#0", + "gt": "[[70, 15], [106, 19], [121, 30], [106, 28], [117, 37], [169, 52], [162, 50], [152, 70], [152, 92], [131, 79]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0037_user_0014_scene_0001_cfg_0007_104122061330#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line', and now you are performing subtask 'pick up the fourth block from the table, transfer the fourth block from the table to the stacking position, place the fourth block on top of the third block.'\nThe trajectory starts at [0.48, 0.59]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.59], [0.52, 0.66], [0.52, 0.69], [0.51, 0.62], [0.47, 0.54], [0.43, 0.51], [0.41, 0.52], [0.4, 0.54], [0.44, 0.39], [0.44, 0.39]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0037_user_0014_scene_0001_cfg_0007_104122061330#373.jpg", + "gt": "[[307, 213], [335, 236], [336, 247], [328, 225], [298, 194], [272, 185], [260, 187], [257, 194], [283, 142], [283, 142]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "12172_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pour the contents of the cup into the bowl', and now you are performing subtask 'pour the contents of the cup into the bowl.'\nFrom start point [0.94, 0.47], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.94, 0.47], [0.84, 0.48], [0.74, 0.59], [0.69, 0.7], [0.7, 0.87], [0.7, 0.6], [0.66, 0.58], [0.65, 0.52], [0.73, 0.72], [0.73, 0.86]]}" + } + ], + "images": "12172_exterior_image_2_left#0", + "gt": "[[301, 85], [269, 87], [238, 106], [222, 126], [225, 156], [223, 108], [211, 104], [209, 94], [234, 130], [234, 154]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "84069_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the flask', and now you are performing subtask 'open the flask.'\nStart at [0.37, 0.31]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.31], [0.38, 0.29], [0.38, 0.33], [0.36, 0.42], [0.39, 0.47], [0.39, 0.51], [0.37, 0.6], [0.35, 0.6], [0.34, 0.44], [0.35, 0.35]]}" + } + ], + "images": "84069_exterior_image_2_left#0", + "gt": "[[118, 55], [122, 52], [123, 60], [114, 75], [124, 84], [125, 92], [117, 108], [113, 108], [108, 79], [111, 63]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "48974_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the first switch from the right', and now you are performing subtask 'turn on the first switch from the right.'\nFrom start point [0.39, 0.25], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.25], [0.49, 0.19], [0.58, 0.11], [0.54, 0.14], [0.57, 0.13], [0.6, 0.18], [0.6, 0.41], [0.62, 0.43], [0.62, 0.47], [0.62, 0.31]]}" + } + ], + "images": "48974_exterior_image_2_left#0", + "gt": "[[125, 45], [157, 34], [185, 19], [174, 26], [183, 23], [191, 33], [193, 73], [197, 78], [197, 84], [197, 56]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "1275_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the lid on the countertop', and now you are performing subtask 'put the lid on the countertop.'\nYour initial coordinate is [0.81, 0.24]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.81, 0.24], [0.76, 0.11], [0.72, 0.13], [0.65, 0.38], [0.66, 0.46], [0.67, 0.33], [0.64, 0.26], [0.54, 0.26], [0.51, 0.54], [0.54, 0.62]]}" + } + ], + "images": "1275_exterior_image_2_left#0", + "gt": "[[258, 43], [243, 20], [230, 24], [207, 69], [211, 82], [215, 59], [206, 46], [174, 47], [164, 97], [174, 112]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0065_user_0015_scene_0001_cfg_0001_038522062547#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with small Lego blocks for the nth time with n blocks', and now you are performing subtask 'press the black block onto the base plate.'\nStart at [0.59, 0.34]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.34], [0.6, 0.35], [0.6, 0.34], [0.58, 0.34], [0.59, 0.34], [0.58, 0.28], [0.59, 0.2], [0.6, 0.14], [0.59, 0.1], [0.6, 0.09]]}" + } + ], + "images": "RH20T_cfg1_task_0065_user_0015_scene_0001_cfg_0001_038522062547#466", + "gt": "[[379, 122], [381, 126], [381, 123], [372, 123], [377, 121], [374, 100], [379, 73], [381, 50], [379, 36], [381, 34]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0032_user_0016_scene_0006_cfg_0004_104122062295#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'slide the outer casing onto the gift box', and now you are performing subtask 'push the gift box lid on the desktop.'\nYou begin at [0.55, 0.68]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.68], [0.63, 0.56], [0.58, 0.54], [0.55, 0.71], [0.56, 0.84], [0.49, 0.67], [0.4, 0.68], [0.39, 0.55], [0.36, 0.56], [0.36, 0.56]]}" + } + ], + "images": "RH20T_cfg4_task_0032_user_0016_scene_0006_cfg_0004_104122062295#221", + "gt": "[[351, 244], [404, 201], [372, 195], [351, 255], [357, 302], [311, 240], [255, 243], [251, 198], [233, 201], [233, 201]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0058_user_0010_scene_0001_cfg_0003_104122062823#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn off the power strips', and now you are performing subtask 'press the button on the third power strip to turn off the power, press the button on the fourth power strip to turn off the power.'\nBegin at [0.68, 0.54] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.54], [0.69, 0.56], [0.69, 0.58], [0.69, 0.56], [0.68, 0.53], [0.65, 0.45], [0.61, 0.38], [0.58, 0.35], [0.59, 0.36], [0.59, 0.36]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg3_task_0058_user_0010_scene_0001_cfg_0003_104122062823#323.jpg", + "gt": "[[434, 194], [442, 200], [441, 208], [441, 201], [437, 192], [419, 162], [391, 136], [374, 126], [379, 128], [378, 131]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "59697_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the objects from the orange bowl to the black pot, then cover the black pot with the glass lid', and now you are performing subtask 'pick up the item from the orange bowl.'\nStart at [0.78, 0.01]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.78, 0.01], [0.72, 0.32], [0.66, 0.41], [0.72, 0.47], [0.69, 0.3], [0.61, 0.46], [0.64, 0.49], [0.73, 0.3], [0.69, 0.36], [0.71, 0.47]]}" + } + ], + "images": "59697_exterior_image_2_left#0", + "gt": "[[249, 1], [231, 58], [211, 74], [230, 85], [222, 54], [195, 83], [206, 89], [234, 54], [222, 64], [227, 84]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "86920_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'cover the pot and fold the cloth', and now you are performing subtask 'fold the cloth on the table.'\nWith start location [0.95, 0.44], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.95, 0.44], [0.83, 0.54], [0.79, 0.61], [0.82, 0.53], [0.8, 0.48], [0.71, 0.48], [0.69, 0.57], [0.69, 0.52], [0.67, 0.48], [0.62, 0.49]]}" + } + ], + "images": "86920_exterior_image_2_left#110", + "gt": "[[304, 80], [267, 97], [253, 110], [261, 95], [255, 86], [227, 86], [222, 102], [221, 93], [213, 86], [198, 88]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0037_user_0017_scene_0006_cfg_0004_104122062823#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the third block on top of the second block on the right side of the table, pick up the fourth block from the left side of the table, transfer the fourth block from the left side of the table to the right side of the table, place the fourth block on top of the third block on the right side of the table.'\nYour starting position is [0.47, 0.2]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.2], [0.47, 0.13], [0.42, 0.09], [0.39, 0.11], [0.37, 0.19], [0.37, 0.22], [0.45, 0.1], [0.49, 0.12], [0.48, 0.13], [0.48, 0.16]]}" + } + ], + "images": "RH20T_cfg4_task_0037_user_0017_scene_0006_cfg_0004_104122062823#321", + "gt": "[[303, 73], [299, 48], [269, 33], [247, 41], [234, 67], [234, 81], [291, 37], [315, 44], [308, 47], [307, 56]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0201_user_0015_scene_0007_cfg_0001_750612070853#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'the medium pipette tip is inserted into the holder for medium pipette tips', and now you are performing subtask 'place the tip of the medium pipette into the holder for medium pipette tips.'\nStarting at [0.65, 0.33], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.33], [0.64, 0.35], [0.64, 0.35], [0.64, 0.36], [0.64, 0.36], [0.64, 0.31], [0.64, 0.31], [0.63, 0.35], [0.63, 0.37], [0.63, 0.4]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0201_user_0015_scene_0007_cfg_0001_750612070853#264.jpg", + "gt": "[[413, 119], [411, 125], [407, 126], [409, 129], [412, 129], [412, 110], [410, 110], [401, 125], [405, 132], [405, 145]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "37061_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the object from the cup and put it on the table', and now you are performing subtask 'remove the object from the cup and put it on the table.'\nYour initial coordinate is [0.99, 0.17]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.99, 0.17], [0.61, 0.12], [0.57, 0.29], [0.59, 0.28], [0.65, 0.09], [0.72, 0.09], [0.81, 0.18], [0.83, 0.28], [0.79, 0.4], [0.79, 0.26]]}" + } + ], + "images": "37061_exterior_image_1_left#0", + "gt": "[[318, 30], [195, 21], [183, 52], [190, 51], [209, 16], [231, 17], [260, 33], [266, 51], [252, 72], [254, 47]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122061330#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'place the lid on the table, pick up the object in the container, transfer the object from the container to the surface, place the object on the surface.'\nYou begin at [0.58, 0.19]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.19], [0.57, 0.16], [0.55, 0.29], [0.58, 0.4], [0.58, 0.36], [0.57, 0.33], [0.53, 0.4], [0.51, 0.56], [0.53, 0.6], [0.52, 0.57]]}" + } + ], + "images": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122061330#186", + "gt": "[[369, 68], [365, 56], [353, 106], [371, 145], [371, 130], [364, 119], [339, 143], [328, 202], [340, 217], [334, 206]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0058_user_0007_scene_0002_cfg_0005_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press a row of two power strip buttons to turn off the power strips', and now you are performing subtask 'press the first button on the power socket.'\nBegin at [0.45, 0.42] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.42], [0.41, 0.52], [0.4, 0.59], [0.39, 0.64], [0.39, 0.66], [0.37, 0.68], [0.37, 0.68], [0.37, 0.68], [0.37, 0.66], [0.39, 0.64]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0058_user_0007_scene_0002_cfg_0005_037522062165#0.jpg", + "gt": "[[289, 152], [263, 186], [253, 214], [251, 231], [248, 238], [238, 244], [239, 246], [236, 246], [237, 239], [247, 229]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "45040_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer and put some of the plush toys in the drawer and close it', and now you are performing subtask 'place the pineapple plush toy in the drawer.'\nYour initial coordinate is [0.5, 0.31]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.31], [0.47, 0.31], [0.41, 0.15], [0.34, 0.17], [0.31, 0.27], [0.3, 0.24], [0.3, 0.3], [0.34, 0.15], [0.43, 0.24], [0.47, 0.23]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/45040_exterior_image_1_left#317.jpg", + "gt": "[[158.5, 55.5], [149.5, 56.5], [132.5, 27.5], [108.5, 30.5], [100.5, 48.5], [96.5, 43.5], [96.5, 53.5], [108.5, 27.5], [136.5, 42.5], [150.5, 41.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "52635_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the towel', and now you are performing subtask 'slide the towel to the left.'\nStart at [0.64, 0.34]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.64, 0.34], [0.62, 0.38], [0.61, 0.38], [0.55, 0.37], [0.62, 0.24], [0.73, 0.17], [0.71, 0.18], [0.77, 0.19], [0.75, 0.2], [0.76, 0.21]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/52635_exterior_image_1_left#252.jpg", + "gt": "[[204, 61], [199, 68], [194, 69], [176, 66], [199, 43], [233, 30], [227, 32], [246, 35], [239, 36], [242, 37]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0009_user_0014_scene_0007_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take out a Hanoi block and throw it aside', and now you are performing subtask 'pick up the ring from the wooden rack.'\nWith start location [0.63, 0.37], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.71, 0.25], [0.7, 0.26], [0.69, 0.22], [0.65, 0.16], [0.63, 0.07], [0.62, 0.08], [0.6, 0.13], [0.63, 0.11], [0.62, 0.2]]}" + } + ], + "images": "RH20T_cfg1_task_0009_user_0014_scene_0007_cfg_0001_038522062547#0", + "gt": "[[406, 133], [453, 91], [449, 93], [442, 78], [419, 58], [403, 26], [397, 27], [386, 48], [404, 40], [399, 73]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0090_user_0007_scene_0001_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the wooden block', and now you are performing subtask 'press the wooden block.'\nBegin at [0.63, 0.37] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.62, 0.47], [0.61, 0.54], [0.6, 0.6], [0.6, 0.6], [0.61, 0.62], [0.6, 0.72], [0.61, 0.76], [0.6, 0.6], [0.6, 0.6]]}" + } + ], + "images": "RH20T_cfg1_task_0090_user_0007_scene_0001_cfg_0001_038522062547#0", + "gt": "[[406, 133], [399, 171], [388, 196], [385, 215], [382, 216], [388, 224], [384, 258], [389, 274], [386, 216], [386, 216]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0096_user_0014_scene_0008_cfg_0004_036422060909#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take something out of a drawer', and now you are performing subtask 'pull the drawer handle to open the drawer, pull the drawer handle to open the drawer, pick up the object from the drawer.'\nYou start from [0.53, 0.22]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.22], [0.56, 0.25], [0.57, 0.14], [0.54, 0.24], [0.49, 0.28], [0.5, 0.23], [0.46, 0.3], [0.51, 0.12], [0.53, 0.1], [0.53, 0.13]]}" + } + ], + "images": "RH20T_cfg4_task_0096_user_0014_scene_0008_cfg_0004_036422060909#0", + "gt": "[[340, 80], [357, 89], [365, 49], [344, 86], [314, 99], [318, 83], [295, 107], [328, 45], [337, 37], [338, 47]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0016_user_0014_scene_0001_cfg_0001_038522063145#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'hit the pool ball', and now you are performing subtask 'hit the pool ball.'\nYou start from [0.69, 0.27]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.27], [0.67, 0.23], [0.67, 0.21], [0.57, 0.12], [0.51, 0.17], [0.5, 0.21], [0.53, 0.19], [0.52, 0.21], [0.5, 0.21], [0.53, 0.26]]}" + } + ], + "images": "RH20T_cfg1_task_0016_user_0014_scene_0001_cfg_0001_038522063145#0", + "gt": "[[443, 97], [429, 84], [431, 74], [368, 43], [329, 61], [318, 77], [337, 67], [336, 77], [317, 74], [340, 94]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "14053_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the small brush from the mug and put it on the table. pick up the mug and slightly move it to the right. then pick up the small brush from the table and put it back in the mug', and now you are performing subtask 'place the brush on the table.'\nYou begin at [0.43, 0.0]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.0], [0.49, 0.08], [0.46, 0.18], [0.48, 0.31], [0.47, 0.22], [0.49, 0.21], [0.52, 0.27], [0.52, 0.37], [0.5, 0.27], [0.5, 0.27]]}" + } + ], + "images": "14053_exterior_image_1_left#0", + "gt": "[[137, 0], [158, 14], [147, 33], [155, 55], [152, 40], [158, 38], [166, 48], [166, 66], [160, 49], [160, 49]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0001_user_0014_scene_0007_cfg_0007_104122061018#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the nearest green button', and now you are performing subtask 'press the nearest green button.'\nYour starting position is [0.57, 0.09]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.09], [0.57, 0.08], [0.53, 0.07], [0.47, 0.1], [0.42, 0.17], [0.42, 0.26], [0.41, 0.23], [0.42, 0.17], [0.49, 0.06], [0.56, 0.05]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0001_user_0014_scene_0007_cfg_0007_104122061018#0.jpg", + "gt": "[[363, 31], [367, 27], [342, 24], [302, 37], [271, 61], [267, 95], [260, 82], [266, 60], [316, 23], [359, 18]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0034_user_0003_scene_0006_cfg_0002_037522062165#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'pick up the second square next to the first square, transfer the second square from the countertop to the left side of the first square, place the second square to the left of the first square.'\nYou start from [0.3, 0.57]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.3, 0.57], [0.31, 0.54], [0.32, 0.56], [0.3, 0.57], [0.32, 0.57], [0.33, 0.5], [0.33, 0.52], [0.32, 0.53], [0.29, 0.5], [0.29, 0.55]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg2_task_0034_user_0003_scene_0006_cfg_0002_037522062165#214.jpg", + "gt": "[[190, 205], [197, 195], [204, 203], [195, 204], [203, 206], [211, 180], [212, 187], [203, 192], [188, 180], [183, 198]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_037522061512#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the first block on the table, transfer the first block from the table to above the second block, place the first block on top of the second block, pick up the third block on the table.'\nBegin at [0.4, 0.23] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.23], [0.38, 0.21], [0.4, 0.21], [0.39, 0.2], [0.39, 0.22], [0.42, 0.3], [0.42, 0.26], [0.46, 0.32], [0.43, 0.27], [0.44, 0.38]]}" + } + ], + "images": "RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_037522061512#0", + "gt": "[[257, 83], [246, 77], [259, 77], [250, 72], [252, 80], [270, 108], [270, 93], [295, 115], [275, 97], [280, 136]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0118_user_0007_scene_0010_cfg_0005_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the water tap', and now you are performing subtask 'turn on the water tap.'\nStart at [0.45, 0.29]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.29], [0.44, 0.26], [0.42, 0.25], [0.39, 0.33], [0.35, 0.34], [0.36, 0.32], [0.39, 0.29], [0.4, 0.35], [0.42, 0.34], [0.42, 0.36]]}" + } + ], + "images": "RH20T_cfg5_task_0118_user_0007_scene_0010_cfg_0005_036422060215#0", + "gt": "[[286, 106], [280, 94], [269, 90], [250, 118], [226, 124], [229, 116], [249, 105], [258, 125], [268, 121], [266, 128]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "50330_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the hand towel out of the second drawer', and now you are performing subtask 'take the hand towel out of the second drawer.'\nYour starting position is [0.15, 0.14]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.15, 0.14], [0.34, 0.41], [0.32, 0.22], [0.49, 0.49], [0.43, 0.69], [0.23, 0.5], [0.22, 0.38], [0.24, 0.27], [0.27, 0.23], [0.25, 0.32]]}" + } + ], + "images": "50330_exterior_image_1_left#0", + "gt": "[[49.5, 26.0], [107.5, 73.0], [102.5, 39.0], [156.5, 88.0], [136.5, 124.0], [74.5, 90.0], [69.5, 68.0], [75.5, 48.0], [86.5, 42.0], [81.5, 58.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0057_user_0016_scene_0006_cfg_0002_037522062165#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn on the power strips', and now you are performing subtask 'press the button on the second socket.'\nYour starting position is [0.39, 0.68]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.68], [0.38, 0.69], [0.38, 0.7], [0.38, 0.71], [0.38, 0.73], [0.39, 0.73], [0.4, 0.71], [0.39, 0.73], [0.38, 0.71], [0.4, 0.71]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg2_task_0057_user_0016_scene_0006_cfg_0002_037522062165#243.jpg", + "gt": "[[247, 246], [240, 249], [245, 253], [242, 257], [245, 263], [251, 262], [253, 257], [248, 261], [243, 256], [253, 257]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0064_user_0016_scene_0003_cfg_0004_036422060909#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the roll of paper down from the shelf', and now you are performing subtask 'pick up the roll of paper from the shelf, transfer the roll of paper from the shelf to the workspace, place the roll of paper in the workspace.'\nYou start from [0.53, 0.23]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.23], [0.52, 0.23], [0.52, 0.18], [0.54, 0.16], [0.55, 0.19], [0.54, 0.22], [0.54, 0.21], [0.44, 0.22], [0.48, 0.2], [0.49, 0.15]]}" + } + ], + "images": "RH20T_cfg4_task_0064_user_0016_scene_0003_cfg_0004_036422060909#168", + "gt": "[[339, 83], [332, 82], [330, 66], [345, 59], [350, 68], [343, 81], [344, 77], [284, 79], [306, 73], [316, 54]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "1389_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the pen out of the cup and put it on the table', and now you are performing subtask 'take the pen out of the cup and put it on the table.'\nStart at [0.53, 0.44]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.44], [0.63, 0.46], [0.69, 0.53], [0.75, 0.53], [0.76, 0.55], [0.74, 0.58], [0.73, 0.51], [0.78, 0.51], [0.8, 0.57], [0.8, 0.74]]}" + } + ], + "images": "1389_exterior_image_1_left#0", + "gt": "[[168.5, 79.0], [202.5, 82.0], [221.5, 96.0], [239.5, 95.0], [243.5, 99.0], [236.5, 104.0], [233.5, 92.0], [249.5, 92.0], [256.5, 103.0], [254.5, 133.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'fold the first lid on the box on the table.'\nYou begin at [0.63, 0.37]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.59, 0.57], [0.56, 0.6], [0.55, 0.63], [0.55, 0.64], [0.55, 0.63], [0.57, 0.71], [0.56, 0.68], [0.57, 0.65], [0.59, 0.61]]}" + } + ], + "images": "RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_038522062547#0", + "gt": "[[406, 132], [379, 206], [357, 216], [349, 228], [354, 231], [349, 226], [364, 254], [360, 245], [363, 233], [377, 218]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0033_user_0010_scene_0010_cfg_0005_104122063678#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp a small block to sweep the other small block into the designated position', and now you are performing subtask 'push another block towards the mark, place the block on the mark.'\nStarting at [0.58, 0.76], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.76], [0.54, 0.73], [0.57, 0.73], [0.62, 0.78], [0.63, 0.84], [0.63, 0.74], [0.6, 0.62], [0.51, 0.49], [0.5, 0.46], [0.52, 0.46]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0033_user_0010_scene_0010_cfg_0005_104122063678#681.jpg", + "gt": "[[373, 274], [348, 262], [366, 264], [397, 281], [406, 301], [405, 266], [384, 224], [329, 177], [323, 166], [333, 165]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0020_user_0014_scene_0001_cfg_0006_037522061512#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the desk lamp by pressing the button', and now you are performing subtask 'turn on the desk lamp by pressing the button.'\nYour starting position is [0.4, 0.23]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.23], [0.39, 0.2], [0.37, 0.39], [0.34, 0.47], [0.32, 0.57], [0.33, 0.66], [0.32, 0.79], [0.35, 0.67], [0.36, 0.46], [0.36, 0.46]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0020_user_0014_scene_0001_cfg_0006_037522061512#0.jpg", + "gt": "[[257, 83], [251, 73], [235, 139], [216, 170], [207, 206], [213, 237], [204, 286], [221, 242], [233, 164], [233, 164]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "90300_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the fork on the white plate', and now you are performing subtask 'put the fork on the white plate.'\nYou begin at [0.85, 0.04]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.85, 0.04], [0.7, 0.21], [0.58, 0.4], [0.58, 0.5], [0.55, 0.56], [0.57, 0.59], [0.58, 0.56], [0.59, 0.58], [0.56, 0.65], [0.55, 0.61]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/90300_exterior_image_1_left#0.jpg", + "gt": "[[270.5, 8.0], [223.5, 37.0], [185.5, 72.0], [186.5, 90.0], [177.5, 100.0], [181.5, 106.0], [186.5, 100.0], [188.5, 104.0], [179.5, 117.0], [177.5, 109.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "47649_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the candy bar from the sink', and now you are performing subtask 'remove the candy bar from the sink.'\nYou begin at [0.71, 0.46]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.71, 0.46], [0.65, 0.68], [0.65, 0.76], [0.67, 0.81], [0.65, 0.72], [0.6, 0.71], [0.54, 0.76], [0.51, 0.82], [0.54, 0.77], [0.63, 0.71]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/47649_exterior_image_2_left#0.jpg", + "gt": "[[226, 83], [207, 123], [208, 137], [213, 145], [208, 129], [192, 127], [174, 136], [163, 147], [172, 138], [201, 128]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg3_task_0018_user_0010_scene_0003_cfg_0003_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'transfer the wooden block from the block pile to the tabletop', and now you are performing subtask 'transfer the wooden block from the block pile to the tabletop.'\nYour initial coordinate is [0.31, 0.11]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.31, 0.11], [0.35, 0.1], [0.3, 0.15], [0.34, 0.15], [0.36, 0.19], [0.33, 0.38], [0.37, 0.26], [0.4, 0.3], [0.41, 0.4], [0.43, 0.33]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg3_task_0018_user_0010_scene_0003_cfg_0003_104122062823#0.jpg", + "gt": "[[197, 41], [224, 37], [189, 53], [216, 54], [228, 67], [213, 137], [234, 93], [254, 108], [261, 144], [275, 119]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_038522062547#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'pick up the fifth square from the right side of the table, transfer the fifth square from the right side of the table to the top of the second square, place the fifth square on top of the second square, pick up the sixth square on the table.'\nFrom start point [0.67, 0.44], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.67, 0.44], [0.71, 0.44], [0.72, 0.46], [0.71, 0.44], [0.65, 0.44], [0.62, 0.47], [0.62, 0.49], [0.6, 0.47], [0.62, 0.44], [0.7, 0.41]]}" + } + ], + "images": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_038522062547#736", + "gt": "[[428, 159], [453, 157], [459, 166], [457, 158], [419, 159], [394, 168], [395, 175], [383, 169], [399, 157], [451, 148]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0010_user_0010_scene_0005_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the handset of the telephone on the corresponding phone cradle', and now you are performing subtask 'place the handset of the telephone on the corresponding phone cradle.'\nStart at [0.5, 0.37]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.5, 0.59], [0.52, 0.64], [0.5, 0.8], [0.53, 0.66], [0.51, 0.45], [0.52, 0.43], [0.53, 0.46], [0.52, 0.49], [0.52, 0.43]]}" + } + ], + "images": "RH20T_cfg5_task_0010_user_0010_scene_0005_cfg_0005_104122063678#0", + "gt": "[[323, 132], [323, 213], [335, 232], [320, 289], [340, 238], [327, 162], [332, 156], [342, 166], [333, 175], [333, 156]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0005_user_0010_scene_0004_cfg_0004_104122063550#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'approach and touch the side of the small block', and now you are performing subtask 'approach and touch the side of the small block.'\nYou begin at [0.43, 0.51]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.51], [0.55, 0.54], [0.5, 0.57], [0.51, 0.55], [0.51, 0.73], [0.53, 0.82], [0.52, 0.72], [0.54, 0.8], [0.55, 0.66], [0.55, 0.66]]}" + } + ], + "images": "RH20T_cfg4_task_0005_user_0010_scene_0004_cfg_0004_104122063550#0", + "gt": "[[278, 183], [352, 194], [321, 207], [326, 197], [328, 261], [339, 296], [331, 260], [346, 288], [350, 239], [350, 239]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0078_user_0014_scene_0008_cfg_0006_037522061512#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the microwave door', and now you are performing subtask 'push the door close the microwave door.'\nYour starting position is [0.4, 0.07]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.08], [0.41, 0.06], [0.42, 0.05], [0.44, 0.05], [0.48, 0.03], [0.51, 0.0], [0.46, 0.0], [0.45, 0.03], [0.47, 0.06], [0.47, 0.08]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0078_user_0014_scene_0008_cfg_0006_037522061512#125.jpg", + "gt": "[[257, 27], [263, 21], [269, 19], [284, 18], [307, 11], [325, 0], [294, 1], [290, 10], [301, 22], [302, 27]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "67326_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the T-shirt on the table', and now you are performing subtask 'unfold the T-shirt on the table.'\nYou start from [0.49, 0.54]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.54], [0.46, 0.54], [0.46, 0.56], [0.48, 0.55], [0.48, 0.56], [0.48, 0.57], [0.47, 0.62], [0.44, 0.72], [0.45, 0.71], [0.43, 0.76]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/67326_exterior_image_2_left#151.jpg", + "gt": "[[157, 98], [146, 97], [148, 100], [155, 99], [153, 101], [155, 102], [151, 112], [140, 130], [143, 127], [138, 137]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0081_user_0010_scene_0009_cfg_0001_819112070044#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the leg of the glasses (with one hand)', and now you are performing subtask 'fold the leg of the glasses to unfold, fold the leg of the glasses to unfold.'\nWith start location [0.52, 0.46], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.46], [0.53, 0.52], [0.52, 0.53], [0.52, 0.51], [0.54, 0.53], [0.53, 0.5], [0.55, 0.5], [0.57, 0.45], [0.59, 0.43], [0.59, 0.44]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0081_user_0010_scene_0009_cfg_0001_819112070044#221.jpg", + "gt": "[[334, 166], [338, 189], [335, 190], [335, 184], [345, 192], [341, 181], [349, 179], [366, 161], [378, 154], [377, 157]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "65454_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up a lemon toy and put it in the transparent bowl', and now you are performing subtask 'pick up a lemon toy and put it in the transparent bowl.'\nYour starting position is [0.37, 0.35]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.35], [0.48, 0.41], [0.49, 0.41], [0.49, 0.41], [0.55, 0.51], [0.57, 0.64], [0.56, 0.47], [0.55, 0.44], [0.54, 0.48], [0.55, 0.41]]}" + } + ], + "images": "65454_exterior_image_2_left#0", + "gt": "[[119, 63], [153, 74], [158, 73], [156, 74], [176, 91], [183, 116], [180, 84], [176, 79], [174, 87], [175, 74]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0081_user_0007_scene_0003_cfg_0005_105422061350#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the leg of the glasses (with one hand)', and now you are performing subtask 'pick up the glasses on the table, fold the legs of the glasses to unfold them.'\nYour starting position is [0.43, 0.13]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.13], [0.38, 0.41], [0.35, 0.38], [0.32, 0.41], [0.32, 0.35], [0.3, 0.32], [0.28, 0.23], [0.27, 0.33], [0.3, 0.36], [0.3, 0.36]]}" + } + ], + "images": "RH20T_cfg5_task_0081_user_0007_scene_0003_cfg_0005_105422061350#0", + "gt": "[[276, 47], [245, 149], [222, 136], [207, 146], [208, 127], [191, 114], [177, 83], [172, 118], [189, 130], [189, 130]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "31362_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the towel twice', and now you are performing subtask 'fold the towel again.'\nThe trajectory starts at [0.44, 0.58]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.58], [0.46, 0.63], [0.43, 0.72], [0.43, 0.53], [0.44, 0.43], [0.49, 0.4], [0.54, 0.43], [0.44, 0.37], [0.43, 0.35], [0.45, 0.36]]}" + } + ], + "images": "31362_exterior_image_1_left#139", + "gt": "[[141, 104], [146, 114], [137, 130], [136, 96], [142, 78], [156, 72], [173, 78], [141, 67], [138, 63], [145, 65]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0071_user_0015_scene_0010_cfg_0004_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close a sliding window', and now you are performing subtask 'push the handle of the sliding window to close the window.'\nStart at [0.45, 0.0]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.0], [0.48, 0.0], [0.46, 0.0], [0.45, 0.01], [0.44, 0.06], [0.47, 0.16], [0.47, 0.21], [0.49, 0.24], [0.52, 0.38], [0.52, 0.52]]}" + } + ], + "images": "RH20T_cfg4_task_0071_user_0015_scene_0010_cfg_0004_104122062823#0", + "gt": "[[291, 0], [310, 0], [297, 0], [290, 3], [282, 22], [301, 59], [299, 75], [315, 88], [332, 137], [334, 186]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "32611_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the pen out of the mug and put it on the counter', and now you are performing subtask 'take the pen out of the mug and put it on the counter.'\nThe trajectory starts at [0.09, 0.0]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.09, 0.0], [0.19, 0.11], [0.27, 0.16], [0.3, 0.27], [0.31, 0.04], [0.32, 0.0], [0.34, 0.0], [0.36, 0.21], [0.38, 0.17], [0.38, 0.17]]}" + } + ], + "images": "32611_exterior_image_1_left#0", + "gt": "[[28.5, 0.0], [60.5, 20.0], [87.5, 28.0], [96.5, 48.0], [98.5, 8.0], [101.5, 0.0], [107.5, 0.0], [114.5, 38.0], [121.5, 31.0], [121.5, 31.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "81736_exterior_image_1_left#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the middle drawer doors', and now you are performing subtask 'push the handle of the left-side middle cabinet door to fully close the cabinet door.'\nYou begin at [0.4, 0.31]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.31], [0.35, 0.34], [0.35, 0.36], [0.41, 0.41], [0.44, 0.39], [0.47, 0.41], [0.48, 0.39], [0.36, 0.36], [0.32, 0.41], [0.25, 0.42]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/81736_exterior_image_1_left#268.jpg", + "gt": "[[127, 56], [111, 61], [113, 64], [131, 73], [141, 70], [149, 73], [155, 70], [115, 65], [101, 73], [80, 75]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0057_user_0014_scene_0008_cfg_0004_038522062288#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn on the power strips', and now you are performing subtask 'press the button on the second power strip.'\nYou begin at [0.43, 0.41]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.41], [0.43, 0.41], [0.43, 0.41], [0.43, 0.43], [0.43, 0.44], [0.43, 0.44], [0.43, 0.42], [0.43, 0.42], [0.41, 0.41], [0.41, 0.4]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0057_user_0014_scene_0008_cfg_0004_038522062288#169.jpg", + "gt": "[[275, 148], [275, 146], [273, 147], [274, 154], [273, 158], [272, 158], [275, 152], [274, 150], [264, 148], [260, 144]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "4908_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull out the top shelf of the nightstand', and now you are performing subtask 'slide the top shelf of the nightstand out.'\nYou start from [0.46, 0.01]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.01], [0.41, 0.17], [0.45, 0.27], [0.42, 0.36], [0.47, 0.37], [0.57, 0.29], [0.51, 0.2], [0.53, 0.28], [0.62, 0.28], [0.62, 0.28]]}" + } + ], + "images": "4908_exterior_image_1_left#0", + "gt": "[[147.0, 2.5], [131.0, 31.5], [143.0, 48.5], [134.0, 64.5], [152.0, 66.5], [184.0, 52.5], [162.0, 36.5], [170.0, 50.5], [197.0, 49.5], [197.0, 49.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "1291_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the cube in the bowl', and now you are performing subtask 'put the cube in the bowl.'\nYou begin at [1.0, 0.84]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[1.0, 0.84], [0.77, 0.44], [0.72, 0.42], [0.68, 0.45], [0.63, 0.67], [0.67, 0.56], [0.62, 0.44], [0.6, 0.56], [0.64, 0.54], [0.63, 0.52]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/1291_exterior_image_2_left#0.jpg", + "gt": "[[319, 151], [246, 80], [231, 75], [217, 81], [202, 121], [214, 100], [198, 80], [193, 101], [206, 97], [201, 94]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "43469_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the sweatshirt twice', and now you are performing subtask 'fold the sweatshirt in half.'\nThe trajectory starts at [0.9, 0.03]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.9, 0.03], [0.88, 0.01], [0.84, 0.2], [0.87, 0.33], [0.86, 0.45], [0.77, 0.14], [0.7, 0.01], [0.68, 0.0], [0.71, 0.0], [0.75, 0.13]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/43469_exterior_image_1_left#0.jpg", + "gt": "[[287, 5], [282, 1], [269, 36], [279, 59], [276, 81], [245, 25], [224, 1], [216, 0], [226, 0], [241, 23]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "9160_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the pen inside the white mug', and now you are performing subtask 'put the pen inside the white mug.'\nBegin at [0.6, 0.2] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.2], [0.48, 0.27], [0.29, 0.51], [0.15, 0.76], [0.14, 0.82], [0.17, 0.87], [0.4, 0.38], [0.43, 0.36], [0.44, 0.42], [0.43, 0.38]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/9160_exterior_image_2_left#0.jpg", + "gt": "[[193, 36], [154, 48], [92, 92], [48, 137], [46, 147], [56, 156], [127, 69], [138, 64], [142, 75], [138, 69]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "67887_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'arrange the cushions on the sofa', and now you are performing subtask 'pick up the horizontally placed sofa cushion on the right side of the sofa.'\nYou begin at [0.46, 0.56]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.56], [0.52, 0.46], [0.46, 0.67], [0.49, 0.54], [0.5, 0.64], [0.61, 0.73], [0.63, 0.69], [0.57, 0.6], [0.49, 0.63], [0.64, 0.89]]}" + } + ], + "images": "67887_exterior_image_2_left#0", + "gt": "[[148, 101], [165, 83], [146, 120], [158, 98], [160, 116], [196, 132], [202, 125], [183, 108], [156, 113], [206, 161]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "79152_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the white marker and put it in the green cup', and now you are performing subtask 'pick up the white marker and put it in the green cup.'\nBegin at [0.58, 0.06] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.06], [0.51, 0.09], [0.45, 0.17], [0.42, 0.31], [0.42, 0.46], [0.37, 0.42], [0.35, 0.5], [0.34, 0.47], [0.37, 0.46], [0.37, 0.46]]}" + } + ], + "images": "79152_exterior_image_2_left#0", + "gt": "[[186, 10], [162, 16], [145, 30], [133, 56], [135, 82], [119, 75], [113, 90], [110, 85], [119, 83], [119, 83]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg3_task_0038_user_0010_scene_0010_cfg_0003_104122062295#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the cup', and now you are performing subtask 'pick up the cup.'\nBegin at [0.6, 0.19] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.19], [0.65, 0.19], [0.52, 0.27], [0.48, 0.35], [0.4, 0.33], [0.34, 0.43], [0.32, 0.38], [0.29, 0.43], [0.32, 0.42], [0.48, 0.17]]}" + } + ], + "images": "RH20T_cfg3_task_0038_user_0010_scene_0010_cfg_0003_104122062295#0", + "gt": "[[381, 70], [415, 70], [336, 96], [305, 125], [258, 120], [220, 155], [207, 135], [184, 156], [203, 151], [307, 63]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_038522062547#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'fold the second lid on the box on the table, fold the third lid on the box on the table.'\nYou begin at [0.59, 0.54]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.54], [0.59, 0.56], [0.57, 0.54], [0.56, 0.49], [0.48, 0.41], [0.49, 0.43], [0.49, 0.45], [0.48, 0.41], [0.52, 0.44], [0.57, 0.54]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_038522062547#162.jpg", + "gt": "[[375, 193], [376, 200], [366, 195], [356, 176], [309, 149], [315, 153], [311, 161], [310, 149], [330, 159], [368, 196]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "81842_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place one block on top of the stacked blue and green blocks', and now you are performing subtask 'place the blue block on the top of the pile of brown and green blocks.'\nYou start from [0.57, 0.32]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.32], [0.55, 0.25], [0.55, 0.25], [0.56, 0.27], [0.55, 0.3], [0.55, 0.19], [0.55, 0.17], [0.56, 0.16], [0.59, 0.15], [0.59, 0.16]]}" + } + ], + "images": "81842_exterior_image_1_left#399", + "gt": "[[181.5, 58.0], [177.5, 45.0], [176.5, 45.0], [178.5, 48.0], [177.5, 54.0], [174.5, 34.0], [175.5, 30.0], [179.5, 28.0], [187.5, 27.0], [188.5, 29.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0201_user_0015_scene_0007_cfg_0001_819112071037#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'the medium pipette tip is inserted into the holder for medium pipette tips', and now you are performing subtask 'pick up the medium pipette from the table, transfer the medium pipette from the table to the holder for medium pipette tips, place the tip of the medium pipette into the holder for medium pipette tips.'\nStarting at [0.57, 0.0], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.0], [0.51, 0.01], [0.47, 0.06], [0.48, 0.0], [0.41, 0.23], [0.44, 0.6], [0.45, 0.37], [0.51, 0.19], [0.52, 0.17], [0.57, 0.24]]}" + } + ], + "images": "RH20T_cfg1_task_0201_user_0015_scene_0007_cfg_0001_819112071037#0", + "gt": "[[368, 0], [329, 3], [304, 21], [306, 0], [265, 83], [280, 217], [290, 134], [324, 68], [331, 62], [366, 88]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "69180_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the first drawer from the right, remove the glue stick from it and put the glue stick on the counter', and now you are performing subtask 'pull the handle of the first drawer on the right to open the drawer.'\nYour initial coordinate is [0.63, 0.21]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.21], [0.54, 0.42], [0.54, 0.63], [0.47, 0.61], [0.55, 0.72], [0.53, 0.59], [0.52, 0.64], [0.52, 0.5], [0.53, 0.51], [0.52, 0.51]]}" + } + ], + "images": "69180_exterior_image_2_left#0", + "gt": "[[202, 37], [174, 76], [172, 114], [151, 109], [176, 129], [170, 106], [165, 116], [166, 90], [170, 91], [165, 92]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0110_user_0015_scene_0003_cfg_0001_750612070851#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the pencil out from the pencil sharpener', and now you are performing subtask 'take the pencil out from the pencil sharpener.'\nYour starting position is [0.49, 0.47]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.47], [0.62, 0.62], [0.6, 0.64], [0.63, 0.77], [0.66, 0.8], [0.7, 0.76], [0.72, 0.71], [0.76, 0.76], [0.72, 0.72], [0.61, 0.71]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0110_user_0015_scene_0003_cfg_0001_750612070851#0.jpg", + "gt": "[[312, 169], [399, 224], [383, 229], [405, 278], [421, 289], [451, 275], [463, 257], [485, 272], [461, 258], [393, 257]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "23167_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'wipe the table with the cloth', and now you are performing subtask 'slide the cloth on the table to wipe the surface.'\nYou start from [0.53, 0.48]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.48], [0.52, 0.52], [0.69, 0.6], [0.61, 0.51], [0.69, 0.53], [0.6, 0.68], [0.5, 0.57], [0.48, 0.51], [0.52, 0.41], [0.48, 0.43]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/23167_exterior_image_1_left#70.jpg", + "gt": "[[168, 86], [167, 94], [221, 108], [194, 92], [221, 96], [193, 123], [159, 103], [154, 91], [166, 74], [154, 77]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "51525_exterior_image_2_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the microwave door, move the pink jar on the counter', and now you are performing subtask 'pick up the pink jar on the microwave.'\nFrom start point [0.49, 0.12], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.12], [0.56, 0.07], [0.5, 0.12], [0.51, 0.12], [0.51, 0.13], [0.56, 0.11], [0.65, 0.18], [0.66, 0.26], [0.67, 0.27], [0.66, 0.3]]}" + } + ], + "images": "51525_exterior_image_2_left#96", + "gt": "[[156, 22], [179, 13], [161, 21], [164, 21], [164, 23], [178, 19], [207, 32], [210, 46], [214, 49], [212, 54]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "12780_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the sachets from the countertop and put them in the box', and now you are performing subtask 'pick up the sachet from the countertop.'\nThe trajectory starts at [0.54, 0.46]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.46], [0.48, 0.37], [0.43, 0.36], [0.44, 0.36], [0.43, 0.38], [0.35, 0.42], [0.36, 0.45], [0.34, 0.49], [0.35, 0.5], [0.35, 0.5]]}" + } + ], + "images": "12780_exterior_image_1_left#0", + "gt": "[[172.5, 83.5], [153.5, 66.5], [137.5, 64.5], [139.5, 65.5], [136.5, 67.5], [110.5, 75.5], [115.5, 80.5], [108.5, 87.5], [110.5, 90.5], [110.5, 90.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0101_user_0007_scene_0002_cfg_0002_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'cleaning the table with a cloth', and now you are performing subtask 'cleaning the table with a cloth.'\nBegin at [0.53, 0.3] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.3], [0.53, 0.39], [0.56, 0.48], [0.58, 0.54], [0.57, 0.54], [0.62, 0.64], [0.65, 0.49], [0.65, 0.4], [0.63, 0.38], [0.56, 0.44]]}" + } + ], + "images": "RH20T_cfg2_task_0101_user_0007_scene_0002_cfg_0002_104122061850#0", + "gt": "[[337, 109], [339, 142], [360, 174], [369, 195], [364, 196], [396, 230], [414, 175], [414, 143], [402, 135], [357, 157]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "46919_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the door of the washing machine and push the tray back in', and now you are performing subtask 'push the tray back into the washing machine.'\nYou start from [0.65, 0.49]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.49], [0.57, 0.59], [0.5, 0.91], [0.59, 1.0], [0.66, 0.99], [0.64, 0.9], [0.7, 0.69], [0.67, 0.67], [0.73, 0.68], [0.73, 0.68]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/46919_exterior_image_2_left#0.jpg", + "gt": "[[209, 89], [184, 107], [160, 163], [188, 180], [211, 179], [206, 162], [224, 124], [214, 120], [234, 122], [234, 122]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0041_user_0007_scene_0006_cfg_0005_104122063678#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'clean the tabletop with a sponge', and now you are performing subtask 'slide the sponge on the tabletop to clean the surface, grab the sponge on the table, let the sponge slide on the tabletop, place the sponge on the tabletop.'\nYou begin at [0.51, 0.62]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.62], [0.52, 0.65], [0.52, 0.61], [0.52, 0.59], [0.53, 0.66], [0.51, 0.6], [0.52, 0.57], [0.51, 0.58], [0.52, 0.62], [0.51, 0.63]]}" + } + ], + "images": "RH20T_cfg5_task_0041_user_0007_scene_0006_cfg_0005_104122063678#573", + "gt": "[[326, 223], [336, 233], [331, 220], [336, 214], [339, 237], [329, 215], [335, 205], [329, 208], [335, 223], [326, 228]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "69884_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the watermelon plush toy in the box', and now you are performing subtask 'put the watermelon plush toy in the box.'\nYour starting position is [0.72, 0.41]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.73, 0.41], [0.69, 0.54], [0.6, 0.72], [0.61, 0.79], [0.67, 0.78], [0.74, 0.61], [0.8, 0.59], [0.85, 0.63], [0.87, 0.58], [0.85, 0.51]]}" + } + ], + "images": "69884_exterior_image_2_left#0", + "gt": "[[232, 74], [221, 98], [193, 130], [196, 143], [214, 140], [237, 110], [255, 107], [273, 114], [277, 104], [272, 92]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "85214_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the spoon to place some of the contents in the bowl into the jug, then pour the contents in the jug into the bowl', and now you are performing subtask 'pour the contents from the jug back into the bowl.'\nYou begin at [0.54, 0.43]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.43], [0.66, 0.26], [0.65, 0.56], [0.66, 0.32], [0.59, 0.17], [0.65, 0.15], [0.7, 0.31], [0.67, 0.52], [0.56, 0.26], [0.59, 0.24]]}" + } + ], + "images": "85214_exterior_image_1_left#348", + "gt": "[[172, 77], [212, 46], [209, 100], [210, 57], [190, 31], [209, 27], [223, 56], [214, 94], [179, 47], [190, 43]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "20481_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn the knob on the toaster oven clockwise', and now you are performing subtask 'turn the knob on the toaster oven clockwise.'\nThe trajectory starts at [0.68, 0.42]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.42], [0.67, 0.32], [0.64, 0.33], [0.65, 0.39], [0.62, 0.4], [0.66, 0.41], [0.68, 0.3], [0.71, 0.28], [0.7, 0.31], [0.69, 0.28]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/20481_exterior_image_1_left#0.jpg", + "gt": "[[217.0, 76.0], [213.0, 58.0], [206.0, 60.0], [209.0, 70.0], [198.0, 72.0], [212.0, 73.0], [218.0, 54.0], [228.0, 51.0], [225.0, 56.0], [222.0, 51.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0034_user_0020_scene_0001_cfg_0006_104122063633#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'pick up the first square on the right, transfer the first square from the right to the center, place the first square in the center.'\nYour initial coordinate is [0.36, 0.33]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.33], [0.35, 0.41], [0.33, 0.39], [0.35, 0.36], [0.33, 0.35], [0.34, 0.4], [0.36, 0.38], [0.33, 0.43], [0.34, 0.47], [0.35, 0.38]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0034_user_0020_scene_0001_cfg_0006_104122063633#0.jpg", + "gt": "[[233, 118], [223, 148], [214, 140], [222, 128], [210, 127], [217, 145], [229, 135], [213, 153], [216, 171], [221, 135]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "44464_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the plush toy out of the box and put it on the black tray', and now you are performing subtask 'throw the green fruit plush toy on the black tray.'\nStarting at [0.39, 0.63], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.63], [0.45, 0.51], [0.47, 0.52], [0.5, 0.59], [0.44, 0.58], [0.41, 0.61], [0.39, 0.65], [0.39, 0.62], [0.42, 0.6], [0.41, 0.56]]}" + } + ], + "images": "44464_exterior_image_2_left#174", + "gt": "[[126, 114], [144, 92], [149, 93], [161, 107], [142, 105], [130, 110], [126, 117], [126, 112], [133, 108], [132, 101]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0014_user_0010_scene_0002_cfg_0004_038522062288#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'the cube on the scale is picked up and placed aside', and now you are performing subtask 'the cube on the scale is picked up and placed aside.'\nYour starting position is [0.36, 0.33]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.33], [0.33, 0.33], [0.37, 0.41], [0.38, 0.42], [0.38, 0.4], [0.36, 0.55], [0.48, 0.71], [0.43, 0.6], [0.35, 0.5], [0.34, 0.53]]}" + } + ], + "images": "RH20T_cfg4_task_0014_user_0010_scene_0002_cfg_0004_038522062288#0", + "gt": "[[232, 120], [211, 120], [236, 146], [246, 152], [246, 143], [230, 198], [306, 254], [274, 215], [221, 179], [220, 191]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "65111_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pour the contents from the blue cup and the orange cup into the plate', and now you are performing subtask 'pick up the orange cup on the table.'\nYou begin at [0.48, 0.24]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.24], [0.72, 0.52], [0.69, 0.57], [0.7, 0.66], [0.68, 0.34], [0.65, 0.47], [0.76, 0.42], [0.69, 0.64], [0.69, 0.53], [0.61, 0.46]]}" + } + ], + "images": "65111_exterior_image_2_left#0", + "gt": "[[154, 43], [230, 94], [220, 102], [223, 118], [218, 62], [207, 84], [244, 75], [221, 116], [222, 96], [196, 82]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122061330#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'place the block on the block board, pick up the block on the flat surface, transfer the block from the flat surface to the other blocks, place the block on the other blocks.'\nWith start location [0.53, 0.43], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.43], [0.6, 0.55], [0.57, 0.49], [0.48, 0.44], [0.46, 0.46], [0.45, 0.43], [0.44, 0.49], [0.43, 0.46], [0.46, 0.42], [0.5, 0.44]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122061330#452.jpg", + "gt": "[[342, 154], [386, 197], [366, 178], [310, 158], [297, 166], [285, 156], [282, 176], [274, 166], [293, 152], [323, 159]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "68845_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the marker out of the pot and put it on the table', and now you are performing subtask 'take the marker out of the pot and put it on the table.'\nStart at [0.59, 0.05]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.05], [0.53, 0.14], [0.46, 0.37], [0.45, 0.26], [0.46, 0.14], [0.47, 0.12], [0.48, 0.14], [0.48, 0.34], [0.48, 0.25], [0.47, 0.26]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/68845_exterior_image_2_left#0.jpg", + "gt": "[[189, 9], [169, 26], [147, 66], [144, 46], [147, 25], [151, 22], [154, 25], [153, 62], [153, 45], [150, 46]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0070_user_0014_scene_0008_cfg_0001_819112071037#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a sliding window', and now you are performing subtask 'open a sliding window.'\nWith start location [0.57, 0.0], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.0], [0.58, 0.2], [0.61, 0.55], [0.59, 0.73], [0.53, 0.7], [0.5, 0.64], [0.7, 0.2], [0.75, 0.03], [0.73, 0.04], [0.73, 0.08]]}" + } + ], + "images": "RH20T_cfg1_task_0070_user_0014_scene_0008_cfg_0001_819112071037#0", + "gt": "[[364, 0], [369, 73], [393, 199], [375, 263], [341, 251], [317, 231], [446, 73], [479, 12], [469, 13], [468, 28]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0037_user_0007_scene_0007_cfg_0002_104422070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the first block from the surface, place the first block on the surface.'\nStart at [0.49, 0.53]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.52], [0.5, 0.51], [0.5, 0.54], [0.51, 0.55], [0.51, 0.51], [0.52, 0.53], [0.51, 0.53], [0.52, 0.55], [0.52, 0.56], [0.51, 0.56]]}" + } + ], + "images": "RH20T_cfg2_task_0037_user_0007_scene_0007_cfg_0002_104422070044#0", + "gt": "[[311, 189], [317, 183], [318, 193], [324, 199], [328, 182], [331, 192], [325, 192], [331, 197], [330, 201], [326, 201]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0016_user_0010_scene_0001_cfg_0005_105422061350#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'hit the pool ball', and now you are performing subtask 'transfer the pool cue from the table to beside the pool ball, strike the pool ball with the pool cue, place the pool cue on the pool table.'\nYou begin at [0.45, 0.33]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.32], [0.43, 0.29], [0.4, 0.27], [0.39, 0.28], [0.43, 0.33], [0.42, 0.32], [0.43, 0.32], [0.4, 0.22], [0.42, 0.23], [0.42, 0.19]]}" + } + ], + "images": "RH20T_cfg5_task_0016_user_0010_scene_0001_cfg_0005_105422061350#294", + "gt": "[[285, 117], [278, 105], [254, 96], [252, 100], [276, 120], [268, 117], [276, 117], [259, 80], [271, 83], [268, 68]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_039422060546#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the second button on the desk, press the third button on the desk.'\nYour initial coordinate is [0.37, 0.69]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.69], [0.43, 0.62], [0.51, 0.44], [0.44, 0.44], [0.46, 0.42], [0.44, 0.42], [0.44, 0.39], [0.47, 0.44], [0.45, 0.38], [0.46, 0.31]]}" + } + ], + "images": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_039422060546#126", + "gt": "[[235, 249], [275, 222], [325, 160], [280, 159], [294, 150], [283, 152], [282, 141], [301, 159], [289, 135], [295, 113]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "361_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the black masking tape on the plate', and now you are performing subtask 'put the black masking tape on the plate.'\nThe trajectory starts at [0.29, 0.03]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.29, 0.03], [0.42, 0.07], [0.53, 0.14], [0.52, 0.16], [0.53, 0.16], [0.57, 0.36], [0.47, 0.24], [0.34, 0.33], [0.34, 0.41], [0.31, 0.33]]}" + } + ], + "images": "361_exterior_image_1_left#0", + "gt": "[[94.0, 5.0], [135.0, 12.0], [171.0, 26.0], [165.0, 29.0], [171.0, 29.0], [183.0, 65.0], [149.0, 44.0], [110.0, 60.0], [108.0, 73.0], [100.0, 59.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0096_user_0007_scene_0008_cfg_0005_104422070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take something out of a drawer', and now you are performing subtask 'pull the drawer handle to the outside of the table.'\nYou begin at [0.58, 0.37]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.37], [0.58, 0.34], [0.63, 0.29], [0.71, 0.36], [0.68, 0.32], [0.73, 0.36], [0.72, 0.46], [0.68, 0.43], [0.69, 0.41], [0.66, 0.39]]}" + } + ], + "images": "RH20T_cfg5_task_0096_user_0007_scene_0008_cfg_0005_104422070044#0", + "gt": "[[374, 132], [374, 124], [406, 104], [454, 130], [432, 117], [464, 128], [461, 165], [438, 153], [439, 149], [421, 142]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0038_user_0010_scene_0005_cfg_0003_038522062288#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the cup', and now you are performing subtask 'pick up the cup.'\nThe trajectory starts at [0.37, 0.36]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.36], [0.36, 0.39], [0.41, 0.38], [0.36, 0.38], [0.36, 0.59], [0.43, 0.68], [0.35, 0.57], [0.34, 0.48], [0.36, 0.45], [0.34, 0.47]]}" + } + ], + "images": "RH20T_cfg3_task_0038_user_0010_scene_0005_cfg_0003_038522062288#0", + "gt": "[[238, 128], [233, 141], [262, 135], [229, 135], [230, 214], [273, 244], [226, 207], [220, 174], [230, 161], [217, 169]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "2790_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the yellow block inside the blue cup', and now you are performing subtask 'place the yellow block inside the blue cup.'\nYou start from [0.32, 0.22]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.22], [0.4, 0.32], [0.43, 0.44], [0.41, 0.49], [0.44, 0.49], [0.48, 0.39], [0.55, 0.34], [0.6, 0.37], [0.62, 0.46], [0.51, 0.41]]}" + } + ], + "images": "2790_exterior_image_2_left#0", + "gt": "[[102, 39], [129, 57], [139, 79], [132, 88], [140, 88], [153, 71], [177, 62], [192, 67], [198, 83], [163, 74]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "41945_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'throw the bottle cap on the table', and now you are performing subtask 'throw the bottle cap on the table.'\nFrom start point [0.68, 0.24], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.24], [0.67, 0.24], [0.62, 0.32], [0.62, 0.33], [0.62, 0.33], [0.56, 0.38], [0.54, 0.42], [0.5, 0.47], [0.49, 0.52], [0.49, 0.52]]}" + } + ], + "images": "41945_exterior_image_1_left#0", + "gt": "[[218, 43], [215, 43], [197, 57], [198, 60], [197, 59], [178, 68], [172, 76], [161, 85], [158, 94], [158, 94]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0123_user_0007_scene_0007_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the knife on the knife rack', and now you are performing subtask 'put the knife on the knife rack.'\nYour starting position is [0.33, 0.33]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.33, 0.33], [0.36, 0.31], [0.39, 0.32], [0.43, 0.41], [0.4, 0.5], [0.41, 0.4], [0.4, 0.54], [0.37, 0.68], [0.37, 0.59], [0.33, 0.37]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0123_user_0007_scene_0007_cfg_0005_104122063678#0.jpg", + "gt": "[[213, 120], [232, 112], [252, 115], [274, 148], [257, 179], [260, 144], [259, 196], [234, 244], [237, 214], [210, 132]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0020_user_0014_scene_0001_cfg_0006_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the desk lamp by pressing the button', and now you are performing subtask 'turn on the desk lamp by pressing the button.'\nBegin at [0.53, 0.31] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.31], [0.53, 0.29], [0.52, 0.26], [0.52, 0.24], [0.51, 0.27], [0.49, 0.24], [0.49, 0.27], [0.51, 0.29], [0.52, 0.29], [0.52, 0.29]]}" + } + ], + "images": "RH20T_cfg6_task_0020_user_0014_scene_0001_cfg_0006_104122064161#0", + "gt": "[[338, 110], [341, 103], [330, 92], [334, 87], [326, 96], [312, 86], [316, 97], [328, 105], [330, 103], [330, 103]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0032_user_0020_scene_0009_cfg_0006_104122060811#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'slide the outer casing onto the gift box', and now you are performing subtask 'slide the outer casing onto the gift box.'\nStarting at [0.6, 0.35], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.35], [0.6, 0.35], [0.61, 0.36], [0.62, 0.35], [0.65, 0.32], [0.67, 0.32], [0.68, 0.31], [0.67, 0.34], [0.63, 0.35], [0.62, 0.34]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0032_user_0020_scene_0009_cfg_0006_104122060811#0.jpg", + "gt": "[[385, 127], [387, 125], [393, 128], [397, 127], [417, 115], [426, 117], [435, 113], [428, 122], [404, 127], [399, 124]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "27099_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the white cloth on the black rod', and now you are performing subtask 'put the white cloth on the black rod.'\nYour initial coordinate is [0.53, 0.22]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.22], [0.59, 0.57], [0.56, 0.56], [0.56, 0.69], [0.46, 0.15], [0.46, 0.26], [0.39, 0.47], [0.43, 0.56], [0.42, 0.62], [0.47, 0.59]]}" + } + ], + "images": "27099_exterior_image_1_left#0", + "gt": "[[171.0, 40.0], [189.0, 103.0], [180.0, 100.0], [178.0, 124.0], [148.0, 27.0], [146.0, 46.0], [126.0, 85.0], [137.0, 100.0], [133.0, 111.0], [152.0, 106.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "53857_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the turquoise toy inside the pot in the sink', and now you are performing subtask 'put the turquoise toy inside the pot in the sink.'\nBegin at [0.17, 0.17] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.17, 0.17], [0.36, 0.32], [0.39, 0.2], [0.45, 0.22], [0.48, 0.32], [0.42, 0.36], [0.46, 0.51], [0.66, 0.61], [0.62, 0.67], [0.58, 0.57]]}" + } + ], + "images": "53857_exterior_image_1_left#0", + "gt": "[[53, 30], [115, 58], [124, 36], [143, 40], [155, 57], [134, 65], [147, 91], [211, 110], [200, 120], [187, 102]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0034_user_0016_scene_0001_cfg_0004_104122062823#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'place the square on top of the square pile in the right center position, pick up the third square, transfer the square from the square pile to the top, place the third square in the center position of the square pile.'\nYour initial coordinate is [0.57, 0.11]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.11], [0.52, 0.0], [0.47, 0.0], [0.44, 0.04], [0.42, 0.11], [0.41, 0.12], [0.42, 0.15], [0.43, 0.08], [0.47, 0.0], [0.55, 0.0]]}" + } + ], + "images": "RH20T_cfg4_task_0034_user_0016_scene_0001_cfg_0004_104122062823#392", + "gt": "[[366, 39], [335, 0], [300, 0], [280, 16], [271, 40], [264, 45], [271, 53], [274, 28], [304, 0], [349, 0]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0034_user_0014_scene_0007_cfg_0007_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'pick up the first block from the desktop, transfer the first block from the desktop to the right side of the table, place the first block on the right side of the table.'\nYour initial coordinate is [0.53, 0.32]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.32], [0.62, 0.47], [0.62, 0.54], [0.61, 0.56], [0.65, 0.59], [0.6, 0.7], [0.55, 0.71], [0.54, 0.77], [0.52, 0.81], [0.58, 0.73]]}" + } + ], + "images": "RH20T_cfg7_task_0034_user_0014_scene_0007_cfg_0007_104122064161#0", + "gt": "[[337, 115], [399, 169], [399, 196], [391, 202], [415, 212], [387, 253], [352, 255], [348, 278], [334, 293], [372, 263]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0016_user_0014_scene_0001_cfg_0001_819112070044#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'hit the pool ball', and now you are performing subtask 'pick up the pool ball from the surface.'\nBegin at [0.38, 0.7] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.7], [0.39, 0.68], [0.39, 0.67], [0.38, 0.68], [0.38, 0.66], [0.39, 0.69], [0.37, 0.72], [0.38, 0.73], [0.45, 0.71], [0.44, 0.7]]}" + } + ], + "images": "RH20T_cfg1_task_0016_user_0014_scene_0001_cfg_0001_819112070044#231", + "gt": "[[246, 252], [249, 246], [247, 242], [245, 244], [246, 237], [248, 247], [239, 259], [245, 264], [285, 256], [284, 251]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "81663_exterior_image_2_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the cup and put it on top of the table, then pick up the cloth and put it in the bowl and close the drawer', and now you are performing subtask 'push the handle of the drawer to close the drawer.'\nStart at [0.52, 0.65]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.65], [0.5, 0.66], [0.51, 0.64], [0.51, 0.66], [0.52, 0.63], [0.52, 0.66], [0.44, 0.78], [0.44, 0.82], [0.44, 0.73], [0.43, 0.69]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/81663_exterior_image_2_left#277.jpg", + "gt": "[[167, 117], [160, 118], [164, 115], [164, 118], [166, 114], [165, 118], [142, 140], [142, 148], [142, 131], [139, 125]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0037_user_0007_scene_0004_cfg_0005_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the first block from the surface, place the first block on the vertical stack on the surface, pick up the second block from the surface.'\nStart at [0.48, 0.14]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.14], [0.48, 0.14], [0.5, 0.19], [0.5, 0.31], [0.51, 0.41], [0.5, 0.48], [0.5, 0.54], [0.5, 0.31], [0.52, 0.25], [0.53, 0.31]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0037_user_0007_scene_0004_cfg_0005_104122061850#0.jpg", + "gt": "[[305, 50], [308, 49], [317, 67], [318, 113], [328, 148], [318, 174], [323, 193], [321, 112], [331, 90], [341, 113]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0003_user_0010_scene_0001_cfg_0005_036422060215#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the red button in the middle.'\nStart at [0.46, 0.34]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.34], [0.44, 0.38], [0.42, 0.38], [0.41, 0.41], [0.42, 0.37], [0.42, 0.36], [0.41, 0.35], [0.4, 0.36], [0.4, 0.37], [0.35, 0.4]]}" + } + ], + "images": "RH20T_cfg5_task_0003_user_0010_scene_0001_cfg_0005_036422060215#224", + "gt": "[[294, 123], [281, 135], [269, 138], [265, 148], [269, 133], [267, 128], [261, 126], [253, 131], [253, 133], [227, 143]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0034_user_0020_scene_0001_cfg_0006_104122063633#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'place the third square above the square, pick up the fourth square on the top, transfer the fourth square from the top to above the third square, place the fourth square above the third square, pick up the fifth square on the upper right.'\nYour initial coordinate is [0.34, 0.51]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.34, 0.51], [0.24, 0.7], [0.2, 0.81], [0.21, 0.87], [0.25, 0.6], [0.29, 0.5], [0.34, 0.48], [0.34, 0.42], [0.31, 0.41], [0.29, 0.52]]}" + } + ], + "images": "RH20T_cfg6_task_0034_user_0020_scene_0001_cfg_0006_104122063633#1402", + "gt": "[[220, 182], [151, 251], [129, 290], [132, 313], [158, 216], [184, 180], [216, 173], [216, 152], [197, 147], [188, 186]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0073_user_0007_scene_0001_cfg_0005_104122063678#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the things in the drawer', and now you are performing subtask 'grab the handle to pull the drawer open.'\nWith start location [0.52, 0.54], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.54], [0.52, 0.55], [0.51, 0.57], [0.51, 0.57], [0.5, 0.52], [0.49, 0.5], [0.5, 0.49], [0.5, 0.48], [0.5, 0.49], [0.51, 0.47]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0073_user_0007_scene_0001_cfg_0005_104122063678#423.jpg", + "gt": "[[331, 195], [333, 198], [328, 204], [328, 206], [318, 189], [315, 181], [317, 177], [322, 174], [322, 175], [328, 169]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0026_user_0010_scene_0001_cfg_0003_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the gripper to push and close the drawer', and now you are performing subtask 'use the gripper to push and close the drawer.'\nYou start from [0.3, 0.08]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.3, 0.08], [0.34, 0.14], [0.32, 0.18], [0.32, 0.24], [0.3, 0.26], [0.14, 0.14], [0.21, 0.14], [0.25, 0.1], [0.28, 0.03], [0.28, 0.03]]}" + } + ], + "images": "RH20T_cfg3_task_0026_user_0010_scene_0001_cfg_0003_104122062823#0", + "gt": "[[191, 30], [220, 52], [205, 64], [203, 86], [193, 94], [92, 50], [132, 50], [158, 36], [176, 9], [176, 9]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "18425_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'switch on the toaster', and now you are performing subtask 'switch on the toaster.'\nYou start from [0.28, 0.21]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.28, 0.21], [0.51, 0.16], [0.62, 0.2], [0.67, 0.32], [0.7, 0.63], [0.69, 0.57], [0.7, 0.63], [0.63, 0.51], [0.56, 0.31], [0.5, 0.28]]}" + } + ], + "images": "18425_exterior_image_1_left#0", + "gt": "[[90.0, 37.0], [163.0, 28.0], [198.0, 36.0], [214.0, 57.0], [225.0, 113.0], [220.0, 102.0], [225.0, 113.0], [202.0, 92.0], [179.0, 55.0], [160.0, 50.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "1087_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the purple and orange plushies from the box', and now you are performing subtask 'throw the purple plush toy in the plate.'\nThe trajectory starts at [0.42, 0.26]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.26], [0.43, 0.41], [0.5, 0.6], [0.5, 0.66], [0.49, 0.62], [0.51, 0.58], [0.62, 0.61], [0.66, 0.61], [0.66, 0.58], [0.64, 0.58]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/1087_exterior_image_2_left#0.jpg", + "gt": "[[133, 47], [138, 73], [161, 108], [160, 118], [158, 111], [164, 104], [197, 110], [211, 109], [212, 104], [204, 104]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0023_user_0010_scene_0007_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the power strip by pressing the button', and now you are performing subtask 'turn on the power strip by pressing the button.'\nStart at [0.5, 0.37]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.45, 0.45], [0.44, 0.41], [0.43, 0.44], [0.45, 0.46], [0.4, 0.51], [0.44, 0.42], [0.48, 0.4], [0.52, 0.42], [0.51, 0.43]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0023_user_0010_scene_0007_cfg_0005_104122063678#0.jpg", + "gt": "[[323, 132], [285, 163], [279, 146], [277, 159], [288, 167], [259, 183], [282, 151], [307, 145], [333, 150], [325, 155]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "34432_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'get one of the bottles from the bowl and set it upright on the purple paper', and now you are performing subtask 'get one of the bottles from the bowl and set it upright on the purple paper.'\nWith start location [0.76, 0.03], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.76, 0.03], [0.63, 0.17], [0.62, 0.28], [0.55, 0.35], [0.55, 0.43], [0.61, 0.3], [0.62, 0.34], [0.62, 0.41], [0.65, 0.42], [0.66, 0.3]]}" + } + ], + "images": "34432_exterior_image_1_left#0", + "gt": "[[242.0, 4.5], [203.0, 30.5], [198.0, 50.5], [177.0, 63.5], [177.0, 77.5], [194.0, 53.5], [200.0, 61.5], [197.0, 74.5], [209.0, 75.5], [211.0, 54.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0033_user_0015_scene_0008_cfg_0001_039422060546#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the block', and now you are performing subtask 'push the square in front of the cardboard to the adjacent square in the middle of the cardboard.'\nThe trajectory starts at [0.41, 0.34]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.34], [0.46, 0.39], [0.58, 0.47], [0.6, 0.51], [0.61, 0.51], [0.6, 0.51], [0.6, 0.49], [0.56, 0.5], [0.5, 0.47], [0.5, 0.46]]}" + } + ], + "images": "RH20T_cfg1_task_0033_user_0015_scene_0008_cfg_0001_039422060546#0", + "gt": "[[265, 124], [296, 141], [372, 169], [382, 183], [390, 182], [385, 182], [384, 176], [358, 181], [317, 170], [320, 166]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0057_user_0007_scene_0003_cfg_0005_104122063678#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn on the power strips', and now you are performing subtask 'press the second button on the power socket, press the third button on the power socket.'\nStart at [0.46, 0.56]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.56], [0.46, 0.56], [0.46, 0.57], [0.46, 0.57], [0.46, 0.57], [0.46, 0.56], [0.45, 0.56], [0.46, 0.53], [0.46, 0.53], [0.47, 0.53]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0057_user_0007_scene_0003_cfg_0005_104122063678#423.jpg", + "gt": "[[297, 201], [296, 203], [297, 205], [295, 206], [296, 205], [292, 203], [291, 200], [293, 191], [297, 190], [304, 192]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "87769_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the green packaging bag from the sink and put it on the countertop', and now you are performing subtask 'remove the green packaging bag from the sink and put it on the countertop.'\nYou begin at [0.23, 0.44]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.23, 0.44], [0.37, 0.49], [0.5, 0.76], [0.5, 0.89], [0.51, 0.82], [0.48, 0.74], [0.37, 0.76], [0.39, 0.71], [0.32, 0.74], [0.32, 0.68]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/87769_exterior_image_1_left#0.jpg", + "gt": "[[75, 79], [118, 89], [160, 137], [159, 161], [162, 147], [153, 134], [117, 137], [124, 127], [103, 133], [102, 122]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "57652_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'hang up the cloth on the glass barrier', and now you are performing subtask 'throw the cloth on the table.'\nStart at [0.43, 0.18]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.18], [0.43, 0.4], [0.62, 0.67], [0.61, 0.76], [0.52, 0.41], [0.6, 0.59], [0.61, 0.74], [0.52, 0.28], [0.63, 0.58], [0.63, 0.68]]}" + } + ], + "images": "57652_exterior_image_1_left#0", + "gt": "[[137, 33], [139, 72], [199, 120], [195, 137], [166, 74], [193, 107], [196, 134], [168, 51], [201, 105], [203, 122]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0013_user_0020_scene_0004_cfg_0007_104122061602#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the block on the scale', and now you are performing subtask 'place the block on the scale.'\nYou start from [0.7, 0.32]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.32], [0.69, 0.37], [0.66, 0.35], [0.67, 0.43], [0.72, 0.45], [0.8, 0.61], [0.78, 0.64], [0.8, 0.54], [0.76, 0.45], [0.71, 0.42]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0013_user_0020_scene_0004_cfg_0007_104122061602#0.jpg", + "gt": "[[448, 116], [441, 133], [424, 127], [428, 154], [463, 163], [511, 218], [501, 229], [509, 196], [489, 162], [456, 152]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0130_user_0020_scene_0002_cfg_0006_104122064161#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb', and now you are performing subtask 'insert the bulb into the socket, press the switch on the socket to turn on the bulb, press the bulb switch to turn on the light.'\nWith start location [0.63, 0.41], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.41], [0.62, 0.42], [0.62, 0.43], [0.63, 0.46], [0.63, 0.47], [0.62, 0.48], [0.62, 0.47], [0.64, 0.4], [0.64, 0.41], [0.64, 0.43]]}" + } + ], + "images": "RH20T_cfg6_task_0130_user_0020_scene_0002_cfg_0006_104122064161#459", + "gt": "[[403, 148], [400, 150], [398, 156], [401, 164], [401, 169], [399, 173], [397, 171], [407, 144], [407, 147], [410, 153]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0002_user_0010_scene_0007_cfg_0004_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull out a napkin', and now you are performing subtask 'pull out a napkin.'\nYou begin at [0.38, 0.0]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.25, 0.0], [0.2, 0.15], [0.16, 0.12], [0.18, 0.12], [0.19, 0.0], [0.31, 0.0], [0.12, 0.0], [0.39, 0.0], [0.39, 0.0]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0002_user_0010_scene_0007_cfg_0004_104122062823#0.jpg", + "gt": "[[243, 0], [161, 0], [127, 53], [100, 42], [117, 42], [119, 0], [198, 0], [78, 0], [247, 0], [247, 0]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "81186_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the lid from the pot and put it on the counter', and now you are performing subtask 'place the lid on the countertop.'\nBegin at [0.1, 0.94] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.1, 0.94], [0.04, 0.82], [0.08, 0.74], [0.12, 0.73], [0.11, 0.82], [0.17, 0.79], [0.25, 0.71], [0.38, 0.75], [0.49, 0.98], [0.53, 0.78]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/81186_exterior_image_1_left#107.jpg", + "gt": "[[33, 170], [12, 147], [24, 134], [40, 132], [34, 147], [55, 143], [79, 128], [120, 135], [158, 176], [169, 140]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0091_user_0007_scene_0004_cfg_0005_105422061350#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move an object from one box to another', and now you are performing subtask 'pick up the block from the pink box, transfer the block from the pink box to the table, throw the block onto the table.'\nStart at [0.43, 0.13]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.13], [0.41, 0.06], [0.39, 0.02], [0.37, 0.01], [0.32, 0.11], [0.33, 0.17], [0.32, 0.1], [0.32, 0.04], [0.33, 0.07], [0.33, 0.07]]}" + } + ], + "images": "RH20T_cfg5_task_0091_user_0007_scene_0004_cfg_0005_105422061350#0", + "gt": "[[276, 47], [264, 22], [250, 7], [237, 4], [208, 38], [211, 61], [206, 35], [207, 15], [211, 25], [211, 25]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "8448_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the marker and put it in the bowl', and now you are performing subtask 'pick up the marker and put it in the bowl.'\nWith start location [0.48, 0.48], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.48], [0.43, 0.4], [0.37, 0.41], [0.34, 0.48], [0.32, 0.72], [0.33, 0.46], [0.35, 0.44], [0.36, 0.49], [0.37, 0.43], [0.39, 0.4]]}" + } + ], + "images": "8448_exterior_image_2_left#0", + "gt": "[[155, 87], [136, 72], [119, 73], [110, 86], [104, 129], [107, 83], [112, 79], [115, 89], [117, 77], [126, 72]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "51625_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the second switch from the main switch on the adapter', and now you are performing subtask 'turn on the second switch from the main switch on the adapter.'\nYou start from [0.78, 0.01]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.78, 0.01], [0.74, 0.18], [0.75, 0.27], [0.78, 0.32], [0.77, 0.37], [0.8, 0.34], [0.8, 0.37], [0.82, 0.38], [0.78, 0.21], [0.78, 0.21]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/51625_exterior_image_1_left#0.jpg", + "gt": "[[251, 2], [236, 32], [239, 48], [251, 57], [247, 66], [255, 61], [255, 66], [264, 69], [251, 37], [251, 37]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "68725_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the plastic box on the table', and now you are performing subtask 'close the plastic box on the table.'\nBegin at [0.39, 0.02] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.02], [0.32, 0.22], [0.24, 0.68], [0.2, 0.72], [0.15, 0.55], [0.17, 0.41], [0.28, 0.29], [0.37, 0.28], [0.42, 0.37], [0.42, 0.37]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/68725_exterior_image_1_left#0.jpg", + "gt": "[[126, 3], [101, 40], [76, 123], [64, 129], [47, 99], [54, 74], [88, 53], [117, 51], [135, 67], [135, 67]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0091_user_0007_scene_0004_cfg_0005_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move an object from one box to another', and now you are performing subtask 'pick up the block from the pink box, transfer the block from the pink box to the table, throw the block onto the table.'\nBegin at [0.48, 0.14] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.14], [0.49, 0.2], [0.5, 0.22], [0.51, 0.24], [0.54, 0.4], [0.58, 0.48], [0.58, 0.51], [0.58, 0.54], [0.58, 0.47], [0.57, 0.27]]}" + } + ], + "images": "RH20T_cfg5_task_0091_user_0007_scene_0004_cfg_0005_104122061850#0", + "gt": "[[305, 50], [312, 73], [323, 79], [327, 86], [345, 143], [371, 172], [374, 185], [372, 195], [374, 168], [362, 98]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "20992_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the candy bar in the sink', and now you are performing subtask 'put the candy bar in the sink.'\nStarting at [0.44, 0.44], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.44], [0.33, 0.54], [0.26, 0.73], [0.28, 0.78], [0.35, 0.68], [0.45, 0.68], [0.47, 0.88], [0.47, 0.82], [0.4, 0.78], [0.39, 0.73]]}" + } + ], + "images": "20992_exterior_image_1_left#0", + "gt": "[[142, 79], [106, 98], [84, 131], [91, 141], [113, 122], [145, 123], [150, 159], [150, 147], [129, 140], [124, 132]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "89315_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the jumper on the armchair seat', and now you are performing subtask 'place the jumper on the seat of the armchair.'\nFrom start point [0.29, 0.47], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.29, 0.47], [0.47, 0.43], [0.5, 0.52], [0.47, 0.62], [0.51, 0.66], [0.5, 0.44], [0.57, 0.16], [0.57, 0.0], [0.62, 0.13], [0.55, 0.36]]}" + } + ], + "images": "89315_exterior_image_2_left#0", + "gt": "[[92, 84], [151, 78], [159, 94], [150, 112], [164, 118], [161, 80], [183, 29], [182, 0], [197, 23], [177, 65]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0044_user_0010_scene_0001_cfg_0005_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up a bag of things', and now you are performing subtask 'pick up a bag of things.'\nFrom start point [0.45, 0.29], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.29], [0.46, 0.25], [0.45, 0.21], [0.5, 0.12], [0.52, 0.24], [0.51, 0.28], [0.53, 0.08], [0.48, 0.19], [0.38, 0.29], [0.43, 0.22]]}" + } + ], + "images": "RH20T_cfg5_task_0044_user_0010_scene_0001_cfg_0005_036422060215#0", + "gt": "[[286, 106], [295, 91], [290, 74], [319, 43], [331, 88], [326, 99], [340, 27], [308, 67], [240, 105], [278, 79]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0047_user_0014_scene_0014_cfg_0004_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the cup off the cup rack', and now you are performing subtask 'take the cup off the cup rack.'\nStart at [0.45, 0.0]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.0], [0.44, 0.03], [0.45, 0.0], [0.35, 0.0], [0.51, 0.0], [0.51, 0.09], [0.53, 0.16], [0.55, 0.14], [0.52, 0.02], [0.45, 0.0]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0047_user_0014_scene_0014_cfg_0004_104122062823#0.jpg", + "gt": "[[289, 0], [284, 11], [289, 0], [224, 0], [328, 0], [324, 32], [339, 59], [349, 51], [335, 8], [291, 0]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "82876_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn off the toaster', and now you are performing subtask 'turn off the toaster.'\nYou start from [0.42, 0.16]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.16], [0.59, 0.65], [0.61, 0.68], [0.63, 0.77], [0.63, 0.75], [0.63, 0.73], [0.62, 0.75], [0.57, 0.73], [0.53, 0.67], [0.52, 0.62]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/82876_exterior_image_1_left#0.jpg", + "gt": "[[135, 29], [188, 117], [196, 122], [201, 138], [203, 135], [201, 132], [197, 135], [182, 132], [170, 121], [166, 111]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0065_user_0007_scene_0003_cfg_0002_104122063678#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with small Lego blocks for the nth time with n blocks', and now you are performing subtask 'place the black Lego block on the Lego base, pick up the green Lego block from the table, transfer the green Lego block from the table to the Lego base, place the green Lego block on the Lego base.'\nFrom start point [0.74, 0.75], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.74, 0.75], [0.73, 0.72], [0.75, 0.75], [0.77, 0.71], [0.78, 0.57], [0.81, 0.55], [0.81, 0.58], [0.74, 0.59], [0.75, 0.65], [0.76, 0.61]]}" + } + ], + "images": "RH20T_cfg2_task_0065_user_0007_scene_0003_cfg_0002_104122063678#640", + "gt": "[[476, 269], [469, 260], [477, 269], [490, 255], [499, 207], [516, 199], [517, 209], [473, 211], [479, 235], [484, 219]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0032_user_0016_scene_0002_cfg_0004_104122062823#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'slide the outer casing onto the gift box', and now you are performing subtask 'slide the outer casing onto the gift box.'\nWith start location [0.45, 0.14], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.14], [0.44, 0.14], [0.43, 0.14], [0.42, 0.14], [0.42, 0.15], [0.42, 0.15], [0.44, 0.15], [0.45, 0.14], [0.46, 0.11], [0.46, 0.12]]}" + } + ], + "images": "RH20T_cfg4_task_0032_user_0016_scene_0002_cfg_0004_104122062823#164", + "gt": "[[285, 52], [284, 51], [278, 51], [270, 52], [269, 53], [270, 54], [280, 53], [287, 49], [297, 41], [292, 43]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0003_user_0016_scene_0006_cfg_0003_104122062823#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the red button in the middle, press the orange button on the left.'\nYou start from [0.43, 0.16]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.16], [0.39, 0.09], [0.36, 0.08], [0.33, 0.04], [0.31, 0.04], [0.32, 0.08], [0.3, 0.15], [0.29, 0.13], [0.3, 0.15], [0.28, 0.11]]}" + } + ], + "images": "RH20T_cfg3_task_0003_user_0016_scene_0006_cfg_0003_104122062823#210", + "gt": "[[273, 56], [251, 32], [230, 29], [210, 14], [200, 15], [207, 28], [190, 53], [187, 47], [192, 55], [180, 39]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0014_user_0014_scene_0001_cfg_0001_819112070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the block from the scale', and now you are performing subtask 'remove the block from the scale.'\nYou start from [0.62, 0.46]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.46], [0.6, 0.46], [0.59, 0.48], [0.54, 0.51], [0.54, 0.53], [0.53, 0.59], [0.54, 0.56], [0.57, 0.53], [0.59, 0.52], [0.62, 0.53]]}" + } + ], + "images": "RH20T_cfg1_task_0014_user_0014_scene_0001_cfg_0001_819112070044#0", + "gt": "[[400, 167], [384, 167], [378, 172], [343, 185], [345, 192], [338, 214], [344, 201], [368, 190], [379, 188], [400, 190]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "13889_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'shift the garment to the right then use the spoon to stir the contents of the bowl', and now you are performing subtask 'place the spoon on the table.'\nFrom start point [0.51, 0.74], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.74], [0.41, 0.58], [0.38, 0.74], [0.41, 0.78], [0.37, 0.68], [0.42, 0.74], [0.39, 0.7], [0.43, 0.62], [0.51, 0.71], [0.48, 0.63]]}" + } + ], + "images": "13889_exterior_image_1_left#156", + "gt": "[[164, 133], [131, 104], [121, 134], [130, 140], [119, 123], [133, 133], [124, 126], [137, 112], [163, 128], [155, 114]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "61794_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the mug with the marker to the left', and now you are performing subtask 'pull the bowl to the left front of the table.'\nStart at [0.51, 0.41]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.41], [0.5, 0.44], [0.52, 0.57], [0.5, 0.6], [0.5, 0.67], [0.51, 0.59], [0.5, 0.49], [0.49, 0.48], [0.49, 0.53], [0.43, 0.51]]}" + } + ], + "images": "61794_exterior_image_1_left#98", + "gt": "[[162, 73], [159, 80], [165, 103], [159, 108], [161, 121], [163, 106], [160, 88], [157, 87], [156, 96], [138, 91]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "73393_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the spoon to above the table', and now you are performing subtask 'move the spoon to above the table.'\nYou start from [0.57, 0.12]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.12], [0.57, 0.31], [0.54, 0.35], [0.54, 0.46], [0.52, 0.5], [0.56, 0.6], [0.57, 0.61], [0.59, 0.48], [0.57, 0.21], [0.61, 0.18]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/73393_exterior_image_2_left#0.jpg", + "gt": "[[181, 21], [184, 55], [174, 63], [172, 83], [165, 90], [178, 108], [183, 109], [189, 86], [181, 38], [194, 33]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "36682_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the brown object in the drawer, then put in the white bowl, then put the knife in the drawer', and now you are performing subtask 'place the brown object in the white bowl.'\nWith start location [0.66, 0.2], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.66, 0.2], [0.67, 0.11], [0.65, 0.2], [0.68, 0.19], [0.7, 0.21], [0.68, 0.28], [0.58, 0.22], [0.55, 0.24], [0.55, 0.28], [0.57, 0.27]]}" + } + ], + "images": "36682_exterior_image_1_left#638", + "gt": "[[212.5, 35.5], [213.5, 20.5], [209.5, 36.5], [217.5, 33.5], [224.5, 38.5], [216.5, 50.5], [185.5, 40.5], [176.5, 43.5], [175.5, 49.5], [181.5, 48.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0094_user_0007_scene_0006_cfg_0005_104422070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'press the part of the box, press the part of the box.'\nYour starting position is [0.58, 0.37]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.37], [0.62, 0.37], [0.62, 0.38], [0.64, 0.41], [0.64, 0.4], [0.64, 0.38], [0.62, 0.37], [0.61, 0.39], [0.62, 0.37], [0.66, 0.39]]}" + } + ], + "images": "RH20T_cfg5_task_0094_user_0007_scene_0006_cfg_0005_104422070044#0", + "gt": "[[374, 132], [395, 132], [400, 135], [409, 147], [412, 144], [410, 138], [395, 133], [393, 139], [394, 132], [422, 142]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0032_user_0020_scene_0009_cfg_0006_104122061018#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'slide the outer casing onto the gift box', and now you are performing subtask 'slide the outer casing onto the gift box.'\nWith start location [0.57, 0.08], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.08], [0.49, 0.02], [0.39, 0.04], [0.38, 0.16], [0.4, 0.22], [0.43, 0.23], [0.4, 0.09], [0.45, 0.07], [0.47, 0.0], [0.47, 0.0]]}" + } + ], + "images": "RH20T_cfg6_task_0032_user_0020_scene_0009_cfg_0006_104122061018#0", + "gt": "[[367, 28], [314, 6], [248, 14], [245, 59], [258, 78], [276, 83], [258, 32], [286, 26], [298, 0], [298, 0]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0045_user_0010_scene_0004_cfg_0005_104422070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the brush on the pen rack', and now you are performing subtask 'place the brush on the pen rack.'\nYour initial coordinate is [0.59, 0.4]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.4], [0.63, 0.44], [0.64, 0.49], [0.65, 0.44], [0.63, 0.44], [0.63, 0.35], [0.61, 0.32], [0.67, 0.34], [0.6, 0.29], [0.6, 0.29]]}" + } + ], + "images": "RH20T_cfg5_task_0045_user_0010_scene_0004_cfg_0005_104422070044#0", + "gt": "[[375, 144], [406, 157], [412, 175], [414, 160], [405, 157], [402, 126], [393, 117], [429, 124], [382, 105], [382, 105]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "10601_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the mug cup to the left', and now you are performing subtask 'move the mug cup to the left.'\nFrom start point [0.45, 0.17], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.17], [0.63, 0.12], [0.62, 0.13], [0.63, 0.16], [0.61, 0.27], [0.58, 0.34], [0.6, 0.47], [0.56, 0.42], [0.53, 0.46], [0.51, 0.41]]}" + } + ], + "images": "10601_exterior_image_2_left#0", + "gt": "[[143, 30], [203, 22], [198, 23], [202, 28], [196, 48], [187, 61], [193, 84], [180, 76], [170, 82], [164, 73]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0012_user_0010_scene_0005_cfg_0004_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push the soccer ball into the goal', and now you are performing subtask 'push the soccer ball toward the goal.'\nFrom start point [0.38, 0.0], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.28, 0.0], [0.29, 0.0], [0.28, 0.02], [0.28, 0.03], [0.27, 0.02], [0.32, 0.06], [0.31, 0.06], [0.32, 0.07], [0.34, 0.1]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0012_user_0010_scene_0005_cfg_0004_104122062823#0.jpg", + "gt": "[[243, 0], [180, 0], [184, 0], [182, 7], [178, 11], [174, 7], [207, 21], [201, 21], [202, 25], [216, 37]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0064_user_0020_scene_0009_cfg_0007_104122061602#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the roll of paper down from the shelf', and now you are performing subtask 'take the roll of paper down from the shelf.'\nFrom start point [0.7, 0.32], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.32], [0.68, 0.33], [0.67, 0.29], [0.62, 0.22], [0.64, 0.23], [0.65, 0.37], [0.68, 0.45], [0.68, 0.37], [0.66, 0.32], [0.66, 0.32]]}" + } + ], + "images": "RH20T_cfg7_task_0064_user_0020_scene_0009_cfg_0007_104122061602#0", + "gt": "[[448, 116], [435, 120], [430, 103], [399, 78], [412, 83], [418, 133], [438, 163], [438, 134], [424, 114], [424, 114]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "62212_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the button of the far right socket', and now you are performing subtask 'press the button of the far right socket.'\nStarting at [0.32, 0.02], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.02], [0.36, 0.11], [0.38, 0.23], [0.38, 0.32], [0.34, 0.39], [0.41, 0.39], [0.42, 0.42], [0.39, 0.46], [0.41, 0.49], [0.41, 0.56]]}" + } + ], + "images": "62212_exterior_image_2_left#0", + "gt": "[[103, 3], [116, 20], [122, 41], [120, 57], [109, 71], [132, 71], [133, 76], [125, 83], [130, 89], [130, 101]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0083_user_0007_scene_0010_cfg_0005_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'swat with a flyswatter', and now you are performing subtask 'swat with a flyswatter.'\nYou begin at [0.45, 0.3]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.3], [0.48, 0.49], [0.49, 0.38], [0.51, 0.61], [0.48, 0.47], [0.48, 0.34], [0.46, 0.36], [0.45, 0.39], [0.47, 0.44], [0.5, 0.41]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0083_user_0007_scene_0010_cfg_0005_036422060215#0.jpg", + "gt": "[[286, 108], [310, 178], [311, 136], [324, 220], [310, 170], [309, 123], [297, 128], [287, 142], [300, 157], [321, 146]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "24724_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the open drawer', and now you are performing subtask 'close the open drawer.'\nBegin at [0.46, 1.0] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 1.0], [0.45, 0.99], [0.42, 0.97], [0.4, 0.97], [0.36, 0.92], [0.34, 0.92], [0.33, 0.91], [0.3, 0.89], [0.3, 0.87], [0.29, 0.87]]}" + } + ], + "images": "24724_exterior_image_1_left#0", + "gt": "[[147.5, 180.0], [142.5, 179.0], [134.5, 175.0], [128.5, 174.0], [114.5, 165.0], [109.5, 165.0], [105.5, 164.0], [97.5, 160.0], [94.5, 156.0], [93.5, 157.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0038_user_0014_scene_0004_cfg_0001_035622060973#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the cup', and now you are performing subtask 'pick up the cup.'\nThe trajectory starts at [0.38, 0.36]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.36], [0.39, 0.38], [0.4, 0.42], [0.45, 0.47], [0.47, 0.54], [0.49, 0.56], [0.48, 0.57], [0.47, 0.48], [0.43, 0.36], [0.44, 0.31]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0038_user_0014_scene_0004_cfg_0001_035622060973#0.jpg", + "gt": "[[246, 130], [247, 136], [259, 152], [285, 171], [299, 195], [311, 200], [309, 206], [299, 174], [276, 130], [282, 110]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "20113_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'roll the object on the cloth', and now you are performing subtask 'roll the object on the cloth.'\nBegin at [0.55, 0.54] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.54], [0.55, 0.42], [0.55, 0.59], [0.53, 0.52], [0.42, 0.71], [0.51, 0.62], [0.44, 0.67], [0.49, 0.55], [0.55, 0.59], [0.54, 0.54]]}" + } + ], + "images": "20113_exterior_image_2_left#0", + "gt": "[[175, 97], [177, 75], [175, 106], [170, 94], [135, 127], [163, 111], [142, 121], [156, 99], [176, 107], [173, 97]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0094_user_0007_scene_0006_cfg_0005_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'press the part of the box, press the part of the box.'\nStart at [0.45, 0.42]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.42], [0.43, 0.43], [0.39, 0.46], [0.38, 0.58], [0.36, 0.64], [0.37, 0.66], [0.4, 0.69], [0.41, 0.73], [0.41, 0.71], [0.37, 0.68]]}" + } + ], + "images": "RH20T_cfg5_task_0094_user_0007_scene_0006_cfg_0005_037522062165#0", + "gt": "[[289, 152], [274, 154], [250, 165], [243, 209], [231, 231], [236, 238], [256, 248], [264, 264], [260, 256], [235, 244]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "24404_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the blue can's bottom drawer and remove the green object', and now you are performing subtask 'pull the drawer completely open.'\nThe trajectory starts at [0.6, 0.39]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.39], [0.59, 0.42], [0.59, 0.47], [0.58, 0.48], [0.6, 0.51], [0.66, 0.51], [0.68, 0.49], [0.68, 0.45], [0.66, 0.42], [0.63, 0.39]]}" + } + ], + "images": "24404_exterior_image_1_left#143", + "gt": "[[191, 71], [188, 75], [188, 85], [185, 87], [192, 92], [210, 92], [218, 89], [217, 81], [212, 75], [201, 70]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "87466_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the packet from the sink and place it on the left side of the counter', and now you are performing subtask 'remove the packet from the sink and place it on the left side of the counter.'\nWith start location [0.33, 0.57], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.33, 0.57], [0.37, 0.59], [0.44, 0.74], [0.43, 0.92], [0.43, 0.75], [0.38, 0.71], [0.34, 0.7], [0.31, 0.76], [0.32, 0.66], [0.3, 0.6]]}" + } + ], + "images": "87466_exterior_image_1_left#0", + "gt": "[[105, 102], [118, 107], [141, 133], [137, 166], [136, 135], [122, 127], [108, 126], [100, 136], [102, 118], [96, 108]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "79681_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the towel and lay it flat on the countertop', and now you are performing subtask 'pull a corner of the towel again to completely unfold the towel.'\nYour initial coordinate is [0.59, 0.57]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.57], [0.62, 0.6], [0.62, 0.63], [0.61, 0.61], [0.62, 0.61], [0.63, 0.52], [0.63, 0.51], [0.65, 0.52], [0.64, 0.49], [0.63, 0.5]]}" + } + ], + "images": "79681_exterior_image_1_left#254", + "gt": "[[188, 102], [200, 108], [199, 114], [195, 110], [197, 110], [201, 94], [203, 91], [208, 93], [206, 89], [201, 90]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "67867_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the pot from the right top plate to the right bottom plate of the stove', and now you are performing subtask 'pick up the pot from the right top plate of the stove.'\nWith start location [0.62, 0.69], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.69], [0.62, 0.69], [0.63, 0.7], [0.64, 0.7], [0.64, 0.71], [0.65, 0.72], [0.65, 0.73], [0.65, 0.72], [0.65, 0.72], [0.65, 0.72]]}" + } + ], + "images": "67867_exterior_image_1_left#186", + "gt": "[[199, 124], [200, 124], [201, 126], [204, 126], [206, 127], [207, 129], [207, 131], [207, 130], [208, 130], [208, 130]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "65467_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the bottle from the table and put it inside the clear bowl, turn off the white extension cable, take out one bottle from the clear bowl and place it on the left side of the table', and now you are performing subtask 'place the bottle on the left side of the table.'\nStart at [0.79, 0.61]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.79, 0.61], [0.7, 0.59], [0.72, 0.49], [0.65, 0.52], [0.59, 0.61], [0.54, 0.47], [0.49, 0.61], [0.46, 0.66], [0.46, 0.6], [0.46, 0.6]]}" + } + ], + "images": "65467_exterior_image_1_left#265", + "gt": "[[253, 110], [224, 107], [229, 88], [207, 93], [190, 109], [172, 84], [157, 110], [147, 118], [148, 108], [148, 108]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "21793_exterior_image_1_left#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn the caps on the bed', and now you are performing subtask 'place the black and white hat on the red hat.'\nStarting at [0.21, 0.72], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.21, 0.72], [0.18, 0.6], [0.08, 0.53], [0.09, 0.49], [0.11, 0.53], [0.09, 0.63], [0.08, 0.57], [0.15, 0.54], [0.28, 0.46], [0.25, 0.44]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/21793_exterior_image_1_left#302.jpg", + "gt": "[[67, 129], [58, 108], [26, 96], [29, 88], [35, 95], [28, 114], [27, 102], [47, 98], [89, 82], [81, 79]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0083_user_0007_scene_0009_cfg_0002_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'swat with a flyswatter', and now you are performing subtask 'swat with a flyswatter.'\nBegin at [0.55, 0.71] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.71], [0.55, 0.64], [0.54, 0.65], [0.53, 0.61], [0.52, 0.51], [0.5, 0.53], [0.52, 0.5], [0.53, 0.58], [0.58, 0.69], [0.58, 0.76]]}" + } + ], + "images": "RH20T_cfg2_task_0083_user_0007_scene_0009_cfg_0002_036422060215#0", + "gt": "[[350, 255], [354, 229], [347, 235], [341, 221], [335, 184], [323, 190], [336, 181], [340, 210], [374, 250], [369, 274]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0050_user_0014_scene_0010_cfg_0004_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the plate on the rack', and now you are performing subtask 'put the plate on the rack.'\nYour initial coordinate is [0.38, 0.0]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.44, 0.01], [0.39, 0.0], [0.42, 0.05], [0.43, 0.22], [0.4, 0.12], [0.49, 0.01], [0.77, 0.0], [0.52, 0.04], [0.39, 0.0]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0050_user_0014_scene_0010_cfg_0004_104122062823#0.jpg", + "gt": "[[243, 0], [279, 3], [252, 0], [270, 18], [276, 80], [254, 43], [314, 3], [494, 0], [330, 15], [250, 0]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0058_user_0010_scene_0006_cfg_0003_036422060909#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn off the power strips', and now you are performing subtask 'press the second button of the power strip, press the third button of the power strip.'\nStarting at [0.33, 0.57], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.33, 0.57], [0.32, 0.57], [0.32, 0.57], [0.32, 0.58], [0.32, 0.59], [0.32, 0.59], [0.32, 0.59], [0.32, 0.58], [0.31, 0.57], [0.29, 0.59]]}" + } + ], + "images": "RH20T_cfg3_task_0058_user_0010_scene_0006_cfg_0003_036422060909#166", + "gt": "[[210, 205], [208, 207], [207, 207], [206, 209], [206, 211], [207, 211], [207, 212], [206, 209], [198, 207], [184, 211]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0034_user_0016_scene_0001_cfg_0004_104122062823#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'push the block next to another block, push the block next to another block to form a line.'\nStart at [0.43, 0.24]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.24], [0.41, 0.21], [0.4, 0.17], [0.43, 0.09], [0.45, 0.07], [0.47, 0.06], [0.46, 0.16], [0.47, 0.21], [0.45, 0.14], [0.44, 0.14]]}" + } + ], + "images": "RH20T_cfg4_task_0034_user_0016_scene_0001_cfg_0004_104122062823#147", + "gt": "[[275, 85], [264, 77], [259, 61], [277, 33], [290, 25], [303, 21], [296, 58], [304, 75], [289, 52], [281, 51]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "43346_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the glue stick out of the cup and place it on the left side of the table', and now you are performing subtask 'take the glue stick out of the cup and place it on the left side of the table.'\nYou start from [0.77, 0.01]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.77, 0.01], [0.74, 0.03], [0.78, 0.06], [0.6, 0.24], [0.52, 0.44], [0.51, 0.31], [0.46, 0.34], [0.45, 0.42], [0.49, 0.47], [0.49, 0.41]]}" + } + ], + "images": "43346_exterior_image_1_left#0", + "gt": "[[245.5, 2.0], [236.5, 6.0], [249.5, 11.0], [192.5, 44.0], [167.5, 80.0], [164.5, 55.0], [146.5, 62.0], [143.5, 76.0], [156.5, 84.0], [155.5, 74.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0050_user_0020_scene_0008_cfg_0007_104122060811#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the plate on the bowl rack', and now you are performing subtask 'put the plate on the bowl rack.'\nStarting at [0.6, 0.36], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.36], [0.57, 0.37], [0.58, 0.35], [0.58, 0.32], [0.54, 0.3], [0.5, 0.29], [0.49, 0.33], [0.54, 0.29], [0.57, 0.34], [0.57, 0.32]]}" + } + ], + "images": "RH20T_cfg7_task_0050_user_0020_scene_0008_cfg_0007_104122060811#0", + "gt": "[[387, 130], [364, 133], [373, 126], [369, 117], [347, 107], [322, 106], [316, 118], [344, 105], [364, 122], [365, 116]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0003_user_0014_scene_0010_cfg_0006_104122060811#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the green button in the middle, press the yellow button at the top.'\nYour initial coordinate is [0.61, 0.33]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.61, 0.32], [0.6, 0.32], [0.59, 0.32], [0.57, 0.32], [0.57, 0.33], [0.56, 0.34], [0.56, 0.34], [0.56, 0.34], [0.57, 0.32], [0.57, 0.32]]}" + } + ], + "images": "RH20T_cfg6_task_0003_user_0014_scene_0010_cfg_0006_104122060811#167", + "gt": "[[388, 117], [382, 114], [375, 114], [368, 117], [367, 119], [361, 121], [360, 123], [361, 123], [363, 117], [365, 116]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "61677_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pour the sweets out of the cup and onto the table', and now you are performing subtask 'pour the sweets out of the cup and onto the table.'\nYour starting position is [0.35, 0.49]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.49], [0.5, 0.46], [0.58, 0.5], [0.55, 0.52], [0.56, 0.58], [0.53, 0.44], [0.59, 0.4], [0.59, 0.51], [0.55, 0.43], [0.55, 0.43]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/61677_exterior_image_1_left#0.jpg", + "gt": "[[112.5, 88.0], [161.5, 83.0], [184.5, 90.0], [176.5, 94.0], [179.5, 104.0], [168.5, 79.0], [188.5, 72.0], [187.5, 91.0], [175.5, 78.0], [175.5, 78.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0022_user_0014_scene_0007_cfg_0006_104122061018#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'wave the flag', and now you are performing subtask 'pick up the flag on the table, shake the flag in the air, place the flag on the table.'\nYour initial coordinate is [0.52, 0.54]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.54], [0.55, 0.36], [0.57, 0.34], [0.57, 0.41], [0.6, 0.01], [0.63, 0.0], [0.65, 0.08], [0.6, 0.0], [0.56, 0.0], [0.63, 0.02]]}" + } + ], + "images": "RH20T_cfg6_task_0022_user_0014_scene_0007_cfg_0006_104122061018#154", + "gt": "[[332, 195], [350, 129], [364, 124], [366, 147], [386, 2], [402, 0], [417, 28], [386, 0], [361, 0], [404, 6]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "56634_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the bottom drawer of the right cabinet', and now you are performing subtask 'close the bottom drawer of the right cabinet.'\nThe trajectory starts at [0.6, 1.0]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 1.0], [0.61, 0.94], [0.68, 0.85], [0.68, 0.78], [0.61, 0.72], [0.6, 0.76], [0.66, 0.79], [0.71, 0.76], [0.66, 0.72], [0.64, 0.66]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/56634_exterior_image_1_left#0.jpg", + "gt": "[[193, 180], [196, 169], [219, 153], [218, 141], [194, 130], [192, 136], [210, 142], [227, 137], [210, 130], [205, 118]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0061_user_0020_scene_0001_cfg_0007_104122063633#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the lid of the gift box on the table, then take everything out of the gift box and place it all on the table', and now you are performing subtask 'place the lid on the tabletop, pick up the cube from the gift box, transfer the cube from the gift box to the tabletop, place the cube on the tabletop.'\nYou begin at [0.26, 0.74]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.26, 0.74], [0.26, 0.56], [0.25, 0.54], [0.25, 0.58], [0.24, 0.58], [0.26, 0.53], [0.25, 0.44], [0.28, 0.38], [0.28, 0.43], [0.28, 0.43]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0061_user_0020_scene_0001_cfg_0007_104122063633#213.jpg", + "gt": "[[167, 267], [166, 200], [163, 193], [160, 209], [156, 208], [166, 190], [163, 157], [180, 136], [177, 153], [177, 153]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "48447_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unhang the orange towel and put it in the sink', and now you are performing subtask 'unhang the orange towel and put it in the sink.'\nStarting at [0.28, 0.06], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.28, 0.06], [0.35, 0.22], [0.39, 0.2], [0.36, 0.26], [0.39, 0.22], [0.43, 0.28], [0.32, 0.41], [0.34, 0.08], [0.38, 0.19], [0.39, 0.16]]}" + } + ], + "images": "48447_exterior_image_1_left#0", + "gt": "[[91, 11], [111, 39], [125, 36], [116, 47], [125, 40], [136, 51], [103, 73], [108, 14], [120, 35], [124, 28]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0105_user_0020_scene_0007_cfg_0006_037522061512#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the object on the shelf', and now you are performing subtask 'pick up the object from the tabletop, transfer the object from the tabletop to the specified location, place the object at the specified location.'\nYou start from [0.33, 0.34]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.33, 0.34], [0.38, 0.28], [0.31, 0.27], [0.26, 0.39], [0.22, 0.4], [0.18, 0.54], [0.18, 0.46], [0.23, 0.33], [0.31, 0.25], [0.31, 0.25]]}" + } + ], + "images": "RH20T_cfg6_task_0105_user_0020_scene_0007_cfg_0006_037522061512#720", + "gt": "[[210, 124], [240, 99], [197, 96], [164, 139], [140, 144], [116, 196], [117, 165], [147, 118], [201, 91], [201, 91]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0071_user_0020_scene_0007_cfg_0006_104122063633#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close a sliding window', and now you are performing subtask 'push the handle of the sliding window to close the lower window.'\nYour starting position is [0.34, 0.37]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.34, 0.37], [0.35, 0.43], [0.36, 0.44], [0.34, 0.4], [0.33, 0.4], [0.35, 0.43], [0.34, 0.38], [0.37, 0.38], [0.37, 0.37], [0.37, 0.37]]}" + } + ], + "images": "RH20T_cfg6_task_0071_user_0020_scene_0007_cfg_0006_104122063633#316", + "gt": "[[217, 134], [221, 153], [229, 157], [216, 144], [210, 145], [222, 153], [219, 135], [238, 138], [239, 132], [239, 132]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "73732_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open and close the microwave door and then close the top fridge door', and now you are performing subtask 'push the microwave door to the closed position.'\nYour initial coordinate is [0.7, 0.24]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.24], [0.69, 0.2], [0.66, 0.19], [0.61, 0.31], [0.56, 0.34], [0.51, 0.28], [0.49, 0.21], [0.46, 0.19], [0.53, 0.23], [0.59, 0.21]]}" + } + ], + "images": "73732_exterior_image_1_left#229", + "gt": "[[223.5, 43.0], [219.5, 36.0], [211.5, 35.0], [196.5, 56.0], [178.5, 61.0], [164.5, 50.0], [156.5, 37.0], [148.5, 35.0], [169.5, 41.0], [189.5, 38.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0057_user_0007_scene_0003_cfg_0005_104122063678#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn on the power strips', and now you are performing subtask 'press the third button on the power socket, press the fourth button on the power socket.'\nBegin at [0.47, 0.53] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.53], [0.45, 0.58], [0.44, 0.6], [0.45, 0.61], [0.45, 0.61], [0.45, 0.61], [0.45, 0.6], [0.45, 0.59], [0.44, 0.59], [0.45, 0.61]]}" + } + ], + "images": "RH20T_cfg5_task_0057_user_0007_scene_0003_cfg_0005_104122063678#503", + "gt": "[[304, 192], [287, 210], [284, 216], [291, 220], [288, 220], [290, 220], [287, 217], [286, 211], [284, 214], [286, 221]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "10945_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the cup to the left', and now you are performing subtask 'move the cup to the left.'\nThe trajectory starts at [0.38, 0.05]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.05], [0.44, 0.11], [0.54, 0.31], [0.56, 0.41], [0.55, 0.48], [0.56, 0.49], [0.52, 0.42], [0.48, 0.42], [0.48, 0.44], [0.47, 0.34]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/10945_exterior_image_1_left#0.jpg", + "gt": "[[122, 9], [142, 19], [172, 56], [178, 74], [177, 86], [179, 88], [166, 75], [154, 75], [155, 79], [152, 62]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "65705_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the towel', and now you are performing subtask 'fold the towel to fully unfold.'\nYour starting position is [0.44, 0.55]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.55], [0.46, 0.53], [0.47, 0.49], [0.47, 0.51], [0.53, 0.52], [0.57, 0.57], [0.56, 0.59], [0.57, 0.55], [0.53, 0.52], [0.45, 0.54]]}" + } + ], + "images": "65705_exterior_image_2_left#0", + "gt": "[[141, 99], [148, 96], [150, 89], [149, 92], [169, 94], [181, 102], [180, 107], [182, 99], [171, 93], [145, 98]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0001_user_0014_scene_0007_cfg_0007_104122060811#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the nearest green button', and now you are performing subtask 'press the nearest green button.'\nBegin at [0.6, 0.36] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.36], [0.61, 0.36], [0.61, 0.34], [0.61, 0.31], [0.6, 0.31], [0.6, 0.34], [0.6, 0.3], [0.61, 0.31], [0.62, 0.36], [0.62, 0.36]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0001_user_0014_scene_0007_cfg_0007_104122060811#0.jpg", + "gt": "[[387, 130], [389, 131], [391, 121], [388, 110], [384, 113], [383, 121], [387, 108], [392, 112], [395, 129], [395, 129]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0015_user_0014_scene_0008_cfg_0007_104122061602#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'play the drum', and now you are performing subtask 'play the drum.'\nYour initial coordinate is [0.7, 0.32]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.32], [0.77, 0.39], [0.81, 0.52], [0.8, 0.65], [0.77, 0.36], [0.7, 0.31], [0.73, 0.31], [0.75, 0.35], [0.72, 0.5], [0.73, 0.34]]}" + } + ], + "images": "RH20T_cfg7_task_0015_user_0014_scene_0008_cfg_0007_104122061602#0", + "gt": "[[448, 116], [491, 142], [520, 186], [512, 233], [494, 129], [449, 110], [466, 110], [477, 127], [459, 181], [467, 124]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "73732_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open and close the microwave door and then close the top fridge door', and now you are performing subtask 'push the microwave door to the closed position.'\nStart at [0.63, 0.47]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.47], [0.55, 0.38], [0.52, 0.22], [0.53, 0.13], [0.5, 0.24], [0.49, 0.24], [0.52, 0.24], [0.57, 0.3], [0.7, 0.24], [0.7, 0.24]]}" + } + ], + "images": "73732_exterior_image_1_left#0", + "gt": "[[201.5, 85.0], [175.5, 69.0], [166.5, 39.0], [169.5, 24.0], [161.5, 44.0], [155.5, 44.0], [165.5, 44.0], [183.5, 54.0], [223.5, 43.0], [223.5, 43.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "75830_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the orange packet from the green box', and now you are performing subtask 'remove the orange packet from the green box.'\nYou begin at [0.56, 0.33]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.56, 0.32], [0.55, 0.38], [0.62, 0.39], [0.57, 0.25], [0.48, 0.2], [0.49, 0.41], [0.58, 0.38], [0.57, 0.42], [0.53, 0.36], [0.48, 0.38]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/75830_exterior_image_1_left#0.jpg", + "gt": "[[178.5, 58.5], [175.5, 67.5], [198.5, 69.5], [181.5, 44.5], [152.5, 36.5], [157.5, 73.5], [184.5, 68.5], [181.5, 75.5], [168.5, 65.5], [152.5, 68.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "55742_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the white towel on the black object on top of the table', and now you are performing subtask 'place the white towel on the black object on the table.'\nStart at [0.59, 0.06]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.06], [0.52, 0.18], [0.55, 0.39], [0.54, 0.51], [0.51, 0.54], [0.49, 0.36], [0.41, 0.1], [0.37, 0.31], [0.38, 0.28], [0.45, 0.32]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/55742_exterior_image_1_left#0.jpg", + "gt": "[[188, 11], [166, 33], [177, 70], [172, 91], [163, 98], [157, 64], [130, 18], [117, 56], [121, 51], [145, 58]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_104122064161#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the fifth block on the table, transfer the fifth block from the table to the fourth block, place the fifth block on top of the fourth block.'\nThe trajectory starts at [0.55, 0.23]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.23], [0.57, 0.16], [0.57, 0.2], [0.59, 0.21], [0.57, 0.33], [0.58, 0.21], [0.56, 0.18], [0.51, 0.29], [0.51, 0.32], [0.55, 0.18]]}" + } + ], + "images": "RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_104122064161#846", + "gt": "[[354, 82], [363, 59], [363, 71], [379, 76], [364, 118], [373, 75], [357, 66], [325, 103], [329, 117], [354, 64]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "91717_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the orange cloth', and now you are performing subtask 'unfold the orange cloth and lay it flat on the table.'\nYour starting position is [0.62, 0.29]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.29], [0.64, 0.22], [0.66, 0.23], [0.66, 0.39], [0.7, 0.46], [0.68, 0.53], [0.65, 0.5], [0.71, 0.35], [0.76, 0.26], [0.76, 0.26]]}" + } + ], + "images": "91717_exterior_image_1_left#0", + "gt": "[[200, 53], [206, 40], [210, 42], [212, 70], [223, 82], [217, 96], [208, 90], [226, 63], [244, 46], [244, 46]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0015_user_0008_scene_0002_cfg_0001_035622060973#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'play the drum', and now you are performing subtask 'play the drum.'\nYou begin at [0.38, 0.36]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.36], [0.4, 0.51], [0.4, 0.5], [0.4, 0.51], [0.34, 0.37], [0.33, 0.34], [0.34, 0.4], [0.32, 0.32], [0.38, 0.48], [0.37, 0.43]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0015_user_0008_scene_0002_cfg_0001_035622060973#0.jpg", + "gt": "[[246, 131], [254, 183], [256, 179], [258, 185], [218, 134], [213, 121], [215, 143], [204, 114], [240, 174], [234, 153]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0005_user_0010_scene_0004_cfg_0004_104422070011#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'approach and touch the side of the small block', and now you are performing subtask 'approach and touch the side of the small block.'\nStarting at [0.53, 0.29], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.29], [0.54, 0.45], [0.5, 0.41], [0.52, 0.42], [0.43, 0.48], [0.38, 0.54], [0.4, 0.56], [0.46, 0.5], [0.47, 0.49], [0.47, 0.51]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0005_user_0010_scene_0004_cfg_0004_104422070011#0.jpg", + "gt": "[[338, 103], [343, 161], [321, 147], [336, 152], [275, 173], [242, 194], [253, 202], [297, 179], [303, 178], [301, 184]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "37562_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the banana and pineapple plush on the table', and now you are performing subtask 'put the banana and pineapple plush on the table.'\nStart at [0.49, 0.21]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.21], [0.5, 0.19], [0.53, 0.23], [0.62, 0.39], [0.61, 0.53], [0.64, 0.39], [0.73, 0.37], [0.78, 0.38], [0.76, 0.4], [0.76, 0.4]]}" + } + ], + "images": "37562_exterior_image_2_left#0", + "gt": "[[156, 38], [160, 35], [171, 41], [198, 71], [195, 95], [205, 70], [232, 67], [250, 69], [243, 72], [243, 72]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0050_user_0020_scene_0008_cfg_0007_037522061512#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the plate on the bowl rack', and now you are performing subtask 'put the plate on the bowl rack.'\nBegin at [0.41, 0.24] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.24], [0.37, 0.31], [0.37, 0.4], [0.37, 0.32], [0.33, 0.25], [0.22, 0.25], [0.21, 0.3], [0.25, 0.34], [0.28, 0.21], [0.34, 0.22]]}" + } + ], + "images": "RH20T_cfg7_task_0050_user_0020_scene_0008_cfg_0007_037522061512#0", + "gt": "[[261, 86], [239, 113], [236, 143], [239, 117], [213, 91], [142, 90], [135, 109], [158, 123], [178, 75], [219, 80]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0032_user_0015_scene_0004_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'slide the outer casing onto the gift box', and now you are performing subtask 'slide the outer casing onto the gift box.'\nYour initial coordinate is [0.63, 0.37]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.7, 0.19], [0.7, 0.14], [0.69, 0.16], [0.72, 0.24], [0.71, 0.3], [0.74, 0.43], [0.74, 0.34], [0.78, 0.44], [0.8, 0.56]]}" + } + ], + "images": "RH20T_cfg1_task_0032_user_0015_scene_0004_cfg_0001_038522062547#0", + "gt": "[[405, 134], [449, 69], [449, 49], [441, 58], [460, 86], [456, 109], [475, 156], [474, 122], [499, 160], [509, 200]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0091_user_0007_scene_0004_cfg_0005_104422070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move an object from one box to another', and now you are performing subtask 'pick up the block from the pink box, transfer the block from the pink box to the table, throw the block onto the table.'\nBegin at [0.58, 0.37] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.37], [0.61, 0.37], [0.63, 0.34], [0.66, 0.36], [0.67, 0.36], [0.7, 0.42], [0.7, 0.46], [0.69, 0.38], [0.68, 0.35], [0.65, 0.33]]}" + } + ], + "images": "RH20T_cfg5_task_0091_user_0007_scene_0004_cfg_0005_104422070044#0", + "gt": "[[374, 132], [390, 132], [406, 123], [425, 129], [429, 131], [447, 150], [447, 166], [443, 136], [433, 127], [413, 120]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0037_user_0014_scene_0009_cfg_0007_104122061330#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the first block from the desktop, transfer the block from the desktop to the second block, place the block on the second block.'\nBegin at [0.42, 0.29] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.29], [0.49, 0.29], [0.5, 0.28], [0.51, 0.24], [0.54, 0.21], [0.56, 0.28], [0.56, 0.3], [0.56, 0.27], [0.57, 0.28], [0.57, 0.31]]}" + } + ], + "images": "RH20T_cfg7_task_0037_user_0014_scene_0009_cfg_0007_104122061330#0", + "gt": "[[266, 103], [314, 103], [320, 101], [325, 86], [347, 77], [361, 100], [357, 109], [359, 98], [364, 102], [365, 110]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "15693_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the metal can into the sink', and now you are performing subtask 'put the metal can into the sink.'\nStarting at [0.35, 0.49], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.49], [0.34, 0.71], [0.37, 0.71], [0.38, 0.66], [0.49, 0.68], [0.51, 0.88], [0.51, 0.76], [0.47, 0.73], [0.41, 0.77], [0.42, 0.78]]}" + } + ], + "images": "15693_exterior_image_1_left#0", + "gt": "[[112, 88], [109, 127], [119, 128], [121, 118], [156, 122], [163, 158], [164, 136], [150, 132], [130, 139], [134, 141]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0009_user_0014_scene_0004_cfg_0001_750612070851#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take out a Hanoi block and throw it aside', and now you are performing subtask 'take out a Hanoi block and throw it aside.'\nFrom start point [0.5, 0.43], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.43], [0.53, 0.43], [0.6, 0.56], [0.62, 0.69], [0.6, 0.77], [0.63, 0.67], [0.53, 0.61], [0.47, 0.72], [0.48, 0.79], [0.55, 0.51]]}" + } + ], + "images": "RH20T_cfg1_task_0009_user_0014_scene_0004_cfg_0001_750612070851#0", + "gt": "[[323, 156], [341, 155], [383, 202], [396, 249], [385, 278], [406, 242], [337, 220], [301, 259], [305, 283], [349, 182]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_104422070044#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the block on the other blocks, transfer the block from the plane to the position, transfer the block from the plane to the other blocks, place the block on the other blocks.'\nYou start from [0.57, 0.45]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.45], [0.61, 0.37], [0.62, 0.36], [0.62, 0.36], [0.63, 0.36], [0.63, 0.38], [0.62, 0.34], [0.61, 0.34], [0.58, 0.38], [0.57, 0.41]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_104422070044#230.jpg", + "gt": "[[367, 161], [388, 133], [395, 129], [394, 131], [402, 131], [402, 135], [397, 123], [393, 121], [372, 136], [363, 147]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0010_user_0010_scene_0007_cfg_0005_105422061350#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the handset of the telephone on the corresponding phone cradle', and now you are performing subtask 'place the handset of the telephone on the corresponding phone cradle.'\nYou begin at [0.43, 0.13]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.13], [0.43, 0.16], [0.46, 0.18], [0.43, 0.19], [0.43, 0.16], [0.48, 0.2], [0.51, 0.27], [0.52, 0.21], [0.47, 0.32], [0.47, 0.32]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0010_user_0010_scene_0007_cfg_0005_105422061350#0.jpg", + "gt": "[[276, 47], [273, 58], [295, 66], [273, 67], [275, 57], [309, 73], [326, 96], [336, 76], [301, 114], [301, 114]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "89980_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the Coca Cola can to the open cabinet', and now you are performing subtask 'move the Coca Cola can to the open cabinet.'\nWith start location [0.74, 0.38], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.74, 0.38], [0.7, 0.62], [0.71, 0.72], [0.68, 0.73], [0.65, 0.51], [0.65, 0.21], [0.63, 0.06], [0.59, 0.11], [0.66, 0.07], [0.68, 0.17]]}" + } + ], + "images": "89980_exterior_image_2_left#0", + "gt": "[[237, 69], [225, 112], [228, 129], [217, 131], [207, 91], [208, 38], [202, 10], [190, 19], [211, 12], [219, 30]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0083_user_0007_scene_0005_cfg_0005_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'swat with a flyswatter', and now you are performing subtask 'swat with a flyswatter.'\nYour initial coordinate is [0.45, 0.42]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.42], [0.47, 0.45], [0.52, 0.79], [0.53, 0.96], [0.53, 1.0], [0.51, 1.0], [0.62, 0.92], [0.61, 0.82], [0.48, 0.72], [0.48, 0.72]]}" + } + ], + "images": "RH20T_cfg5_task_0083_user_0007_scene_0005_cfg_0005_037522062165#0", + "gt": "[[288, 152], [301, 162], [334, 283], [340, 345], [337, 360], [328, 360], [394, 331], [391, 295], [308, 259], [308, 259]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "24684_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer, pick up the green stuffed animal and put it on the table', and now you are performing subtask 'open the drawer, pick up the green stuffed animal and put it on the table.'\nBegin at [0.39, 0.0] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.0], [0.6, 0.49], [0.49, 0.48], [0.53, 0.39], [0.57, 0.47], [0.52, 0.28], [0.45, 0.29], [0.44, 0.42], [0.45, 0.35], [0.45, 0.35]]}" + } + ], + "images": "24684_exterior_image_1_left#0", + "gt": "[[126, 0], [192, 89], [158, 86], [169, 71], [182, 84], [165, 50], [143, 53], [142, 75], [143, 63], [143, 63]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0043_user_0014_scene_0008_cfg_0006_104122064161#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unscrew the lid from the jar', and now you are performing subtask 'twist the lid of the jar counterclockwise.'\nFrom start point [0.5, 0.49], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.49], [0.5, 0.5], [0.5, 0.51], [0.5, 0.51], [0.5, 0.51], [0.5, 0.51], [0.5, 0.51], [0.5, 0.51], [0.5, 0.51], [0.51, 0.52]]}" + } + ], + "images": "RH20T_cfg6_task_0043_user_0014_scene_0008_cfg_0006_104122064161#430", + "gt": "[[320, 177], [320, 180], [319, 182], [322, 182], [322, 185], [322, 183], [322, 184], [321, 184], [322, 184], [324, 187]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "22851_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'cover the lid of the pink object on the table', and now you are performing subtask 'place the lid on the object on the table.'\nBegin at [0.67, 0.53] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.67, 0.53], [0.67, 0.5], [0.64, 0.54], [0.66, 0.49], [0.75, 0.46], [0.79, 0.54], [0.84, 0.59], [0.85, 0.68], [0.82, 0.57], [0.84, 0.58]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/22851_exterior_image_2_left#199.jpg", + "gt": "[[213, 96], [215, 90], [206, 97], [211, 89], [240, 82], [252, 97], [270, 106], [272, 123], [264, 103], [269, 105]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0042_user_0007_scene_0002_cfg_0002_037522062165#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'screw the lid onto the jar (needs to be rotated)', and now you are performing subtask 'twist the lid clockwise to the jar.'\nYour starting position is [0.39, 0.69]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.69], [0.39, 0.71], [0.39, 0.7], [0.38, 0.7], [0.38, 0.71], [0.38, 0.71], [0.38, 0.71], [0.38, 0.72], [0.4, 0.72], [0.39, 0.73]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg2_task_0042_user_0007_scene_0002_cfg_0002_037522062165#796.jpg", + "gt": "[[251, 249], [248, 254], [247, 252], [243, 253], [245, 255], [244, 257], [246, 257], [244, 260], [253, 260], [249, 264]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0058_user_0015_scene_0003_cfg_0001_035622060973#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn off the power strips', and now you are performing subtask 'press the first button on the power socket.'\nBegin at [0.38, 0.36] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.36], [0.39, 0.36], [0.38, 0.33], [0.37, 0.27], [0.38, 0.17], [0.36, 0.17], [0.37, 0.16], [0.36, 0.18], [0.37, 0.21], [0.37, 0.21]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0058_user_0015_scene_0003_cfg_0001_035622060973#0.jpg", + "gt": "[[246, 130], [249, 129], [246, 118], [235, 97], [240, 62], [233, 60], [235, 59], [232, 65], [236, 75], [236, 75]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "62555_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick the marker and put it in the cup', and now you are performing subtask 'pick the marker and put it in the cup.'\nWith start location [0.53, 0.0], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.0], [0.47, 0.23], [0.49, 0.54], [0.5, 0.22], [0.55, 0.13], [0.63, 0.14], [0.58, 0.24], [0.57, 0.16], [0.47, 0.11], [0.47, 0.11]]}" + } + ], + "images": "62555_exterior_image_2_left#0", + "gt": "[[168, 0], [151, 41], [157, 98], [160, 39], [177, 23], [201, 26], [187, 43], [183, 28], [152, 19], [152, 19]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0074_user_0015_scene_0006_cfg_0004_038522062288#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the button horizontally', and now you are performing subtask 'press the button horizontally.'\nStart at [0.37, 0.33]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.33], [0.35, 0.36], [0.31, 0.29], [0.27, 0.39], [0.23, 0.32], [0.21, 0.3], [0.19, 0.33], [0.17, 0.29], [0.23, 0.3], [0.37, 0.56]]}" + } + ], + "images": "RH20T_cfg4_task_0074_user_0015_scene_0006_cfg_0004_038522062288#0", + "gt": "[[236, 118], [222, 131], [197, 103], [175, 140], [150, 115], [132, 108], [123, 118], [109, 104], [150, 109], [234, 202]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0108_user_0014_scene_0004_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'sharpen the pencil with a pencil sharpener', and now you are performing subtask 'sharpen the pencil with a pencil sharpener.'\nWith start location [0.63, 0.37], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.65, 0.24], [0.62, 0.21], [0.62, 0.26], [0.6, 0.27], [0.62, 0.31], [0.65, 0.31], [0.65, 0.29], [0.63, 0.26], [0.63, 0.16]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0108_user_0014_scene_0004_cfg_0001_038522062547#0.jpg", + "gt": "[[406, 132], [417, 85], [395, 74], [398, 92], [386, 97], [398, 110], [417, 112], [417, 103], [403, 92], [402, 57]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0059_user_0010_scene_0004_cfg_0003_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn the knob to increase the volume of the speaker', and now you are performing subtask 'twist the knob on the speaker clockwise to increase the volume.'\nYour starting position is [0.3, 0.12]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.3, 0.12], [0.21, 0.12], [0.32, 0.11], [0.34, 0.15], [0.38, 0.15], [0.38, 0.17], [0.4, 0.16], [0.4, 0.2], [0.38, 0.23], [0.39, 0.25]]}" + } + ], + "images": "RH20T_cfg3_task_0059_user_0010_scene_0004_cfg_0003_104122062823#0", + "gt": "[[193, 42], [132, 43], [204, 39], [218, 55], [240, 54], [245, 62], [258, 56], [257, 72], [242, 83], [249, 90]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "26790_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push the first small block to the left and use the second small block to push the third small block to the right', and now you are performing subtask 'push the third small object to the right with the second small object.'\nStarting at [0.42, 0.5], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.5], [0.45, 0.36], [0.48, 0.37], [0.46, 0.42], [0.47, 0.47], [0.49, 0.43], [0.5, 0.37], [0.53, 0.41], [0.57, 0.28], [0.57, 0.28]]}" + } + ], + "images": "26790_exterior_image_1_left#118", + "gt": "[[134.0, 90.0], [143.0, 64.0], [155.0, 67.0], [147.0, 75.0], [149.0, 85.0], [158.0, 77.0], [159.0, 66.0], [171.0, 74.0], [184.0, 50.0], [184.0, 50.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0034_user_0016_scene_0001_cfg_0004_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'push the block next to another block.'\nFrom start point [0.45, 0.0], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.0], [0.48, 0.0], [0.44, 0.0], [0.45, 0.04], [0.41, 0.12], [0.44, 0.22], [0.41, 0.26], [0.44, 0.28], [0.49, 0.23], [0.43, 0.24]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0034_user_0016_scene_0001_cfg_0004_104122062823#0.jpg", + "gt": "[[290, 0], [305, 0], [279, 0], [285, 15], [262, 44], [279, 78], [265, 94], [284, 102], [314, 82], [275, 85]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0118_user_0010_scene_0004_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn the water tap handle counterclockwise', and now you are performing subtask 'turn the water tap handle counterclockwise.'\nStarting at [0.33, 0.33], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.33, 0.33], [0.4, 0.43], [0.41, 0.52], [0.47, 0.54], [0.45, 0.64], [0.42, 0.67], [0.41, 0.71], [0.42, 0.79], [0.43, 0.54], [0.41, 0.42]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0118_user_0010_scene_0004_cfg_0005_104122063678#0.jpg", + "gt": "[[213, 120], [256, 155], [261, 186], [299, 193], [291, 230], [267, 241], [261, 255], [267, 286], [272, 196], [263, 150]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "45999_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick the purple ball and put it in the bowl', and now you are performing subtask 'pick the purple ball and put it in the bowl.'\nStart at [0.52, 0.34]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.34], [0.59, 0.47], [0.58, 0.58], [0.53, 0.67], [0.54, 0.59], [0.61, 0.57], [0.68, 0.61], [0.69, 0.7], [0.68, 0.66], [0.62, 0.57]]}" + } + ], + "images": "45999_exterior_image_2_left#0", + "gt": "[[165, 62], [189, 85], [187, 105], [171, 121], [173, 107], [194, 102], [216, 109], [221, 126], [219, 119], [197, 102]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "6005_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'lay the charger across the table', and now you are performing subtask 'move the power cable from the middle of the table to below the table.'\nBegin at [0.68, 0.0] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.0], [0.52, 0.12], [0.54, 0.13], [0.51, 0.44], [0.52, 0.49], [0.54, 0.36], [0.62, 0.33], [0.74, 0.41], [0.72, 0.38], [0.57, 0.41]]}" + } + ], + "images": "6005_exterior_image_1_left#0", + "gt": "[[219, 0], [168, 22], [174, 23], [163, 79], [168, 88], [172, 65], [198, 59], [237, 74], [231, 68], [181, 74]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_750612070851#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'putting a portion of the Lego blocks on the Lego baseplate', and now you are performing subtask 'insert the red Lego brick into the Lego base, pick up the purple Lego brick from the table, transfer the purple Lego brick from the table to the Lego base, insert the purple Lego brick into the Lego base.'\nYour initial coordinate is [0.32, 0.61]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.61], [0.29, 0.6], [0.28, 0.73], [0.24, 0.82], [0.24, 0.86], [0.26, 0.89], [0.29, 0.68], [0.32, 0.66], [0.38, 0.46], [0.4, 0.57]]}" + } + ], + "images": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_750612070851#142", + "gt": "[[206, 218], [187, 215], [178, 262], [154, 294], [152, 308], [168, 322], [184, 245], [203, 236], [245, 165], [253, 207]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0016_user_0014_scene_0008_cfg_0001_038522063145#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'hit the pool ball', and now you are performing subtask 'hit the pool ball.'\nFrom start point [0.69, 0.27], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.27], [0.69, 0.26], [0.7, 0.26], [0.71, 0.28], [0.71, 0.31], [0.71, 0.35], [0.71, 0.33], [0.7, 0.4], [0.71, 0.36], [0.68, 0.34]]}" + } + ], + "images": "RH20T_cfg1_task_0016_user_0014_scene_0008_cfg_0001_038522063145#0", + "gt": "[[443, 97], [441, 95], [446, 94], [453, 99], [456, 113], [452, 125], [457, 120], [445, 145], [452, 131], [435, 124]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0013_user_0020_scene_0010_cfg_0007_104122061018#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the block on the scale', and now you are performing subtask 'place the block on the scale.'\nStarting at [0.57, 0.09], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.09], [0.53, 0.18], [0.5, 0.16], [0.36, 0.28], [0.42, 0.29], [0.45, 0.38], [0.42, 0.24], [0.44, 0.31], [0.43, 0.19], [0.45, 0.12]]}" + } + ], + "images": "RH20T_cfg7_task_0013_user_0020_scene_0010_cfg_0007_104122061018#0", + "gt": "[[364, 31], [341, 66], [319, 59], [231, 99], [268, 105], [285, 135], [268, 86], [282, 112], [272, 68], [290, 45]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_038522062547#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'fold the first lid on the box on the table, fold the second lid on the box on the table.'\nYour starting position is [0.59, 0.61]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.61], [0.61, 0.56], [0.65, 0.48], [0.67, 0.45], [0.67, 0.47], [0.66, 0.46], [0.66, 0.48], [0.65, 0.5], [0.59, 0.54], [0.59, 0.54]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_038522062547#129.jpg", + "gt": "[[377, 218], [388, 201], [419, 174], [426, 161], [426, 171], [423, 166], [422, 174], [415, 179], [375, 193], [375, 193]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the block on the plane, transfer the block from the plane to the other blocks, place the block on the other blocks.'\nBegin at [0.45, 0.3] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.3], [0.45, 0.38], [0.46, 0.41], [0.46, 0.45], [0.46, 0.41], [0.45, 0.41], [0.41, 0.47], [0.4, 0.47], [0.38, 0.39], [0.38, 0.39]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_036422060215#0.jpg", + "gt": "[[286, 108], [290, 137], [296, 147], [293, 163], [297, 149], [287, 147], [262, 168], [255, 168], [246, 139], [246, 139]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0073_user_0015_scene_0007_cfg_0004_104122062823#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the things in the drawer', and now you are performing subtask 'place the button in the drawer, push the drawer inward to close.'\nWith start location [0.58, 0.17], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.17], [0.58, 0.09], [0.59, 0.08], [0.59, 0.06], [0.59, 0.1], [0.59, 0.13], [0.59, 0.14], [0.57, 0.14], [0.56, 0.11], [0.58, 0.04]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0073_user_0015_scene_0007_cfg_0004_104122062823#237.jpg", + "gt": "[[370, 62], [374, 34], [378, 29], [376, 21], [380, 37], [380, 46], [376, 50], [367, 50], [357, 39], [371, 16]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0013_user_0016_scene_0007_cfg_0004_104422070011#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the block on the scale', and now you are performing subtask 'place the block on the scale.'\nStart at [0.53, 0.29]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.29], [0.57, 0.3], [0.57, 0.37], [0.59, 0.35], [0.63, 0.5], [0.61, 0.58], [0.59, 0.57], [0.59, 0.62], [0.5, 0.5], [0.56, 0.52]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0013_user_0016_scene_0007_cfg_0004_104422070011#0.jpg", + "gt": "[[340, 104], [366, 109], [366, 133], [379, 125], [401, 181], [392, 210], [377, 204], [377, 223], [320, 180], [360, 189]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "71190_exterior_image_2_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the orange cloth', and now you are performing subtask 'fold the orange cloth on the table.'\nYou begin at [0.5, 0.79]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.79], [0.51, 0.8], [0.5, 0.82], [0.5, 0.85], [0.49, 0.87], [0.42, 0.97], [0.41, 0.97], [0.42, 0.94], [0.42, 0.91], [0.44, 0.86]]}" + } + ], + "images": "71190_exterior_image_2_left#324", + "gt": "[[159, 142], [164, 144], [160, 147], [160, 153], [158, 157], [134, 174], [131, 175], [133, 170], [133, 163], [142, 154]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0010_user_0007_scene_0003_cfg_0005_105422061350#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the handset of the telephone on the corresponding phone cradle', and now you are performing subtask 'place the handset of the telephone on the corresponding phone cradle.'\nBegin at [0.43, 0.14] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.14], [0.36, 0.34], [0.31, 0.39], [0.3, 0.61], [0.31, 0.38], [0.32, 0.31], [0.37, 0.24], [0.4, 0.26], [0.34, 0.32], [0.37, 0.25]]}" + } + ], + "images": "RH20T_cfg5_task_0010_user_0007_scene_0003_cfg_0005_105422061350#0", + "gt": "[[273, 51], [228, 123], [197, 141], [191, 218], [197, 138], [208, 113], [239, 85], [257, 93], [219, 117], [235, 89]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "40628_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the wooden spatula to mix the contents in the right silver dish', and now you are performing subtask 'use the wooden spatula to mix the contents in the right silver dish.'\nStarting at [0.36, 0.31], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.31], [0.35, 0.46], [0.37, 0.49], [0.46, 0.18], [0.46, 0.35], [0.52, 0.4], [0.48, 0.42], [0.41, 0.28], [0.41, 0.4], [0.38, 0.35]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/40628_exterior_image_1_left#0.jpg", + "gt": "[[114.5, 56.5], [113.5, 82.5], [119.5, 87.5], [148.5, 32.5], [148.5, 63.5], [165.5, 71.5], [154.5, 75.5], [130.5, 50.5], [131.5, 72.5], [121.5, 62.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0009_user_0010_scene_0006_cfg_0004_104122062295#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take out a Hanoi block and throw it aside', and now you are performing subtask 'pick up the Hanoi block from the surface of the wooden table.'\nYou start from [0.41, 0.19]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.19], [0.36, 0.18], [0.36, 0.27], [0.38, 0.28], [0.39, 0.32], [0.38, 0.37], [0.32, 0.4], [0.33, 0.46], [0.33, 0.49], [0.32, 0.47]]}" + } + ], + "images": "RH20T_cfg4_task_0009_user_0010_scene_0006_cfg_0004_104122062295#0", + "gt": "[[262, 68], [231, 64], [230, 97], [246, 101], [249, 114], [240, 132], [206, 144], [214, 165], [210, 175], [204, 169]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "32646_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the cloth on the table', and now you are performing subtask 'again unfold the cloth on the table.'\nWith start location [0.62, 0.38], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.38], [0.62, 0.39], [0.62, 0.37], [0.61, 0.38], [0.55, 0.37], [0.55, 0.38], [0.65, 0.24], [0.73, 0.14], [0.74, 0.13], [0.73, 0.15]]}" + } + ], + "images": "32646_exterior_image_1_left#153", + "gt": "[[199, 68], [197, 70], [199, 66], [195, 68], [175, 67], [176, 68], [208, 43], [234, 25], [236, 24], [233, 27]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0046_user_0020_scene_0004_cfg_0006_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the cup on the cup rack', and now you are performing subtask 'put the cup on the cup rack.'\nStarting at [0.53, 0.31], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.31], [0.56, 0.31], [0.56, 0.36], [0.57, 0.34], [0.6, 0.37], [0.58, 0.42], [0.64, 0.2], [0.67, 0.2], [0.64, 0.21], [0.66, 0.24]]}" + } + ], + "images": "RH20T_cfg6_task_0046_user_0020_scene_0004_cfg_0006_104122064161#0", + "gt": "[[338, 110], [356, 113], [358, 131], [363, 122], [386, 133], [372, 151], [412, 73], [427, 73], [411, 76], [422, 87]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0037_user_0014_scene_0001_cfg_0007_104122063633#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line', and now you are performing subtask 'pick up the first block from the table, transfer the first block from the table to the stacking position, place the first block at the stacking position.'\nFrom start point [0.37, 0.34], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.34], [0.34, 0.38], [0.33, 0.41], [0.33, 0.44], [0.33, 0.43], [0.33, 0.45], [0.33, 0.41], [0.34, 0.41], [0.35, 0.41], [0.34, 0.4]]}" + } + ], + "images": "RH20T_cfg7_task_0037_user_0014_scene_0001_cfg_0007_104122063633#0", + "gt": "[[234, 121], [217, 135], [212, 148], [214, 157], [212, 156], [211, 162], [213, 149], [216, 147], [226, 149], [220, 145]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0071_user_0014_scene_0003_cfg_0002_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close a sliding window', and now you are performing subtask 'close a sliding window.'\nStart at [0.53, 0.31]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.31], [0.51, 0.31], [0.51, 0.28], [0.52, 0.29], [0.55, 0.36], [0.56, 0.48], [0.54, 0.55], [0.59, 0.64], [0.62, 0.57], [0.53, 0.43]]}" + } + ], + "images": "RH20T_cfg2_task_0071_user_0014_scene_0003_cfg_0002_104122061850#0", + "gt": "[[336, 110], [329, 110], [327, 100], [334, 106], [350, 130], [358, 173], [348, 198], [378, 229], [399, 205], [342, 155]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "5907_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker on the desk', and now you are performing subtask 'put the marker on the desk.'\nThe trajectory starts at [0.29, 0.23]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.29, 0.23], [0.45, 0.46], [0.52, 0.44], [0.54, 0.51], [0.51, 0.58], [0.51, 0.47], [0.49, 0.49], [0.53, 0.71], [0.48, 0.47], [0.48, 0.47]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/5907_exterior_image_2_left#0.jpg", + "gt": "[[92, 42], [145, 83], [166, 79], [174, 91], [164, 105], [164, 84], [157, 89], [171, 128], [155, 85], [155, 85]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the block on the plane, transfer the block from the plane to the other blocks, place the block on the other blocks.'\nYour initial coordinate is [0.45, 0.42]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.42], [0.34, 0.55], [0.34, 0.59], [0.35, 0.57], [0.36, 0.61], [0.37, 0.55], [0.44, 0.53], [0.47, 0.56], [0.48, 0.61], [0.41, 0.69]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_037522062165#0.jpg", + "gt": "[[289, 152], [219, 198], [219, 212], [224, 207], [228, 218], [238, 198], [283, 192], [301, 201], [307, 218], [264, 248]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0067_user_0010_scene_0010_cfg_0003_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press a button from top to bottom with obstacles', and now you are performing subtask 'press a button from top to bottom with obstacles.'\nFrom start point [0.38, 0.0], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.0, 0.0], [0.03, 0.16], [0.0, 0.15], [0.05, 0.28], [0.01, 0.31], [0.03, 0.36], [0.0, 0.21], [0.0, 0.0], [0.43, 0.0]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg3_task_0067_user_0010_scene_0010_cfg_0003_104122062823#0.jpg", + "gt": "[[242, 0], [0, 0], [16, 57], [0, 54], [35, 101], [8, 113], [17, 128], [1, 76], [1, 0], [278, 0]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0078_user_0014_scene_0008_cfg_0006_104122063633#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the microwave door', and now you are performing subtask 'close the microwave door.'\nFrom start point [0.36, 0.33], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.33], [0.29, 0.31], [0.28, 0.24], [0.3, 0.19], [0.38, 0.13], [0.39, 0.1], [0.37, 0.08], [0.36, 0.1], [0.35, 0.13], [0.38, 0.17]]}" + } + ], + "images": "RH20T_cfg6_task_0078_user_0014_scene_0008_cfg_0006_104122063633#0", + "gt": "[[232, 119], [183, 113], [176, 87], [192, 69], [242, 47], [249, 36], [239, 29], [228, 35], [226, 47], [242, 63]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "33388_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the mouth of the flask', and now you are performing subtask 'open the mouth of the flask.'\nYou start from [0.83, 0.01]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.83, 0.01], [0.85, 0.13], [0.84, 0.11], [0.82, 0.21], [0.82, 0.38], [0.8, 0.39], [0.81, 0.41], [0.77, 0.41], [0.82, 0.31], [0.83, 0.17]]}" + } + ], + "images": "33388_exterior_image_1_left#0", + "gt": "[[265, 2], [272, 23], [269, 20], [264, 38], [264, 68], [255, 71], [259, 74], [247, 74], [262, 55], [267, 30]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "58247_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the microwave door, place the bowl inside the microwave', and now you are performing subtask 'pull the microwave door.'\nWith start location [0.45, 0.35], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.35], [0.46, 0.43], [0.51, 0.5], [0.51, 0.64], [0.51, 0.61], [0.47, 0.59], [0.48, 0.62], [0.53, 0.43], [0.54, 0.6], [0.53, 0.6]]}" + } + ], + "images": "58247_exterior_image_1_left#188", + "gt": "[[145, 63], [147, 77], [163, 90], [163, 116], [163, 109], [151, 106], [153, 111], [170, 78], [174, 108], [170, 108]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0013_user_0014_scene_0007_cfg_0006_104122061602#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the block on the scale', and now you are performing subtask 'place the block on the scale.'\nYou start from [0.7, 0.29]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.29], [0.62, 0.46], [0.61, 0.56], [0.61, 0.41], [0.61, 0.4], [0.59, 0.42], [0.61, 0.39], [0.67, 0.36], [0.68, 0.33], [0.67, 0.33]]}" + } + ], + "images": "RH20T_cfg6_task_0013_user_0014_scene_0007_cfg_0006_104122061602#0", + "gt": "[[446, 106], [400, 164], [389, 201], [393, 149], [391, 143], [378, 151], [390, 139], [429, 129], [435, 120], [428, 119]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "87512_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the duster to wipe the writing on the white board', and now you are performing subtask 'use the duster to wipe the writing on the white board.'\nThe trajectory starts at [0.89, 0.04]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.89, 0.04], [0.61, 0.23], [0.5, 0.42], [0.62, 0.18], [0.6, 0.01], [0.56, 0.0], [0.55, 0.29], [0.5, 0.39], [0.62, 0.25], [0.62, 0.25]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/87512_exterior_image_2_left#0.jpg", + "gt": "[[284, 7], [194, 41], [159, 75], [199, 33], [192, 2], [179, 0], [177, 52], [159, 70], [200, 45], [200, 45]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "2362_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the wallet from the table and put it in the black bag', and now you are performing subtask 'pick up the wallet from the table.'\nYour starting position is [0.89, 0.38]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.89, 0.38], [0.71, 0.73], [0.62, 0.81], [0.62, 0.85], [0.6, 0.74], [0.63, 0.76], [0.64, 0.71], [0.67, 0.75], [0.69, 0.74], [0.69, 0.74]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/2362_exterior_image_1_left#0.jpg", + "gt": "[[286.0, 68.5], [227.0, 131.5], [198.0, 145.5], [197.0, 153.5], [191.0, 132.5], [201.0, 136.5], [204.0, 127.5], [214.0, 134.5], [222.0, 132.5], [222.0, 132.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "11748_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the white plate from the dishwasher and place it on top of the white plates on the counter', and now you are performing subtask 'remove the white plate from the dishwasher and place it on top of the white plates on the counter.'\nWith start location [0.58, 0.29], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.29], [0.39, 0.46], [0.31, 0.76], [0.36, 0.84], [0.42, 0.55], [0.44, 0.5], [0.44, 0.56], [0.45, 0.45], [0.43, 0.55], [0.38, 0.54]]}" + } + ], + "images": "11748_exterior_image_1_left#0", + "gt": "[[187, 53], [124, 83], [100, 137], [115, 152], [133, 99], [142, 90], [142, 101], [143, 81], [139, 99], [122, 98]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0054_user_0007_scene_0009_cfg_0002_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp the moving object', and now you are performing subtask 'pick up the moving object directly under the robotic arm.'\nBegin at [0.59, 0.38] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.38], [0.6, 0.38], [0.6, 0.4], [0.63, 0.44], [0.65, 0.42], [0.64, 0.39], [0.64, 0.39], [0.63, 0.42], [0.65, 0.44], [0.66, 0.46]]}" + } + ], + "images": "RH20T_cfg2_task_0054_user_0007_scene_0009_cfg_0002_104122061850#0", + "gt": "[[380, 137], [383, 135], [387, 144], [401, 158], [413, 151], [412, 142], [409, 142], [405, 150], [417, 158], [421, 164]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0037_user_0014_scene_0001_cfg_0007_037522061512#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line', and now you are performing subtask 'transfer the first block from the table to the stacking position, place the first block at the stacking position, pick up the second block from the table, transfer the second block from the table to the stacking position.'\nYou start from [0.45, 0.32]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.32], [0.44, 0.31], [0.42, 0.31], [0.41, 0.32], [0.41, 0.34], [0.41, 0.34], [0.41, 0.31], [0.42, 0.27], [0.43, 0.26], [0.45, 0.28]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0037_user_0014_scene_0001_cfg_0007_037522061512#178.jpg", + "gt": "[[289, 115], [282, 110], [270, 111], [264, 115], [262, 124], [263, 124], [263, 110], [271, 96], [277, 93], [287, 102]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0037_user_0014_scene_0008_cfg_0006_104122064161#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the block on the other blocks, pick up the block on the table, transfer the block from the table to the stacked blocks, place the block on the stacked blocks, pick up the block on the table.'\nYour initial coordinate is [0.48, 0.22]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.22], [0.5, 0.22], [0.5, 0.27], [0.48, 0.41], [0.51, 0.28], [0.5, 0.16], [0.48, 0.16], [0.48, 0.18], [0.52, 0.19], [0.55, 0.28]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0037_user_0014_scene_0008_cfg_0006_104122064161#206.jpg", + "gt": "[[309, 78], [317, 78], [322, 97], [305, 147], [328, 101], [320, 57], [308, 56], [306, 65], [333, 69], [353, 101]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0061_user_0007_scene_0008_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the first block out of the box and place it on the table, then take the second block and place it on top of the first block', and now you are performing subtask 'pick up the first block in the box, transfer the first block from the box to above the table, place the first block on the table.'\nFrom start point [0.5, 0.37], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.5, 0.43], [0.51, 0.45], [0.51, 0.52], [0.5, 0.58], [0.51, 0.51], [0.49, 0.41], [0.49, 0.34], [0.49, 0.38], [0.49, 0.42]]}" + } + ], + "images": "RH20T_cfg5_task_0061_user_0007_scene_0008_cfg_0005_104122063678#0", + "gt": "[[323, 132], [319, 153], [324, 161], [324, 187], [319, 210], [325, 183], [316, 149], [315, 123], [313, 138], [314, 152]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0058_user_0010_scene_0010_cfg_0003_036422060909#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn off the power strips', and now you are performing subtask 'press the button on the power strip above.'\nYour starting position is [0.35, 0.28]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.28], [0.38, 0.27], [0.37, 0.28], [0.38, 0.29], [0.37, 0.29], [0.37, 0.3], [0.38, 0.3], [0.37, 0.3], [0.38, 0.29], [0.36, 0.26]]}" + } + ], + "images": "RH20T_cfg3_task_0058_user_0010_scene_0010_cfg_0003_036422060909#193", + "gt": "[[221, 100], [241, 98], [236, 102], [240, 106], [237, 104], [238, 108], [240, 108], [239, 109], [241, 106], [231, 92]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0032_user_0011_scene_0001_cfg_0003_104122062823#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'slide the outer casing onto the gift box', and now you are performing subtask 'slide the outer casing onto the gift box.'\nBegin at [0.41, 0.29] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.29], [0.47, 0.35], [0.49, 0.4], [0.47, 0.44], [0.41, 0.41], [0.35, 0.27], [0.31, 0.23], [0.41, 0.29], [0.43, 0.32], [0.4, 0.35]]}" + } + ], + "images": "RH20T_cfg3_task_0032_user_0011_scene_0001_cfg_0003_104122062823#267", + "gt": "[[260, 103], [300, 127], [313, 144], [301, 158], [261, 147], [223, 96], [196, 82], [265, 103], [274, 115], [259, 127]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "37646_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the glass lid from the top of the black pot', and now you are performing subtask 'remove the glass lid from the top of the black pot.'\nBegin at [0.68, 0.13] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.13], [0.61, 0.48], [0.59, 0.51], [0.58, 0.56], [0.57, 0.52], [0.55, 0.52], [0.52, 0.56], [0.49, 0.69], [0.51, 0.73], [0.5, 0.69]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/37646_exterior_image_2_left#0.jpg", + "gt": "[[219, 24], [196, 86], [189, 92], [186, 101], [183, 93], [175, 94], [167, 100], [157, 124], [162, 131], [160, 125]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "33838_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the lid off of the pot', and now you are performing subtask 'take the lid off of the pot.'\nYou start from [1.0, 0.14]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[1.0, 0.14], [0.81, 0.45], [0.78, 0.57], [0.75, 0.62], [0.77, 0.57], [0.77, 0.59], [0.74, 0.73], [0.71, 0.83], [0.71, 0.86], [0.78, 0.82]]}" + } + ], + "images": "33838_exterior_image_2_left#0", + "gt": "[[320, 25], [258, 81], [249, 102], [241, 112], [246, 103], [247, 106], [238, 131], [226, 150], [227, 155], [250, 148]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0050_user_0020_scene_0008_cfg_0007_104122061602#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the plate on the bowl rack', and now you are performing subtask 'put the plate on the bowl rack.'\nThe trajectory starts at [0.7, 0.32]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.32], [0.73, 0.51], [0.74, 0.43], [0.7, 0.13], [0.65, 0.0], [0.63, 0.0], [0.64, 0.09], [0.69, 0.12], [0.69, 0.08], [0.69, 0.17]]}" + } + ], + "images": "RH20T_cfg7_task_0050_user_0020_scene_0008_cfg_0007_104122061602#0", + "gt": "[[448, 116], [470, 184], [472, 156], [449, 48], [419, 0], [401, 0], [411, 32], [439, 44], [443, 29], [443, 62]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "82330_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the marker from the cup to the table', and now you are performing subtask 'move the marker from the cup to the table.'\nFrom start point [0.37, 0.14], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.14], [0.62, 0.25], [0.67, 0.35], [0.67, 0.5], [0.7, 0.45], [0.7, 0.53], [0.68, 0.29], [0.6, 0.27], [0.55, 0.34], [0.56, 0.39]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/82330_exterior_image_1_left#0.jpg", + "gt": "[[119.0, 25.5], [198.0, 45.5], [215.0, 62.5], [214.0, 90.5], [225.0, 81.5], [224.0, 95.5], [218.0, 51.5], [193.0, 48.5], [175.0, 61.5], [178.0, 69.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0010_user_0010_scene_0002_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the handset of the telephone on the corresponding phone cradle', and now you are performing subtask 'place the handset of the telephone on the corresponding phone cradle.'\nYou begin at [0.5, 0.37]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.53, 0.48], [0.51, 0.52], [0.5, 0.65], [0.45, 0.73], [0.46, 0.79], [0.43, 0.66], [0.46, 0.59], [0.4, 0.52], [0.44, 0.48]]}" + } + ], + "images": "RH20T_cfg5_task_0010_user_0010_scene_0002_cfg_0005_104122063678#0", + "gt": "[[323, 132], [338, 173], [324, 186], [317, 234], [287, 263], [293, 286], [273, 236], [295, 213], [256, 186], [284, 173]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "66631_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the towel to wipe the table', and now you are performing subtask 'unfold the towel.'\nFrom start point [1.0, 0.34], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[1.0, 0.34], [0.8, 0.42], [0.76, 0.49], [0.74, 0.26], [0.72, 0.17], [0.7, 0.16], [0.66, 0.31], [0.69, 0.31], [0.74, 0.43], [0.68, 0.4]]}" + } + ], + "images": "66631_exterior_image_1_left#0", + "gt": "[[319, 61], [257, 76], [243, 89], [238, 47], [231, 31], [224, 29], [210, 55], [221, 56], [236, 78], [216, 72]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "592_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer then place the green pear inside the drawer', and now you are performing subtask 'pick up the green pear from the box on the countertop.'\nYou start from [0.75, 0.44]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.75, 0.44], [0.59, 0.19], [0.65, 0.26], [0.61, 0.22], [0.64, 0.32], [0.57, 0.43], [0.57, 0.51], [0.71, 0.47], [0.63, 0.29], [0.53, 0.22]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/592_exterior_image_1_left#0.jpg", + "gt": "[[240, 79], [189, 35], [209, 46], [195, 40], [204, 57], [183, 78], [181, 91], [228, 84], [203, 53], [169, 40]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0013_user_0020_scene_0004_cfg_0007_037522061512#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the block on the scale', and now you are performing subtask 'place the block on the scale.'\nYou start from [0.41, 0.24]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.24], [0.34, 0.43], [0.33, 0.5], [0.37, 0.36], [0.43, 0.3], [0.43, 0.28], [0.43, 0.31], [0.41, 0.24], [0.43, 0.27], [0.43, 0.27]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0013_user_0020_scene_0004_cfg_0007_037522061512#0.jpg", + "gt": "[[261, 86], [217, 153], [211, 180], [239, 130], [275, 108], [272, 102], [277, 112], [263, 87], [273, 97], [273, 97]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "25731_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the rubik's cube on the top of the yellow block', and now you are performing subtask 'put the rubik's cube on the top of the yellow block.'\nBegin at [0.2, 0.21] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.2, 0.21], [0.34, 0.34], [0.39, 0.53], [0.37, 0.6], [0.34, 0.29], [0.39, 0.34], [0.38, 0.46], [0.42, 0.48], [0.38, 0.38], [0.33, 0.09]]}" + } + ], + "images": "25731_exterior_image_2_left#0", + "gt": "[[64, 38], [108, 61], [125, 95], [119, 108], [109, 53], [124, 61], [123, 83], [133, 87], [120, 69], [106, 17]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0037_user_0007_scene_0007_cfg_0002_105422061350#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the fifth block from the surface, transfer the fifth block from the surface to the top of the fourth block, place the fifth block on top of the fourth block.'\nYour initial coordinate is [0.42, 0.45]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.45], [0.41, 0.47], [0.42, 0.49], [0.43, 0.45], [0.45, 0.44], [0.51, 0.46], [0.52, 0.48], [0.5, 0.48], [0.51, 0.44], [0.54, 0.42]]}" + } + ], + "images": "RH20T_cfg2_task_0037_user_0007_scene_0007_cfg_0002_105422061350#824", + "gt": "[[267, 162], [263, 169], [270, 177], [276, 162], [288, 158], [326, 165], [330, 173], [321, 172], [326, 157], [343, 150]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0007_user_0016_scene_0008_cfg_0003_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'holding a small block with the gripper and sweeping it from left to right on the table', and now you are performing subtask 'holding a small block with the gripper and sweeping it from left to right on the table.'\nYou begin at [0.38, 0.0]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.49, 0.02], [0.47, 0.08], [0.49, 0.08], [0.52, 0.14], [0.51, 0.19], [0.53, 0.26], [0.38, 0.21], [0.37, 0.16], [0.41, 0.14]]}" + } + ], + "images": "RH20T_cfg3_task_0007_user_0016_scene_0008_cfg_0003_104122062823#0", + "gt": "[[242, 0], [316, 8], [304, 27], [316, 28], [332, 51], [326, 69], [338, 94], [241, 77], [236, 58], [264, 52]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "53935_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push the chair, press the box', and now you are performing subtask 'press the box.'\nYou begin at [0.73, 0.29]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.73, 0.29], [0.69, 0.29], [0.74, 0.37], [0.74, 0.34], [0.7, 0.33], [0.67, 0.38], [0.66, 0.36], [0.69, 0.4], [0.64, 0.38], [0.72, 0.51]]}" + } + ], + "images": "53935_exterior_image_1_left#0", + "gt": "[[233, 52], [222, 53], [238, 66], [237, 61], [224, 59], [215, 69], [212, 65], [221, 72], [205, 68], [230, 92]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0044_user_0010_scene_0002_cfg_0005_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up a bag of things', and now you are performing subtask 'pick up a bag of things.'\nThe trajectory starts at [0.47, 0.4]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.4], [0.38, 0.52], [0.34, 0.55], [0.3, 0.64], [0.3, 0.65], [0.3, 0.61], [0.32, 0.47], [0.32, 0.45], [0.35, 0.44], [0.4, 0.41]]}" + } + ], + "images": "RH20T_cfg5_task_0044_user_0010_scene_0002_cfg_0005_037522062165#0", + "gt": "[[298, 143], [245, 187], [215, 197], [190, 231], [194, 234], [192, 218], [208, 171], [203, 163], [227, 159], [253, 146]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "71776_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the yellow mug to the left', and now you are performing subtask 'transfer the yellow mug from the right side of the table to the left side.'\nBegin at [0.73, 0.15] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.73, 0.15], [0.64, 0.16], [0.62, 0.21], [0.55, 0.23], [0.56, 0.23], [0.53, 0.25], [0.52, 0.35], [0.47, 0.35], [0.5, 0.36], [0.5, 0.36]]}" + } + ], + "images": "71776_exterior_image_1_left#0", + "gt": "[[235, 27], [206, 29], [197, 37], [175, 41], [179, 42], [171, 45], [167, 63], [151, 63], [159, 64], [159, 64]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0073_user_0014_scene_0001_cfg_0007_037522061512#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the things in the drawer, close the drawer', and now you are performing subtask 'place the item in the drawer, pick up the round object from the table, place the prototype object in the drawer, pick up the square object from the table.'\nThe trajectory starts at [0.35, 0.28]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.28], [0.28, 0.45], [0.27, 0.53], [0.28, 0.59], [0.33, 0.29], [0.33, 0.25], [0.35, 0.21], [0.4, 0.17], [0.35, 0.17], [0.29, 0.35]]}" + } + ], + "images": "RH20T_cfg7_task_0073_user_0014_scene_0001_cfg_0007_037522061512#208", + "gt": "[[224, 102], [179, 162], [172, 192], [179, 214], [210, 106], [209, 91], [221, 75], [253, 60], [227, 62], [187, 127]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_037522061512#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'pick up the block on the flat surface, transfer the block from the flat surface to the block board, place the block on the block board, pick up the block on the flat surface.'\nWith start location [0.41, 0.24], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.24], [0.35, 0.21], [0.38, 0.27], [0.38, 0.29], [0.39, 0.23], [0.4, 0.3], [0.42, 0.36], [0.39, 0.31], [0.36, 0.29], [0.36, 0.3]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_037522061512#0.jpg", + "gt": "[[261, 86], [226, 76], [246, 98], [245, 105], [249, 82], [256, 108], [270, 131], [249, 110], [233, 105], [230, 109]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_038522062547#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'pick up the second square from the right side of the table, transfer the second square from the right side of the table to the top of the first square, place the second square beside the first square.'\nThe trajectory starts at [0.67, 0.56]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.67, 0.56], [0.74, 0.57], [0.74, 0.59], [0.73, 0.55], [0.64, 0.49], [0.62, 0.49], [0.62, 0.51], [0.61, 0.52], [0.62, 0.49], [0.67, 0.5]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_038522062547#226.jpg", + "gt": "[[431, 200], [472, 205], [472, 212], [464, 198], [410, 176], [397, 176], [400, 183], [391, 188], [397, 178], [426, 180]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "31401_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the purple packet out of the sink and place it on the countertop', and now you are performing subtask 'take the purple packet out of the sink and place it on the countertop.'\nWith start location [0.3, 0.34], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.3, 0.34], [0.39, 0.57], [0.48, 0.9], [0.52, 0.9], [0.52, 0.95], [0.52, 0.87], [0.51, 0.83], [0.47, 0.82], [0.33, 0.86], [0.35, 0.77]]}" + } + ], + "images": "31401_exterior_image_1_left#0", + "gt": "[[95, 61], [125, 103], [154, 162], [167, 162], [168, 171], [167, 157], [163, 150], [150, 148], [106, 154], [112, 139]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "63529_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn off the light', and now you are performing subtask 'turn off the light.'\nStarting at [0.0, 0.64], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.0, 0.64], [0.15, 0.64], [0.25, 0.6], [0.35, 0.52], [0.46, 0.33], [0.44, 0.42], [0.41, 0.38], [0.36, 0.39], [0.27, 0.58], [0.29, 0.57]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/63529_exterior_image_2_left#0.jpg", + "gt": "[[0, 116], [47, 116], [80, 108], [112, 93], [147, 59], [142, 76], [130, 68], [115, 71], [85, 105], [92, 103]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "5154_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the marker from the center of the table to the bottom of the table', and now you are performing subtask 'move the marker from the center of the table to the bottom of the table.'\nStart at [0.85, 0.02]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.85, 0.02], [0.73, 0.09], [0.61, 0.27], [0.55, 0.41], [0.51, 0.47], [0.51, 0.49], [0.52, 0.52], [0.63, 0.48], [0.65, 0.4], [0.64, 0.41]]}" + } + ], + "images": "5154_exterior_image_2_left#0", + "gt": "[[271, 4], [235, 16], [194, 48], [177, 73], [163, 84], [163, 89], [166, 93], [202, 87], [209, 72], [204, 74]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_037522061512#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'place the object on the surface, pick up the object in the container, transfer the object from the container to the surface, place the item on the table.'\nYour starting position is [0.36, 0.31]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.31], [0.35, 0.38], [0.35, 0.44], [0.36, 0.46], [0.35, 0.35], [0.39, 0.29], [0.4, 0.33], [0.41, 0.37], [0.41, 0.34], [0.39, 0.32]]}" + } + ], + "images": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_037522061512#450", + "gt": "[[231, 110], [222, 135], [224, 157], [230, 165], [227, 126], [251, 105], [259, 118], [261, 133], [260, 124], [251, 117]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0058_user_0015_scene_0003_cfg_0001_038522063145#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn off the power strips', and now you are performing subtask 'press the fourth button on the power socket.'\nYour initial coordinate is [0.79, 0.56]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.79, 0.56], [0.78, 0.54], [0.78, 0.54], [0.77, 0.57], [0.78, 0.54], [0.78, 0.54], [0.78, 0.55], [0.78, 0.54], [0.73, 0.42], [0.73, 0.42]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0058_user_0015_scene_0003_cfg_0001_038522063145#265.jpg", + "gt": "[[504, 201], [501, 194], [498, 194], [493, 205], [499, 195], [502, 195], [502, 197], [498, 195], [464, 152], [464, 152]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_038522063145#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the second block on top of the first block, pick up the third block from the desktop, transfer the third block from the desktop to the second block above the second position, place the third block on top of the second block.'\nStart at [0.71, 0.24]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.71, 0.24], [0.71, 0.13], [0.64, 0.15], [0.57, 0.22], [0.57, 0.34], [0.61, 0.14], [0.64, 0.12], [0.72, 0.17], [0.72, 0.14], [0.72, 0.14]]}" + } + ], + "images": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_038522063145#305", + "gt": "[[452, 85], [457, 47], [411, 53], [368, 81], [368, 121], [390, 51], [410, 45], [458, 63], [459, 52], [459, 52]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "56568_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'slide the black lid off', and now you are performing subtask 'slide the black lid to one side of the table.'\nYou begin at [0.54, 0.46]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.46], [0.54, 0.44], [0.53, 0.44], [0.52, 0.45], [0.52, 0.46], [0.53, 0.46], [0.53, 0.46], [0.53, 0.46], [0.53, 0.4], [0.54, 0.39]]}" + } + ], + "images": "56568_exterior_image_2_left#132", + "gt": "[[172, 82], [172, 79], [170, 79], [167, 81], [167, 82], [169, 82], [169, 83], [171, 82], [171, 72], [172, 71]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "19968_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer on the right, take out the watermelon toy and the green toy and place them on the countertop, close the drawer', and now you are performing subtask 'put the green toy on the countertop.'\nYour initial coordinate is [0.41, 0.91]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.91], [0.46, 0.85], [0.5, 0.78], [0.59, 0.7], [0.66, 0.74], [0.67, 0.79], [0.65, 0.76], [0.57, 0.72], [0.47, 0.71], [0.46, 0.73]]}" + } + ], + "images": "19968_exterior_image_1_left#285", + "gt": "[[130, 164], [148, 153], [160, 141], [189, 126], [210, 133], [213, 143], [208, 137], [184, 129], [152, 128], [146, 132]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0045_user_0010_scene_0004_cfg_0005_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the brush on the pen rack', and now you are performing subtask 'place the brush on the pen rack.'\nYou start from [0.45, 0.29]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.29], [0.45, 0.33], [0.39, 0.45], [0.4, 0.4], [0.44, 0.36], [0.47, 0.21], [0.5, 0.19], [0.49, 0.21], [0.49, 0.18], [0.43, 0.2]]}" + } + ], + "images": "RH20T_cfg5_task_0045_user_0010_scene_0004_cfg_0005_036422060215#0", + "gt": "[[285, 106], [286, 120], [250, 162], [253, 145], [280, 129], [301, 74], [321, 69], [313, 75], [315, 66], [274, 71]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "18044_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the packet on the box', and now you are performing subtask 'pick up the second packet from the counter.'\nYou begin at [0.5, 0.42]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.42], [0.55, 0.47], [0.58, 0.46], [0.65, 0.56], [0.69, 0.79], [0.66, 0.83], [0.62, 0.84], [0.61, 0.73], [0.66, 0.73], [0.66, 0.73]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/18044_exterior_image_2_left#0.jpg", + "gt": "[[159, 75], [175, 84], [187, 83], [208, 101], [221, 142], [212, 150], [200, 152], [196, 132], [211, 132], [211, 132]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "63177_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the objects on the plate and put them in the bin', and now you are performing subtask 'take the object from the plate and move it to the bowl.'\nThe trajectory starts at [0.62, 0.27]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.27], [0.6, 0.34], [0.59, 0.29], [0.57, 0.19], [0.58, 0.19], [0.57, 0.2], [0.57, 0.19], [0.58, 0.16], [0.61, 0.13], [0.61, 0.13]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/63177_exterior_image_1_left#202.jpg", + "gt": "[[199, 49], [192, 62], [190, 52], [181, 35], [185, 34], [184, 36], [183, 34], [186, 29], [195, 24], [195, 24]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "14782_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'uncover the lunch box, pour the sweets inside it into a mug and then put it back on the table', and now you are performing subtask 'place the lunch box back on the table.'\nStart at [0.57, 0.48]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.48], [0.59, 0.47], [0.59, 0.41], [0.82, 0.46], [0.77, 0.33], [0.64, 0.37], [0.7, 0.42], [0.7, 0.44], [0.67, 0.4], [0.62, 0.42]]}" + } + ], + "images": "14782_exterior_image_1_left#207", + "gt": "[[182, 86], [189, 85], [190, 74], [262, 83], [247, 59], [206, 66], [224, 76], [223, 80], [215, 72], [200, 76]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "65809_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the lid on the pot', and now you are performing subtask 'place the lid on the pot.'\nFrom start point [0.22, 0.04], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.22, 0.04], [0.31, 0.04], [0.47, 0.21], [0.5, 0.31], [0.51, 0.39], [0.46, 0.33], [0.4, 0.41], [0.38, 0.39], [0.38, 0.38], [0.37, 0.36]]}" + } + ], + "images": "65809_exterior_image_1_left#0", + "gt": "[[71, 7], [98, 7], [150, 37], [161, 55], [162, 70], [148, 59], [127, 73], [120, 71], [123, 68], [119, 64]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "63429_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker in the cup', and now you are performing subtask 'place the marker in the cup.'\nYour initial coordinate is [0.31, 0.17]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.31, 0.17], [0.35, 0.26], [0.34, 0.29], [0.29, 0.33], [0.29, 0.37], [0.29, 0.24], [0.32, 0.2], [0.35, 0.23], [0.34, 0.17], [0.36, 0.16]]}" + } + ], + "images": "63429_exterior_image_2_left#69", + "gt": "[[98, 31], [112, 46], [110, 53], [93, 59], [93, 66], [93, 44], [103, 36], [111, 42], [109, 31], [115, 29]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "20220_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the tap', and now you are performing subtask 'move the tap.'\nYou begin at [0.68, 0.0]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.0], [0.73, 0.16], [0.74, 0.29], [0.77, 0.31], [0.72, 0.28], [0.7, 0.3], [0.75, 0.34], [0.68, 0.32], [0.64, 0.33], [0.66, 0.34]]}" + } + ], + "images": "20220_exterior_image_1_left#0", + "gt": "[[217, 0], [233, 28], [238, 53], [247, 56], [230, 51], [223, 54], [240, 61], [216, 58], [204, 59], [211, 61]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0061_user_0020_scene_0001_cfg_0007_104122063633#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the lid of the gift box on the table, then take everything out of the gift box and place it all on the table', and now you are performing subtask 'place the object on the tabletop, pick up the block from the gift box, transfer the block from the gift box to the tabletop, place the block on the tabletop.'\nYour initial coordinate is [0.32, 0.44]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.44], [0.3, 0.41], [0.29, 0.41], [0.28, 0.44], [0.3, 0.61], [0.3, 0.36], [0.31, 0.33], [0.31, 0.39], [0.31, 0.32], [0.34, 0.31]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0061_user_0020_scene_0001_cfg_0007_104122063633#492.jpg", + "gt": "[[205, 160], [192, 148], [185, 148], [178, 159], [192, 218], [189, 131], [196, 119], [196, 141], [199, 116], [217, 110]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "51023_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the rectangular block on the table', and now you are performing subtask 'put the rectangular block on the table.'\nStarting at [0.43, 0.01], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.01], [0.45, 0.16], [0.45, 0.25], [0.47, 0.32], [0.47, 0.36], [0.5, 0.22], [0.49, 0.35], [0.48, 0.2], [0.49, 0.23], [0.49, 0.2]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/51023_exterior_image_1_left#0.jpg", + "gt": "[[139, 2], [145, 28], [144, 45], [152, 58], [149, 64], [159, 40], [156, 63], [153, 36], [157, 41], [157, 36]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0096_user_0020_scene_0002_cfg_0006_037522061512#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take something out of a drawer', and now you are performing subtask 'place the item on the surface.'\nStart at [0.41, 0.27]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.27], [0.42, 0.33], [0.41, 0.29], [0.37, 0.25], [0.36, 0.25], [0.34, 0.28], [0.33, 0.26], [0.34, 0.23], [0.33, 0.25], [0.33, 0.23]]}" + } + ], + "images": "RH20T_cfg6_task_0096_user_0020_scene_0002_cfg_0006_037522061512#413", + "gt": "[[265, 98], [267, 120], [261, 103], [237, 90], [231, 89], [216, 102], [213, 95], [217, 84], [212, 89], [211, 84]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "66631_exterior_image_2_left#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the towel to wipe the table', and now you are performing subtask 'wipe the table.'\nStarting at [0.59, 0.61], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.61], [0.52, 0.63], [0.57, 0.64], [0.54, 0.54], [0.52, 0.58], [0.54, 0.63], [0.66, 0.64], [0.55, 0.58], [0.54, 0.61], [0.55, 0.46]]}" + } + ], + "images": "66631_exterior_image_2_left#381", + "gt": "[[188, 109], [168, 114], [184, 115], [174, 97], [166, 105], [173, 113], [211, 115], [176, 104], [172, 109], [177, 82]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_104122061850#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the third block on top of the first block, pick up the fourth block on the table, transfer the fourth block from the table to the position above the third block, place the fourth block on top of the third block.'\nYou start from [0.49, 0.43]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.43], [0.56, 0.31], [0.62, 0.37], [0.61, 0.42], [0.6, 0.36], [0.55, 0.3], [0.51, 0.28], [0.5, 0.3], [0.49, 0.38], [0.47, 0.37]]}" + } + ], + "images": "RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_104122061850#521", + "gt": "[[315, 154], [361, 113], [396, 133], [392, 151], [382, 131], [354, 109], [328, 101], [318, 108], [312, 138], [303, 132]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "275_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the white mug on the left side of the counter, put it under the running faucet then place it down on the right side of the counter', and now you are performing subtask 'place the white mug on the right side of the counter.'\nFrom start point [0.1, 0.76], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.1, 0.76], [0.17, 0.63], [0.36, 0.53], [0.49, 0.56], [0.44, 0.46], [0.44, 0.54], [0.47, 0.57], [0.52, 0.42], [0.56, 0.48], [0.6, 0.42]]}" + } + ], + "images": "275_exterior_image_1_left#175", + "gt": "[[31, 137], [54, 113], [114, 95], [156, 100], [140, 83], [142, 98], [152, 102], [167, 76], [180, 86], [193, 75]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0053_user_0014_scene_0003_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use a clamp', and now you are performing subtask 'use a clamp.'\nBegin at [0.63, 0.37] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.65, 0.34], [0.62, 0.05], [0.6, 0.06], [0.59, 0.2], [0.6, 0.17], [0.61, 0.2], [0.62, 0.25], [0.63, 0.06], [0.63, 0.06]]}" + } + ], + "images": "RH20T_cfg1_task_0053_user_0014_scene_0003_cfg_0001_038522062547#0", + "gt": "[[405, 132], [414, 121], [397, 17], [384, 20], [377, 71], [381, 63], [393, 72], [396, 91], [406, 20], [406, 20]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "26333_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the button atop the lid then pick up the lid and put it on the table', and now you are performing subtask 'pick up the lid from the top of the container.'\nBegin at [0.1, 0.2] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.1, 0.2], [0.29, 0.27], [0.47, 0.28], [0.48, 0.3], [0.45, 0.33], [0.46, 0.4], [0.44, 0.46], [0.45, 0.49], [0.46, 0.33], [0.45, 0.43]]}" + } + ], + "images": "26333_exterior_image_2_left#0", + "gt": "[[32, 36], [93, 49], [152, 50], [153, 54], [145, 59], [148, 72], [141, 82], [144, 89], [147, 60], [145, 78]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "71329_exterior_image_2_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker in the mug and then remove it and put it back on the table', and now you are performing subtask 'place the marker back on the table.'\nStart at [0.4, 0.39]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.39], [0.4, 0.3], [0.37, 0.43], [0.31, 0.37], [0.32, 0.44], [0.38, 0.47], [0.44, 0.3], [0.51, 0.4], [0.48, 0.68], [0.47, 0.52]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/71329_exterior_image_2_left#193.jpg", + "gt": "[[129, 71], [127, 54], [119, 77], [99, 67], [102, 80], [121, 85], [140, 54], [164, 72], [155, 123], [152, 93]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "10406_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the green marker inside the red cup', and now you are performing subtask 'place the green marker inside the red cup.'\nYou start from [0.39, 0.23]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.23], [0.53, 0.29], [0.55, 0.55], [0.62, 0.58], [0.63, 0.68], [0.64, 0.54], [0.58, 0.44], [0.63, 0.43], [0.59, 0.5], [0.61, 0.31]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/10406_exterior_image_1_left#0.jpg", + "gt": "[[124.0, 41.5], [171.0, 51.5], [177.0, 98.5], [198.0, 104.5], [201.0, 122.5], [204.0, 97.5], [185.0, 79.5], [202.0, 76.5], [190.0, 90.5], [196.0, 55.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "57625_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'cover the pan with the silver lid', and now you are performing subtask 'cover the pan with the silver lid.'\nWith start location [0.69, 0.02], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.02], [0.68, 0.61], [0.67, 0.49], [0.62, 0.62], [0.6, 0.57], [0.59, 0.64], [0.62, 0.67], [0.6, 0.62], [0.6, 0.51], [0.6, 0.51]]}" + } + ], + "images": "57625_exterior_image_1_left#0", + "gt": "[[221.5, 4.0], [218.5, 109.0], [214.5, 89.0], [197.5, 112.0], [190.5, 103.0], [187.5, 116.0], [197.5, 120.0], [190.5, 111.0], [192.5, 92.0], [192.5, 92.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "77919_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn off the toaster', and now you are performing subtask 'turn off the toaster.'\nWith start location [0.02, 0.22], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.02, 0.22], [0.33, 0.36], [0.43, 0.44], [0.45, 0.52], [0.45, 0.48], [0.45, 0.51], [0.42, 0.53], [0.29, 0.46], [0.27, 0.43], [0.29, 0.42]]}" + } + ], + "images": "77919_exterior_image_1_left#0", + "gt": "[[5, 39], [107, 65], [136, 79], [145, 94], [144, 87], [145, 92], [134, 95], [93, 83], [87, 77], [92, 75]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "44864_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the cabinet door', and now you are performing subtask 'close the cabinet door.'\nStarting at [0.6, 0.21], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.21], [0.63, 0.24], [0.67, 0.33], [0.74, 0.39], [0.74, 0.42], [0.88, 0.42], [0.86, 0.43], [0.81, 0.38], [0.83, 0.39], [0.83, 0.39]]}" + } + ], + "images": "44864_exterior_image_2_left#0", + "gt": "[[193, 38], [203, 44], [214, 60], [236, 70], [238, 76], [282, 75], [275, 77], [258, 68], [266, 71], [266, 71]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "14782_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'uncover the lunch box, pour the sweets inside it into a mug and then put it back on the table', and now you are performing subtask 'pick up the lunch box.'\nThe trajectory starts at [0.33, 0.39]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.39], [0.4, 0.34], [0.46, 0.34], [0.57, 0.45], [0.56, 0.42], [0.64, 0.36], [0.65, 0.38], [0.57, 0.44], [0.57, 0.48], [0.57, 0.48]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/14782_exterior_image_1_left#0.jpg", + "gt": "[[104, 70], [128, 62], [146, 62], [181, 81], [178, 75], [206, 65], [207, 69], [182, 80], [182, 86], [182, 86]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0003_user_0014_scene_0006_cfg_0005_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the red button on the black and white base, press the green button on the gray and white base.'\nThe trajectory starts at [0.48, 0.14]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.14], [0.48, 0.1], [0.55, 0.0], [0.55, 0.11], [0.54, 0.15], [0.53, 0.44], [0.55, 0.5], [0.57, 0.51], [0.56, 0.49], [0.55, 0.44]]}" + } + ], + "images": "RH20T_cfg5_task_0003_user_0014_scene_0006_cfg_0005_104122061850#0", + "gt": "[[305, 50], [307, 37], [353, 0], [350, 40], [343, 53], [339, 159], [355, 181], [366, 185], [356, 178], [351, 157]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0085_user_0007_scene_0002_cfg_0005_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the bubble ring from the assembled bubble ring and ball', and now you are performing subtask 'pick up the bubble ring to separate it from the assembled bubble ring with the ball.'\nWith start location [0.45, 0.3], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.3], [0.38, 0.37], [0.37, 0.4], [0.36, 0.41], [0.35, 0.42], [0.35, 0.46], [0.35, 0.47], [0.35, 0.47], [0.35, 0.48], [0.35, 0.45]]}" + } + ], + "images": "RH20T_cfg5_task_0085_user_0007_scene_0002_cfg_0005_036422060215#0", + "gt": "[[286, 108], [245, 134], [237, 145], [229, 147], [224, 152], [226, 164], [223, 168], [224, 171], [221, 172], [226, 163]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0002_user_0014_scene_0010_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull out a napkin', and now you are performing subtask 'pull out a napkin.'\nBegin at [0.5, 0.37] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.5, 0.37], [0.54, 0.41], [0.57, 0.49], [0.59, 0.56], [0.6, 0.61], [0.6, 0.57], [0.6, 0.54], [0.57, 0.37], [0.56, 0.35]]}" + } + ], + "images": "RH20T_cfg5_task_0002_user_0014_scene_0010_cfg_0005_104122063678#0", + "gt": "[[323, 132], [320, 133], [345, 149], [367, 177], [380, 201], [381, 221], [384, 207], [382, 193], [364, 134], [357, 126]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0089_user_0014_scene_0001_cfg_0004_104122062823#TrajectoryPrediction_Task#14", + "task": "TrajectoryPrediction_Task", + "p_index": 14, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push an object with obstacles', and now you are performing subtask 'push the block to the left, push the block to the left, push the block downward.'\nYour starting position is [0.41, 0.07]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.07], [0.4, 0.13], [0.35, 0.18], [0.54, 0.0], [0.37, 0.0], [0.33, 0.11], [0.39, 0.07], [0.33, 0.11], [0.32, 0.19], [0.42, 0.02]]}" + } + ], + "images": "RH20T_cfg4_task_0089_user_0014_scene_0001_cfg_0004_104122062823#438", + "gt": "[[260, 24], [257, 48], [222, 64], [343, 0], [236, 0], [210, 41], [251, 24], [211, 39], [203, 68], [266, 8]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "58095_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the bottle from the tray and put it on the table', and now you are performing subtask 'remove the bottle from the tray and put it on the table.'\nYou start from [0.61, 0.01]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.61, 0.01], [0.57, 0.1], [0.58, 0.13], [0.57, 0.28], [0.52, 0.34], [0.55, 0.28], [0.61, 0.24], [0.64, 0.27], [0.74, 0.43], [0.76, 0.31]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/58095_exterior_image_1_left#0.jpg", + "gt": "[[195, 2], [184, 18], [187, 24], [182, 51], [167, 61], [177, 50], [196, 44], [205, 48], [237, 77], [244, 55]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0020_user_0016_scene_0002_cfg_0004_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the desk lamp by pressing the button', and now you are performing subtask 'press the button on the desk lamp to turn on the light.'\nYour starting position is [0.38, 0.0]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.3, 0.0], [0.32, 0.0], [0.29, 0.04], [0.3, 0.1], [0.28, 0.03], [0.3, 0.09], [0.3, 0.17], [0.25, 0.24], [0.25, 0.18]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0020_user_0016_scene_0002_cfg_0004_104122062823#0.jpg", + "gt": "[[243, 0], [190, 0], [203, 0], [184, 14], [191, 36], [181, 10], [194, 32], [190, 60], [158, 85], [160, 66]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "49916_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the lemon from the white plate and put it on the nightstand', and now you are performing subtask 'pick up the lemon from the white plate.'\nFrom start point [0.26, 0.26], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.26, 0.26], [0.28, 0.31], [0.28, 0.35], [0.28, 0.37], [0.28, 0.39], [0.34, 0.43], [0.35, 0.47], [0.36, 0.48], [0.35, 0.48], [0.35, 0.51]]}" + } + ], + "images": "49916_exterior_image_2_left#0", + "gt": "[[84, 47], [88, 56], [88, 63], [90, 67], [89, 71], [108, 78], [112, 85], [115, 87], [113, 87], [113, 91]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0009_user_0014_scene_0004_cfg_0001_038522063145#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take out a Hanoi block and throw it aside', and now you are performing subtask 'take out a Hanoi block and throw it aside.'\nThe trajectory starts at [0.69, 0.27]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.27], [0.68, 0.25], [0.69, 0.18], [0.65, 0.3], [0.66, 0.22], [0.68, 0.22], [0.71, 0.26], [0.72, 0.36], [0.71, 0.39], [0.69, 0.22]]}" + } + ], + "images": "RH20T_cfg1_task_0009_user_0014_scene_0004_cfg_0001_038522063145#0", + "gt": "[[441, 97], [433, 89], [442, 66], [418, 109], [423, 80], [436, 80], [455, 93], [462, 129], [452, 141], [444, 81]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "19968_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer on the right, take out the watermelon toy and the green toy and place them on the countertop, close the drawer', and now you are performing subtask 'pull open the right drawer.'\nYou start from [0.39, 0.0]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.0], [0.4, 0.25], [0.56, 0.42], [0.6, 0.44], [0.69, 0.63], [0.68, 0.71], [0.64, 0.7], [0.57, 0.67], [0.56, 0.56], [0.56, 0.56]]}" + } + ], + "images": "19968_exterior_image_2_left#0", + "gt": "[[126, 0], [128, 45], [178, 75], [193, 79], [220, 113], [217, 128], [204, 126], [183, 120], [178, 101], [178, 101]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "59921_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the spoon to stir in the plastic box', and now you are performing subtask 'use the spoon to stir in the plastic box.'\nYou start from [0.82, 0.73]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.82, 0.73], [0.61, 0.79], [0.59, 0.69], [0.54, 0.78], [0.62, 0.73], [0.59, 0.79], [0.54, 0.75], [0.6, 0.86], [0.52, 0.76], [0.57, 0.78]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/59921_exterior_image_1_left#0.jpg", + "gt": "[[262.5, 132.0], [195.5, 142.0], [188.5, 125.0], [172.5, 140.0], [197.5, 131.0], [189.5, 143.0], [173.5, 135.0], [192.5, 154.0], [166.5, 136.0], [182.5, 141.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "4968_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the marker out of the mug and put it on the table', and now you are performing subtask 'take the marker out of the mug and put it on the table.'\nFrom start point [0.16, 0.0], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.16, 0.0], [0.48, 0.24], [0.53, 0.37], [0.51, 0.22], [0.45, 0.07], [0.41, 0.04], [0.4, 0.11], [0.41, 0.51], [0.44, 0.57], [0.44, 0.42]]}" + } + ], + "images": "4968_exterior_image_1_left#0", + "gt": "[[50.5, 0.0], [153.5, 44.0], [169.5, 66.0], [163.5, 40.0], [142.5, 12.0], [130.5, 8.0], [126.5, 20.0], [130.5, 91.0], [141.5, 103.0], [141.5, 75.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0003_user_0010_scene_0001_cfg_0005_105422061350#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the yellow button on the right, press the red button in the middle.'\nStart at [0.43, 0.13]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.13], [0.42, 0.1], [0.44, 0.09], [0.41, 0.07], [0.41, 0.11], [0.42, 0.07], [0.41, 0.07], [0.4, 0.17], [0.39, 0.1], [0.38, 0.24]]}" + } + ], + "images": "RH20T_cfg5_task_0003_user_0010_scene_0001_cfg_0005_105422061350#0", + "gt": "[[276, 47], [266, 35], [280, 31], [264, 26], [264, 38], [267, 24], [260, 26], [257, 60], [252, 35], [240, 86]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_038522063145#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the first button on the desk, press the second button on the desk.'\nYour initial coordinate is [0.85, 0.46]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.85, 0.46], [0.86, 0.45], [0.83, 0.38], [0.76, 0.31], [0.72, 0.28], [0.7, 0.32], [0.68, 0.31], [0.68, 0.33], [0.68, 0.3], [0.68, 0.3]]}" + } + ], + "images": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_038522063145#131", + "gt": "[[547, 167], [550, 162], [532, 138], [486, 113], [460, 101], [450, 116], [437, 112], [432, 118], [436, 107], [436, 107]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0037_user_0007_scene_0004_cfg_0005_036422060215#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the second block from the surface, place the second block on top of the first block, pick up the third block from the surface.'\nWith start location [0.4, 0.33], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.33], [0.4, 0.31], [0.41, 0.31], [0.41, 0.31], [0.41, 0.36], [0.42, 0.28], [0.43, 0.32], [0.43, 0.33], [0.42, 0.36], [0.42, 0.34]]}" + } + ], + "images": "RH20T_cfg5_task_0037_user_0007_scene_0004_cfg_0005_036422060215#437", + "gt": "[[254, 120], [259, 110], [261, 110], [263, 113], [262, 130], [268, 99], [275, 116], [272, 119], [270, 128], [267, 123]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "30133_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the yellow cob of maize into the white bowl', and now you are performing subtask 'move the yellow cob of maize into the white bowl.'\nWith start location [0.59, 0.01], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.01], [0.6, 0.21], [0.59, 0.46], [0.63, 0.53], [0.62, 0.23], [0.55, 0.13], [0.5, 0.14], [0.46, 0.26], [0.45, 0.4], [0.47, 0.18]]}" + } + ], + "images": "30133_exterior_image_1_left#0", + "gt": "[[188, 2], [192, 37], [189, 82], [201, 96], [198, 41], [175, 24], [160, 26], [146, 46], [143, 72], [149, 33]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "74077_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push down the lever on the toaster', and now you are performing subtask 'push down the lever on the toaster.'\nStart at [0.79, 0.26]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.79, 0.26], [0.63, 0.18], [0.52, 0.15], [0.45, 0.22], [0.45, 0.35], [0.46, 0.32], [0.51, 0.29], [0.62, 0.29], [0.64, 0.27], [0.64, 0.27]]}" + } + ], + "images": "74077_exterior_image_2_left#0", + "gt": "[[254, 47], [203, 32], [167, 27], [143, 39], [143, 63], [147, 58], [163, 53], [197, 53], [204, 49], [204, 49]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0015_user_0008_scene_0002_cfg_0001_038522063145#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'play the drum', and now you are performing subtask 'play the drum.'\nYou start from [0.69, 0.27]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.27], [0.63, 0.41], [0.68, 0.29], [0.75, 0.27], [0.74, 0.37], [0.76, 0.16], [0.75, 0.21], [0.69, 0.27], [0.67, 0.36], [0.69, 0.25]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0015_user_0008_scene_0002_cfg_0001_038522063145#0.jpg", + "gt": "[[442, 98], [405, 149], [437, 105], [483, 98], [471, 133], [487, 56], [479, 75], [444, 97], [429, 128], [442, 91]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "7985_exterior_image_1_left#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the towel', and now you are performing subtask 'fold the towel.'\nWith start location [0.52, 0.5], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.5], [0.51, 0.46], [0.53, 0.45], [0.52, 0.5], [0.58, 0.52], [0.63, 0.62], [0.57, 0.5], [0.58, 0.49], [0.56, 0.5], [0.56, 0.5]]}" + } + ], + "images": "7985_exterior_image_1_left#357", + "gt": "[[165, 90], [163, 82], [169, 81], [167, 90], [186, 94], [202, 112], [182, 90], [185, 88], [179, 90], [179, 90]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "48707_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick the lid from the tray and put it on the pot', and now you are performing subtask 'pick the lid from the tray and put it on the pot.'\nStart at [0.48, 0.34]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.34], [0.52, 0.33], [0.54, 0.38], [0.54, 0.48], [0.5, 0.55], [0.51, 0.62], [0.5, 0.69], [0.51, 0.61], [0.55, 0.72], [0.55, 0.72]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/48707_exterior_image_1_left#0.jpg", + "gt": "[[155, 61], [167, 60], [173, 68], [174, 87], [161, 99], [164, 112], [160, 125], [162, 109], [176, 129], [176, 129]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "42894_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull the white bowl to the left, take out the fork and the spoon from the bowl, then place them on the table, move the white cup to the left', and now you are performing subtask 'pick up the cup on the right side of the table.'\nYou begin at [0.41, 0.65]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.65], [0.41, 0.59], [0.37, 0.61], [0.41, 0.72], [0.42, 0.78], [0.41, 0.82], [0.42, 0.81], [0.43, 0.63], [0.45, 0.59], [0.49, 0.57]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/42894_exterior_image_1_left#469.jpg", + "gt": "[[130, 117], [131, 106], [117, 109], [131, 129], [134, 140], [131, 147], [135, 146], [136, 114], [143, 106], [156, 103]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "59693_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the brown cup from the edge of the countertop and put it in the sink', and now you are performing subtask 'pick up the brown cup from the edge of the countertop and put it in the sink.'\nBegin at [0.76, 0.22] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.76, 0.22], [0.71, 0.37], [0.68, 0.41], [0.69, 0.48], [0.67, 0.52], [0.68, 0.57], [0.61, 0.51], [0.58, 0.54], [0.58, 0.61], [0.63, 0.4]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/59693_exterior_image_1_left#0.jpg", + "gt": "[[243, 39], [226, 67], [217, 73], [222, 87], [213, 93], [217, 102], [194, 91], [187, 97], [185, 110], [203, 72]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0003_user_0010_scene_0005_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the green button on the right.'\nStart at [0.5, 0.37]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.53, 0.46], [0.54, 0.58], [0.57, 0.61], [0.56, 0.62], [0.56, 0.58], [0.54, 0.59], [0.53, 0.62], [0.56, 0.64], [0.54, 0.68]]}" + } + ], + "images": "RH20T_cfg5_task_0003_user_0010_scene_0005_cfg_0005_104122063678#0", + "gt": "[[323, 132], [340, 165], [345, 208], [366, 219], [361, 222], [360, 210], [345, 213], [342, 224], [359, 230], [347, 244]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0055_user_0015_scene_0009_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'transfer liquid using a dropper', and now you are performing subtask 'pick up the dropper from the cup, transfer the dropper from the cup to the container with liquid, place the dropper in the container with liquid.'\nYour initial coordinate is [0.63, 0.37]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.6, 0.31], [0.55, 0.3], [0.56, 0.33], [0.6, 0.28], [0.61, 0.18], [0.71, 0.08], [0.73, 0.08], [0.72, 0.14], [0.72, 0.11]]}" + } + ], + "images": "RH20T_cfg1_task_0055_user_0015_scene_0009_cfg_0001_038522062547#0", + "gt": "[[405, 132], [383, 110], [354, 108], [359, 118], [381, 101], [390, 65], [457, 28], [470, 30], [463, 49], [458, 40]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0039_user_0007_scene_0007_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pour the water from a cup into another empty cup', and now you are performing subtask 'pour the water from a cup into another empty cup.'\nFrom start point [0.5, 0.37], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.5, 0.3], [0.52, 0.35], [0.5, 0.41], [0.51, 0.37], [0.54, 0.38], [0.56, 0.53], [0.57, 0.54], [0.52, 0.37], [0.51, 0.46]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0039_user_0007_scene_0007_cfg_0005_104122063678#0.jpg", + "gt": "[[323, 132], [317, 108], [332, 126], [322, 147], [329, 133], [348, 138], [356, 192], [368, 194], [336, 134], [324, 167]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "36386_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the second last switch of the adapter', and now you are performing subtask 'turn on the second last switch of the adapter.'\nStarting at [0.97, 0.0], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.97, 0.0], [0.56, 0.03], [0.52, 0.09], [0.53, 0.12], [0.44, 0.19], [0.4, 0.3], [0.37, 0.32], [0.37, 0.24], [0.39, 0.31], [0.39, 0.31]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/36386_exterior_image_1_left#0.jpg", + "gt": "[[309, 0], [179, 6], [166, 17], [169, 22], [140, 34], [128, 54], [119, 57], [118, 44], [124, 56], [124, 56]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0025_user_0010_scene_0006_cfg_0005_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold a piece of paper that has been folded once (one end is pressed down by a heavy object)', and now you are performing subtask 'unfold a piece of paper that has been folded once (one end is pressed down by a heavy object).'\nYou begin at [0.45, 0.3]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.3], [0.45, 0.32], [0.4, 0.35], [0.42, 0.46], [0.44, 0.42], [0.48, 0.42], [0.52, 0.52], [0.46, 0.36], [0.46, 0.32], [0.47, 0.29]]}" + } + ], + "images": "RH20T_cfg5_task_0025_user_0010_scene_0006_cfg_0005_036422060215#0", + "gt": "[[286, 108], [291, 116], [259, 127], [268, 164], [279, 150], [306, 152], [336, 187], [296, 128], [293, 115], [299, 105]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "70471_exterior_image_2_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the white and black mugs from the drawer, push the drawer closed, close the open bottom cabinet', and now you are performing subtask 'push the drawer closed.'\nYour initial coordinate is [0.49, 0.56]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.56], [0.47, 0.58], [0.45, 0.55], [0.35, 0.69], [0.34, 0.75], [0.36, 0.77], [0.35, 0.79], [0.41, 0.75], [0.46, 0.67], [0.42, 0.71]]}" + } + ], + "images": "70471_exterior_image_2_left#600", + "gt": "[[158, 101], [149, 105], [144, 99], [111, 125], [109, 135], [114, 139], [113, 143], [130, 135], [147, 120], [133, 128]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg3_task_0058_user_0010_scene_0010_cfg_0003_104122063550#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn off the power strips', and now you are performing subtask 'press the button on the power strip above, press the button on the power strip above.'\nYou start from [0.55, 0.37]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.37], [0.57, 0.38], [0.56, 0.38], [0.56, 0.39], [0.56, 0.38], [0.56, 0.38], [0.57, 0.37], [0.57, 0.37], [0.57, 0.38], [0.62, 0.51]]}" + } + ], + "images": "RH20T_cfg3_task_0058_user_0010_scene_0010_cfg_0003_104122063550#75", + "gt": "[[353, 134], [362, 136], [361, 138], [361, 139], [358, 138], [358, 137], [362, 132], [365, 133], [368, 136], [395, 182]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "40415_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the banana from the pot', and now you are performing subtask 'remove the banana from the pot.'\nFrom start point [0.07, 0.01], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.07, 0.01], [0.25, 0.22], [0.33, 0.36], [0.35, 0.39], [0.34, 0.45], [0.38, 0.47], [0.33, 0.44], [0.32, 0.46], [0.33, 0.51], [0.3, 0.46]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/40415_exterior_image_1_left#0.jpg", + "gt": "[[22.5, 1.5], [81.5, 40.5], [106.5, 64.5], [110.5, 69.5], [109.5, 81.5], [120.5, 85.5], [106.5, 79.5], [103.5, 82.5], [106.5, 91.5], [94.5, 82.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0011_user_0016_scene_0006_cfg_0004_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'water the plant', and now you are performing subtask 'water the plant.'\nYour initial coordinate is [0.38, 0.0]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.41, 0.0], [0.4, 0.0], [0.4, 0.02], [0.35, 0.0], [0.46, 0.0], [0.48, 0.03], [0.46, 0.0], [0.51, 0.0], [0.48, 0.0]]}" + } + ], + "images": "RH20T_cfg4_task_0011_user_0016_scene_0006_cfg_0004_104122062823#0", + "gt": "[[243, 0], [264, 0], [256, 0], [257, 7], [222, 0], [297, 0], [308, 10], [296, 0], [325, 0], [308, 0]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0076_user_0010_scene_0001_cfg_0001_750612070853#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stacking blocks', and now you are performing subtask 'place the Lego on the Lego board.'\nWith start location [0.47, 0.71], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.71], [0.48, 0.66], [0.46, 0.7], [0.47, 0.7], [0.46, 0.67], [0.47, 0.61], [0.45, 0.63], [0.43, 0.58], [0.42, 0.52], [0.42, 0.53]]}" + } + ], + "images": "RH20T_cfg1_task_0076_user_0010_scene_0001_cfg_0001_750612070853#273", + "gt": "[[299, 255], [307, 238], [293, 252], [299, 251], [293, 241], [299, 218], [291, 226], [276, 209], [271, 188], [267, 192]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0013_user_0020_scene_0010_cfg_0007_104122061330#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the block on the scale', and now you are performing subtask 'place the big block on the scale, pick up the small block from the table, transfer the small block from the table to the big block, place the small block on the big block.'\nThe trajectory starts at [0.53, 0.49]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.49], [0.5, 0.53], [0.46, 0.63], [0.47, 0.64], [0.48, 0.61], [0.5, 0.54], [0.52, 0.49], [0.51, 0.5], [0.49, 0.53], [0.44, 0.52]]}" + } + ], + "images": "RH20T_cfg7_task_0013_user_0020_scene_0010_cfg_0007_104122061330#252", + "gt": "[[337, 178], [319, 190], [297, 227], [300, 230], [310, 219], [319, 195], [336, 176], [326, 179], [311, 191], [282, 189]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0096_user_0020_scene_0003_cfg_0006_104122060811#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take something out of a drawer', and now you are performing subtask 'pull the handle of the drawer, pick up the object from inside the drawer.'\nYou start from [0.6, 0.35]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.35], [0.57, 0.34], [0.58, 0.31], [0.62, 0.36], [0.65, 0.34], [0.68, 0.37], [0.65, 0.4], [0.68, 0.35], [0.61, 0.38], [0.63, 0.39]]}" + } + ], + "images": "RH20T_cfg6_task_0096_user_0020_scene_0003_cfg_0006_104122060811#0", + "gt": "[[385, 127], [366, 124], [371, 113], [395, 129], [414, 121], [434, 134], [413, 143], [434, 126], [392, 136], [406, 141]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "51525_exterior_image_2_left#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the microwave door, move the pink jar on the counter', and now you are performing subtask 'transfer the pink jar from the microwave to the countertop.'\nWith start location [0.42, 0.34], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.34], [0.43, 0.36], [0.39, 0.4], [0.37, 0.43], [0.38, 0.46], [0.41, 0.51], [0.41, 0.53], [0.41, 0.52], [0.4, 0.46], [0.36, 0.4]]}" + } + ], + "images": "51525_exterior_image_2_left#205", + "gt": "[[134, 62], [139, 64], [125, 72], [119, 78], [120, 82], [131, 92], [132, 95], [130, 94], [128, 83], [116, 72]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "35107_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the towel, then lay it out on the countertop', and now you are performing subtask 'place the towel flat on the countertop.'\nYour initial coordinate is [0.47, 0.71]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.71], [0.49, 0.61], [0.53, 0.57], [0.52, 0.57], [0.53, 0.53], [0.53, 0.6], [0.54, 0.59], [0.58, 0.67], [0.6, 0.52], [0.57, 0.54]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/35107_exterior_image_2_left#0.jpg", + "gt": "[[150, 128], [158, 110], [170, 102], [167, 103], [170, 95], [170, 108], [172, 107], [185, 120], [192, 93], [184, 98]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0017_user_0014_scene_0002_cfg_0007_104122061330#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the pen into the pen holder', and now you are performing subtask 'put the pen into the pen holder.'\nFrom start point [0.42, 0.29], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.29], [0.49, 0.29], [0.53, 0.36], [0.53, 0.4], [0.52, 0.38], [0.43, 0.46], [0.38, 0.63], [0.41, 0.66], [0.43, 0.47], [0.43, 0.37]]}" + } + ], + "images": "RH20T_cfg7_task_0017_user_0014_scene_0002_cfg_0007_104122061330#0", + "gt": "[[266, 103], [315, 105], [339, 128], [337, 144], [333, 138], [278, 165], [245, 226], [261, 236], [275, 169], [274, 132]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0096_user_0007_scene_0003_cfg_0005_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer and place the cube on the table', and now you are performing subtask 'pull the handle of the drawer to pull open the drawer.'\nYour starting position is [0.45, 0.3]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.3], [0.45, 0.29], [0.45, 0.3], [0.46, 0.3], [0.45, 0.32], [0.45, 0.38], [0.45, 0.39], [0.44, 0.37], [0.41, 0.32], [0.4, 0.28]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0096_user_0007_scene_0003_cfg_0005_036422060215#0.jpg", + "gt": "[[286, 108], [287, 105], [290, 108], [292, 107], [285, 114], [287, 136], [285, 140], [281, 133], [260, 115], [259, 101]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0003_user_0006_scene_0007_cfg_0002_105422061350#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the three buttons in sequence', and now you are performing subtask 'press the gray base green button, press the gray base red button.'\nThe trajectory starts at [0.58, 0.42]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.43], [0.6, 0.37], [0.61, 0.19], [0.6, 0.22], [0.62, 0.2], [0.63, 0.24], [0.59, 0.32], [0.62, 0.28], [0.64, 0.29], [0.65, 0.36]]}" + } + ], + "images": "RH20T_cfg2_task_0003_user_0006_scene_0007_cfg_0002_105422061350#0", + "gt": "[[369, 153], [381, 132], [389, 70], [386, 78], [398, 71], [405, 85], [380, 116], [396, 101], [411, 103], [417, 131]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0201_user_0015_scene_0007_cfg_0001_819112071037#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'the medium pipette tip is inserted into the holder for medium pipette tips', and now you are performing subtask 'place the tip of the medium pipette into the holder for medium pipette tips, push the tip of the medium pipette.'\nYour starting position is [0.55, 0.23]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.23], [0.62, 0.19], [0.63, 0.11], [0.61, 0.17], [0.64, 0.24], [0.58, 0.24], [0.62, 0.27], [0.63, 0.26], [0.56, 0.22], [0.43, 0.19]]}" + } + ], + "images": "RH20T_cfg1_task_0201_user_0015_scene_0007_cfg_0001_819112071037#268", + "gt": "[[351, 84], [398, 69], [401, 40], [393, 62], [407, 87], [373, 86], [395, 97], [404, 93], [360, 81], [274, 70]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0002_user_0010_scene_0001_cfg_0005_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull out a napkin', and now you are performing subtask 'pull out a napkin.'\nYour initial coordinate is [0.4, 0.35]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.35], [0.4, 0.36], [0.4, 0.43], [0.39, 0.49], [0.39, 0.45], [0.46, 0.24], [0.49, 0.22], [0.5, 0.25], [0.52, 0.26], [0.52, 0.27]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0002_user_0010_scene_0001_cfg_0005_036422060215#0.jpg", + "gt": "[[255, 125], [253, 129], [256, 156], [247, 176], [248, 161], [292, 86], [314, 79], [322, 90], [332, 92], [331, 96]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0023_user_0010_scene_0006_cfg_0003_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the power strip by pressing the button', and now you are performing subtask 'turn on the power strip by pressing the button.'\nStarting at [0.38, 0.0], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.52, 0.12], [0.5, 0.12], [0.5, 0.24], [0.48, 0.28], [0.5, 0.28], [0.46, 0.38], [0.46, 0.32], [0.44, 0.2], [0.45, 0.09]]}" + } + ], + "images": "RH20T_cfg3_task_0023_user_0010_scene_0006_cfg_0003_104122062823#0", + "gt": "[[246, 0], [336, 43], [321, 45], [318, 87], [308, 100], [317, 99], [295, 136], [294, 115], [279, 73], [289, 34]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "79049_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the spoon from the table and use it to mix the objects in the bowl', and now you are performing subtask 'pick up the spoon from the table and use it to mix the objects in the bowl.'\nWith start location [0.54, 0.43], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.43], [0.61, 0.51], [0.6, 0.64], [0.53, 0.45], [0.5, 0.61], [0.47, 0.59], [0.47, 0.63], [0.51, 0.54], [0.58, 0.58], [0.58, 0.58]]}" + } + ], + "images": "79049_exterior_image_1_left#0", + "gt": "[[171.5, 77.5], [195.5, 91.5], [193.5, 115.5], [169.5, 80.5], [158.5, 110.5], [149.5, 105.5], [150.5, 113.5], [164.5, 96.5], [186.5, 104.5], [186.5, 104.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "54874_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker in the cup', and now you are performing subtask 'put the marker in the cup.'\nYou start from [0.67, 0.28]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.67, 0.28], [0.63, 0.13], [0.52, 0.18], [0.52, 0.36], [0.56, 0.45], [0.55, 0.51], [0.57, 0.29], [0.51, 0.23], [0.45, 0.26], [0.45, 0.32]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/54874_exterior_image_1_left#0.jpg", + "gt": "[[214.0, 50.0], [202.0, 24.0], [167.0, 32.0], [166.0, 64.0], [179.0, 81.0], [175.0, 92.0], [182.0, 53.0], [162.0, 42.0], [145.0, 46.0], [143.0, 58.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0038_user_0014_scene_0010_cfg_0006_104122061602#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the cup', and now you are performing subtask 'pick up the cup.'\nYou begin at [0.7, 0.29]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.29], [0.68, 0.27], [0.65, 0.27], [0.65, 0.28], [0.65, 0.29], [0.64, 0.29], [0.63, 0.32], [0.62, 0.35], [0.63, 0.3], [0.68, 0.26]]}" + } + ], + "images": "RH20T_cfg6_task_0038_user_0014_scene_0010_cfg_0006_104122061602#0", + "gt": "[[446, 106], [433, 97], [419, 97], [415, 101], [418, 104], [408, 106], [405, 117], [395, 127], [405, 107], [433, 95]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "51039_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the orange cup from the table, pour its contents into the black pot, and put the cup back on the table', and now you are performing subtask 'pick up the orange cup from the table, pour its contents into the black pot, and put the cup back on the table.'\nStart at [0.82, 0.13]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.82, 0.13], [0.59, 0.27], [0.5, 0.54], [0.52, 0.36], [0.43, 0.19], [0.49, 0.26], [0.5, 0.41], [0.41, 0.51], [0.44, 0.56], [0.48, 0.37]]}" + } + ], + "images": "51039_exterior_image_1_left#0", + "gt": "[[261, 23], [189, 49], [159, 98], [165, 65], [138, 35], [157, 47], [160, 73], [132, 92], [141, 101], [153, 67]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "25326_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the lid on the clear container on the left then turn on the extension cable', and now you are performing subtask 'put the lid on the clear container.'\nYou begin at [0.73, 0.0]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.73, 0.0], [0.69, 0.03], [0.55, 0.28], [0.52, 0.41], [0.48, 0.48], [0.45, 0.46], [0.41, 0.35], [0.38, 0.36], [0.38, 0.41], [0.36, 0.37]]}" + } + ], + "images": "25326_exterior_image_1_left#0", + "gt": "[[235, 0], [222, 6], [176, 50], [166, 73], [153, 87], [143, 82], [130, 63], [123, 64], [121, 74], [114, 66]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "79387_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the top drawer of the white file cabinet', and now you are performing subtask 'close the top drawer of the white file cabinet.'\nStarting at [0.76, 0.99], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.76, 0.99], [0.75, 0.99], [0.7, 0.87], [0.65, 0.69], [0.62, 0.63], [0.59, 0.6], [0.54, 0.59], [0.47, 0.56], [0.42, 0.58], [0.42, 0.58]]}" + } + ], + "images": "79387_exterior_image_1_left#0", + "gt": "[[244, 179], [241, 178], [223, 156], [208, 125], [199, 113], [190, 108], [174, 107], [150, 100], [134, 104], [134, 104]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0004_user_0005_scene_0003_cfg_0001_819112070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the small block on the left and move it to the right', and now you are performing subtask 'pick up the small block on the left and move it to the right.'\nYour starting position is [0.62, 0.47]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.47], [0.66, 0.43], [0.69, 0.44], [0.71, 0.48], [0.69, 0.48], [0.7, 0.53], [0.69, 0.53], [0.69, 0.57], [0.64, 0.46], [0.62, 0.46]]}" + } + ], + "images": "RH20T_cfg1_task_0004_user_0005_scene_0003_cfg_0001_819112070044#0", + "gt": "[[400, 169], [425, 154], [440, 158], [454, 172], [439, 172], [450, 190], [441, 190], [443, 204], [407, 165], [394, 166]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "52523_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the cloth on the table', and now you are performing subtask 'fold the cloth on the table.'\nStarting at [0.73, 0.64], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.73, 0.64], [0.78, 0.67], [0.75, 0.71], [0.78, 0.62], [0.75, 0.42], [0.7, 0.31], [0.64, 0.36], [0.64, 0.43], [0.57, 0.22], [0.57, 0.22]]}" + } + ], + "images": "52523_exterior_image_2_left#145", + "gt": "[[234, 115], [250, 120], [239, 128], [251, 111], [239, 75], [225, 56], [204, 65], [206, 77], [181, 40], [181, 40]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "45921_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the white book into the gap between the books on the left side of the bookshelf', and now you are performing subtask 'move the white book into the gap between the books on the left side of the bookshelf.'\nBegin at [0.66, 0.52] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.66, 0.52], [0.68, 0.39], [0.57, 0.09], [0.47, 0.26], [0.56, 0.25], [0.41, 0.17], [0.34, 0.33], [0.41, 0.23], [0.6, 0.41], [0.6, 0.41]]}" + } + ], + "images": "45921_exterior_image_1_left#0", + "gt": "[[211.0, 93.0], [219.0, 70.0], [184.0, 17.0], [149.0, 47.0], [180.0, 45.0], [130.0, 30.0], [110.0, 59.0], [132.0, 41.0], [191.0, 74.0], [191.0, 74.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0065_user_0014_scene_0002_cfg_0001_750612070853#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with small Lego blocks for the nth time with n blocks', and now you are performing subtask 'pick up the pink block on the desktop, transfer the pink block from the desktop to the building area, place the pink block in the building area.'\nStart at [0.56, 0.23]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.56, 0.22], [0.58, 0.24], [0.45, 0.36], [0.44, 0.35], [0.41, 0.43], [0.41, 0.48], [0.43, 0.4], [0.47, 0.38], [0.52, 0.43], [0.52, 0.46]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0065_user_0014_scene_0002_cfg_0001_750612070853#0.jpg", + "gt": "[[356, 81], [371, 85], [288, 131], [281, 126], [261, 155], [262, 172], [273, 144], [298, 137], [334, 154], [334, 166]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "83829_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the green block from the blue table and place it on the green block', and now you are performing subtask 'move the green block from the blue table and place it on the green block.'\nWith start location [0.21, 0.12], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.21, 0.12], [0.32, 0.58], [0.35, 0.42], [0.4, 0.44], [0.43, 0.38], [0.55, 0.42], [0.51, 0.29], [0.48, 0.28], [0.45, 0.32], [0.45, 0.32]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/83829_exterior_image_1_left#0.jpg", + "gt": "[[67, 22], [101, 105], [112, 75], [128, 80], [138, 68], [176, 75], [163, 53], [154, 51], [145, 58], [145, 58]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "6142_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the lid of the transparent box, then put the transparent box inside the container', and now you are performing subtask 'place the bottle in the transparent box.'\nYour initial coordinate is [0.47, 0.51]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.51], [0.48, 0.56], [0.49, 0.57], [0.5, 0.44], [0.49, 0.39], [0.47, 0.39], [0.46, 0.43], [0.46, 0.42], [0.46, 0.36], [0.47, 0.32]]}" + } + ], + "images": "6142_exterior_image_2_left#150", + "gt": "[[151, 91], [154, 101], [158, 102], [161, 80], [156, 70], [150, 71], [148, 77], [147, 75], [146, 64], [151, 57]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "54536_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the green towel', and now you are performing subtask 'fold the green towel.'\nBegin at [0.33, 0.02] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.02], [0.43, 0.35], [0.47, 0.55], [0.48, 0.54], [0.5, 0.4], [0.48, 0.43], [0.49, 0.37], [0.47, 0.33], [0.49, 0.34], [0.47, 0.33]]}" + } + ], + "images": "54536_exterior_image_1_left#0", + "gt": "[[104, 4], [136, 63], [150, 99], [155, 98], [159, 72], [155, 78], [157, 67], [151, 59], [157, 61], [152, 60]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0105_user_0020_scene_0007_cfg_0006_104122061330#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the object on the shelf', and now you are performing subtask 'pick up the object from the table, transfer the object from the table to the shelf, place the object on the shelf.'\nStarting at [0.42, 0.27], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.27], [0.46, 0.33], [0.48, 0.33], [0.53, 0.4], [0.54, 0.4], [0.58, 0.51], [0.58, 0.37], [0.56, 0.23], [0.57, 0.16], [0.55, 0.07]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0105_user_0020_scene_0007_cfg_0006_104122061330#0.jpg", + "gt": "[[266, 97], [294, 118], [305, 119], [340, 143], [347, 143], [372, 184], [370, 134], [359, 83], [368, 57], [351, 24]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "47071_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the wooden spatula and mix the contents in the silver bowl', and now you are performing subtask 'pick up the spoon from the pot.'\nThe trajectory starts at [0.62, 0.33]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.33], [0.57, 0.31], [0.5, 0.32], [0.45, 0.33], [0.39, 0.37], [0.38, 0.38], [0.38, 0.39], [0.38, 0.39], [0.37, 0.42], [0.33, 0.42]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/47071_exterior_image_2_left#0.jpg", + "gt": "[[197, 60], [182, 56], [160, 57], [144, 60], [125, 66], [121, 69], [123, 71], [121, 70], [118, 75], [107, 76]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "52870_exterior_image_1_left#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the lemons and an apple out of the glass bowl and place them on the countertop', and now you are performing subtask 'place the fourth lemon on the tabletop.'\nFrom start point [0.56, 0.66], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.56, 0.66], [0.6, 0.56], [0.7, 0.51], [0.73, 0.54], [0.71, 0.62], [0.74, 0.67], [0.69, 0.49], [0.62, 0.57], [0.58, 0.71], [0.59, 0.63]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/52870_exterior_image_1_left#330.jpg", + "gt": "[[180, 119], [191, 100], [223, 92], [232, 97], [226, 112], [238, 120], [220, 88], [199, 102], [187, 127], [189, 114]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "11777_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the marker from the cup and put it on the table', and now you are performing subtask 'take the marker from the cup and put it on the table.'\nBegin at [0.76, 0.07] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.76, 0.07], [0.77, 0.13], [0.8, 0.1], [0.8, 0.13], [0.73, 0.11], [0.68, 0.23], [0.71, 0.13], [0.7, 0.09], [0.68, 0.09], [0.6, 0.21]]}" + } + ], + "images": "11777_exterior_image_2_left#0", + "gt": "[[244, 13], [248, 23], [257, 18], [256, 23], [234, 20], [218, 41], [228, 24], [225, 16], [218, 16], [193, 38]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "70472_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the top right cabinet door', and now you are performing subtask 'close the top right cabinet door.'\nBegin at [0.99, 0.36] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.99, 0.36], [0.92, 0.24], [0.96, 0.05], [0.88, 0.11], [0.82, 0.0], [0.8, 0.09], [0.74, 0.03], [0.64, 0.01], [0.76, 0.24], [0.82, 0.24]]}" + } + ], + "images": "70472_exterior_image_1_left#0", + "gt": "[[318, 65], [293, 43], [307, 9], [281, 20], [264, 0], [256, 17], [237, 5], [206, 1], [244, 44], [262, 44]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "11997_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the pen out of the mug and put it on the counter', and now you are performing subtask 'take the pen out of the mug and put it on the counter.'\nYou start from [0.87, 0.0]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.87, 0.0], [0.86, 0.1], [0.88, 0.16], [0.86, 0.32], [0.78, 0.41], [0.84, 0.27], [0.83, 0.23], [0.79, 0.25], [0.72, 0.45], [0.72, 0.45]]}" + } + ], + "images": "11997_exterior_image_2_left#0", + "gt": "[[277, 0], [275, 18], [283, 28], [276, 57], [251, 73], [269, 48], [265, 41], [252, 45], [231, 81], [231, 81]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "24096_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the glass lid on the black pot', and now you are performing subtask 'put the glass lid on the black pot.'\nYou start from [0.25, 0.14]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.25, 0.14], [0.33, 0.46], [0.3, 0.33], [0.35, 0.32], [0.38, 0.4], [0.37, 0.47], [0.28, 0.46], [0.32, 0.51], [0.32, 0.46], [0.32, 0.46]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/24096_exterior_image_1_left#0.jpg", + "gt": "[[79, 26], [106, 83], [97, 60], [111, 57], [123, 72], [118, 85], [88, 83], [103, 92], [101, 82], [101, 82]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0051_user_0014_scene_0002_cfg_0006_037522061512#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the dish off the dish rack', and now you are performing subtask 'take the dish off the dish rack.'\nYou begin at [0.4, 0.23]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.23], [0.39, 0.17], [0.42, 0.11], [0.38, 0.01], [0.31, 0.18], [0.31, 0.29], [0.28, 0.34], [0.28, 0.32], [0.32, 0.23], [0.3, 0.27]]}" + } + ], + "images": "RH20T_cfg6_task_0051_user_0014_scene_0002_cfg_0006_037522061512#0", + "gt": "[[257, 83], [248, 62], [267, 41], [244, 2], [198, 66], [198, 106], [182, 124], [176, 115], [207, 83], [190, 96]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0057_user_0014_scene_0009_cfg_0004_038522062288#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn on the power strips', and now you are performing subtask 'press the first button from the left on the power socket.'\nWith start location [0.3, 0.55], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.3, 0.55], [0.32, 0.52], [0.32, 0.52], [0.32, 0.49], [0.31, 0.51], [0.3, 0.52], [0.3, 0.52], [0.3, 0.45], [0.31, 0.43], [0.31, 0.46]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0057_user_0014_scene_0009_cfg_0004_038522062288#267.jpg", + "gt": "[[190, 199], [203, 187], [208, 189], [202, 176], [199, 182], [192, 188], [190, 187], [195, 161], [199, 156], [201, 166]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "84608_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the lid on the pot', and now you are performing subtask 'put the lid on the pot.'\nStart at [0.09, 0.24]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.09, 0.24], [0.19, 0.22], [0.27, 0.42], [0.33, 0.42], [0.33, 0.57], [0.31, 0.38], [0.35, 0.41], [0.36, 0.48], [0.4, 0.48], [0.39, 0.56]]}" + } + ], + "images": "84608_exterior_image_1_left#0", + "gt": "[[30, 43], [61, 39], [87, 75], [105, 75], [105, 102], [99, 68], [112, 73], [115, 86], [129, 86], [126, 100]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "18559_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the lid on the table', and now you are performing subtask 'put the lid on the table.'\nYou start from [0.96, 0.14]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.96, 0.14], [0.65, 0.43], [0.57, 0.57], [0.55, 0.68], [0.52, 0.68], [0.55, 0.65], [0.55, 0.72], [0.53, 0.64], [0.52, 0.7], [0.55, 0.6]]}" + } + ], + "images": "18559_exterior_image_2_left#0", + "gt": "[[307, 25], [209, 78], [183, 102], [177, 122], [165, 123], [177, 117], [175, 129], [171, 115], [165, 126], [175, 108]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0037_user_0015_scene_0004_cfg_0001_035622060973#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the third block on top of the first block, pick up the fourth block on the table, transfer the fourth block from the table to above the third block, place the fourth block on top of the third block, pick up the fifth block on the table.'\nWith start location [0.38, 0.36], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.36], [0.35, 0.42], [0.32, 0.56], [0.35, 0.58], [0.36, 0.66], [0.36, 0.52], [0.42, 0.31], [0.38, 0.4], [0.37, 0.63], [0.37, 0.63]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0037_user_0015_scene_0004_cfg_0001_035622060973#520.jpg", + "gt": "[[246, 131], [224, 151], [208, 203], [225, 209], [229, 236], [228, 187], [271, 110], [242, 144], [239, 226], [239, 226]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "70472_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the top right cabinet door', and now you are performing subtask 'push the doorknob to the position of closing the top right cabinet door.'\nWith start location [0.48, 0.02], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.02], [0.49, 0.04], [0.47, 0.08], [0.48, 0.16], [0.46, 0.21], [0.43, 0.44], [0.38, 0.52], [0.36, 0.53], [0.19, 0.49], [0.22, 0.51]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/70472_exterior_image_2_left#118.jpg", + "gt": "[[154, 4], [156, 7], [151, 14], [154, 28], [147, 38], [136, 80], [122, 93], [114, 96], [62, 89], [72, 92]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "57664_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the spoon out of the cup, place it on the table, and pour the contents of the cup into another cup', and now you are performing subtask 'place the cup on the table.'\nYour starting position is [0.55, 0.67]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.67], [0.52, 0.59], [0.5, 0.76], [0.52, 0.64], [0.59, 0.66], [0.59, 0.6], [0.59, 0.66], [0.55, 0.66], [0.57, 0.72], [0.54, 0.66]]}" + } + ], + "images": "57664_exterior_image_1_left#185", + "gt": "[[176.5, 120.0], [167.5, 106.0], [158.5, 136.0], [165.5, 116.0], [189.5, 119.0], [188.5, 108.0], [187.5, 118.0], [175.5, 119.0], [183.5, 129.0], [171.5, 118.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "35524_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the switch for the fourth slot from the right', and now you are performing subtask 'press the switch for the fourth slot from the right.'\nYour starting position is [0.69, 0.0]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.0], [0.68, 0.05], [0.66, 0.29], [0.65, 0.3], [0.65, 0.27], [0.65, 0.29], [0.65, 0.28], [0.62, 0.33], [0.58, 0.36], [0.58, 0.44]]}" + } + ], + "images": "35524_exterior_image_1_left#0", + "gt": "[[222, 0], [218, 9], [210, 52], [208, 54], [208, 49], [209, 52], [208, 50], [197, 60], [185, 64], [186, 79]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0096_user_0020_scene_0002_cfg_0006_104122064161#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take something out of a drawer', and now you are performing subtask 'pick up the item from the drawer, transfer the item from the drawer to the surface, place the item on the surface.'\nThe trajectory starts at [0.55, 0.22]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.22], [0.53, 0.29], [0.53, 0.32], [0.55, 0.21], [0.58, 0.41], [0.58, 0.54], [0.57, 0.59], [0.6, 0.5], [0.59, 0.31], [0.6, 0.32]]}" + } + ], + "images": "RH20T_cfg6_task_0096_user_0020_scene_0002_cfg_0006_104122064161#289", + "gt": "[[351, 79], [340, 104], [342, 115], [350, 74], [371, 149], [370, 196], [362, 213], [382, 181], [377, 110], [387, 114]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "14849_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push the tap to the right, remove the toy from the orange plate and place it on the right side of the orange plate', and now you are performing subtask 'push the tap to the right.'\nFrom start point [0.51, 0.0], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.0], [0.55, 0.03], [0.59, 0.14], [0.6, 0.24], [0.59, 0.27], [0.57, 0.28], [0.71, 0.29], [0.64, 0.28], [0.55, 0.21], [0.51, 0.22]]}" + } + ], + "images": "14849_exterior_image_1_left#0", + "gt": "[[164, 0], [175, 6], [190, 25], [191, 44], [189, 49], [184, 51], [226, 52], [206, 50], [177, 38], [164, 40]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "32511_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the towel', and now you are performing subtask 'pick up the towel.'\nStart at [0.8, 0.01]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.8, 0.01], [0.73, 0.17], [0.67, 0.47], [0.63, 0.53], [0.63, 0.25], [0.61, 0.14], [0.64, 0.26], [0.62, 0.5], [0.62, 0.33], [0.61, 0.36]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/32511_exterior_image_2_left#0.jpg", + "gt": "[[257, 2], [235, 30], [214, 84], [201, 96], [201, 45], [195, 25], [205, 46], [197, 90], [200, 60], [195, 65]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "91638_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the shirt', and now you are performing subtask 'fold the shirt's other sleeve.'\nBegin at [0.63, 0.09] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.09], [0.79, 0.13], [0.85, 0.28], [0.84, 0.35], [0.8, 0.41], [0.77, 0.28], [0.61, 0.23], [0.59, 0.24], [0.65, 0.28], [0.65, 0.28]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/91638_exterior_image_1_left#0.jpg", + "gt": "[[203, 16], [252, 23], [271, 50], [269, 63], [255, 73], [248, 51], [194, 41], [189, 43], [209, 51], [209, 51]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0106_user_0020_scene_0004_cfg_0006_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the object down from the shelf', and now you are performing subtask 'pick up the object from the top shelf, transfer the object from the top shelf to the table, place the object on the table.'\nYour initial coordinate is [0.53, 0.31]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.31], [0.53, 0.29], [0.56, 0.28], [0.57, 0.49], [0.59, 0.38], [0.58, 0.33], [0.56, 0.3], [0.55, 0.44], [0.57, 0.41], [0.57, 0.41]]}" + } + ], + "images": "RH20T_cfg6_task_0106_user_0020_scene_0004_cfg_0006_104122064161#0", + "gt": "[[338, 111], [342, 103], [361, 100], [368, 176], [376, 137], [371, 118], [359, 108], [355, 159], [362, 149], [362, 149]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0081_user_0007_scene_0003_cfg_0005_104122061850#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the leg of the glasses (with one hand)', and now you are performing subtask 'fold the legs of the glasses to unfold them.'\nYour initial coordinate is [0.6, 0.25]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.25], [0.6, 0.24], [0.61, 0.21], [0.62, 0.22], [0.62, 0.24], [0.61, 0.29], [0.61, 0.31], [0.6, 0.34], [0.59, 0.34], [0.59, 0.34]]}" + } + ], + "images": "RH20T_cfg5_task_0081_user_0007_scene_0003_cfg_0005_104122061850#275", + "gt": "[[381, 89], [386, 85], [391, 74], [396, 80], [399, 88], [393, 106], [389, 113], [381, 122], [379, 122], [379, 122]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "3391_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the left bottom cabinet door', and now you are performing subtask 'open the left bottom cabinet door.'\nYour initial coordinate is [0.98, 0.0]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.98, 0.0], [0.78, 0.04], [0.7, 0.14], [0.57, 0.38], [0.52, 0.57], [0.49, 0.53], [0.52, 0.54], [0.52, 0.63], [0.58, 0.62], [0.57, 0.65]]}" + } + ], + "images": "3391_exterior_image_2_left#0", + "gt": "[[315, 0], [251, 8], [224, 26], [183, 68], [165, 102], [157, 96], [166, 97], [165, 114], [187, 111], [183, 117]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "65660_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the object on the upper half of the box', and now you are performing subtask 'put the object on the upper half of the box.'\nYou begin at [0.63, 0.46]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.46], [0.56, 0.69], [0.57, 0.68], [0.58, 0.75], [0.56, 0.84], [0.57, 0.9], [0.56, 0.81], [0.58, 0.82], [0.57, 0.83], [0.55, 0.73]]}" + } + ], + "images": "65660_exterior_image_1_left#0", + "gt": "[[203.0, 82.0], [178.0, 124.0], [182.0, 123.0], [185.0, 135.0], [179.0, 152.0], [184.0, 162.0], [179.0, 145.0], [186.0, 147.0], [182.0, 149.0], [175.0, 131.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "71329_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker in the mug and then remove it and put it back on the table', and now you are performing subtask 'pick up the marker on the table.'\nStarting at [0.36, 0.02], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.02], [0.28, 0.19], [0.27, 0.22], [0.29, 0.21], [0.22, 0.4], [0.22, 0.56], [0.18, 0.72], [0.2, 0.71], [0.24, 0.74], [0.24, 0.68]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/71329_exterior_image_2_left#0.jpg", + "gt": "[[116, 3], [89, 35], [87, 39], [92, 38], [72, 72], [70, 101], [58, 129], [65, 127], [76, 133], [78, 122]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0008_user_0010_scene_0010_cfg_0005_105422061350#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grab the block and place it at the black dot', and now you are performing subtask 'grab the block and place it at the black dot.'\nWith start location [0.43, 0.13], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.13], [0.45, 0.17], [0.5, 0.12], [0.49, 0.0], [0.44, 0.0], [0.45, 0.13], [0.38, 0.34], [0.39, 0.35], [0.4, 0.18], [0.44, 0.07]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0008_user_0010_scene_0010_cfg_0005_105422061350#0.jpg", + "gt": "[[276, 47], [287, 62], [319, 45], [316, 0], [281, 0], [286, 46], [243, 123], [249, 126], [253, 66], [284, 24]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0043_user_0014_scene_0008_cfg_0006_104122064161#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unscrew the lid from the jar', and now you are performing subtask 'twist the lid of the jar counterclockwise.'\nStart at [0.52, 0.48]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.48], [0.52, 0.49], [0.52, 0.49], [0.51, 0.51], [0.51, 0.5], [0.51, 0.5], [0.51, 0.5], [0.51, 0.5], [0.51, 0.49], [0.51, 0.49]]}" + } + ], + "images": "RH20T_cfg6_task_0043_user_0014_scene_0008_cfg_0006_104122064161#204", + "gt": "[[333, 174], [335, 175], [332, 176], [329, 182], [327, 180], [325, 181], [326, 180], [325, 181], [325, 176], [324, 177]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0105_user_0020_scene_0007_cfg_0006_104122061330#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the object on the shelf', and now you are performing subtask 'place the object on the shelf, pick up the object from the tabletop, transfer the object from the tabletop to the specified location, place the object at the specified location.'\nYou begin at [0.55, 0.07]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.07], [0.56, 0.04], [0.56, 0.06], [0.47, 0.4], [0.47, 0.54], [0.5, 0.63], [0.51, 0.37], [0.5, 0.35], [0.52, 0.23], [0.51, 0.12]]}" + } + ], + "images": "RH20T_cfg6_task_0105_user_0020_scene_0007_cfg_0006_104122061330#380", + "gt": "[[351, 24], [356, 15], [360, 20], [303, 145], [304, 195], [320, 227], [325, 132], [318, 127], [334, 82], [327, 43]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "5462_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the brown plush toy to the left', and now you are performing subtask 'move the brown plush toy to the left.'\nStarting at [0.37, 0.49], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.49], [0.44, 0.35], [0.56, 0.32], [0.65, 0.65], [0.62, 0.7], [0.64, 0.62], [0.65, 0.65], [0.55, 0.64], [0.52, 0.6], [0.52, 0.64]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/5462_exterior_image_1_left#0.jpg", + "gt": "[[118.5, 87.5], [141.5, 62.5], [178.5, 58.5], [209.5, 116.5], [199.5, 125.5], [204.5, 111.5], [206.5, 117.5], [176.5, 114.5], [165.5, 107.5], [166.5, 114.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "10020_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the black lid from the top of the can and put it on the table', and now you are performing subtask 'take the black lid from the top of the can and put it on the table.'\nStart at [0.57, 0.0]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.0], [0.56, 0.15], [0.54, 0.17], [0.53, 0.09], [0.52, 0.09], [0.48, 0.15], [0.44, 0.32], [0.45, 0.32], [0.46, 0.19], [0.46, 0.19]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/10020_exterior_image_2_left#0.jpg", + "gt": "[[182, 0], [179, 27], [173, 30], [170, 17], [165, 16], [153, 27], [140, 57], [143, 57], [147, 35], [147, 35]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "41068_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer and then put the brown object from the bowl on the green cloth', and now you are performing subtask 'place the brown object on the green towel.'\nBegin at [0.61, 0.41] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.61, 0.41], [0.6, 0.52], [0.62, 0.6], [0.58, 0.62], [0.57, 0.59], [0.56, 0.64], [0.53, 0.56], [0.53, 0.52], [0.53, 0.58], [0.5, 0.48]]}" + } + ], + "images": "41068_exterior_image_2_left#123", + "gt": "[[194, 73], [192, 93], [198, 108], [187, 111], [181, 106], [179, 115], [170, 100], [171, 94], [171, 105], [161, 86]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0034_user_0003_scene_0006_cfg_0002_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'pick up the square from the left side of the stack, transfer the square from the left side to above the center of the stack, place the square on top of the existing square in the center.'\nBegin at [0.53, 0.29] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.29], [0.55, 0.33], [0.55, 0.39], [0.57, 0.46], [0.55, 0.51], [0.54, 0.48], [0.53, 0.48], [0.52, 0.54], [0.51, 0.66], [0.51, 0.66]]}" + } + ], + "images": "RH20T_cfg2_task_0034_user_0003_scene_0006_cfg_0002_104122061850#0", + "gt": "[[340, 105], [350, 119], [350, 139], [363, 164], [351, 184], [348, 173], [342, 173], [330, 195], [328, 236], [328, 236]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "59697_exterior_image_2_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the objects from the orange bowl to the black pot, then cover the black pot with the glass lid', and now you are performing subtask 'place the glass lid on the black pot.'\nThe trajectory starts at [0.83, 0.49]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.83, 0.49], [0.9, 0.32], [0.9, 0.19], [0.76, 0.07], [0.69, 0.11], [0.67, 0.43], [0.63, 0.39], [0.63, 0.46], [0.68, 0.37], [0.71, 0.17]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/59697_exterior_image_2_left#398.jpg", + "gt": "[[266, 88], [288, 57], [287, 35], [244, 12], [220, 20], [215, 78], [201, 71], [202, 83], [218, 66], [227, 31]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "76067_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'tip the white mug over', and now you are performing subtask 'tip the white mug over.'\nBegin at [0.27, 0.15] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.27, 0.15], [0.32, 0.17], [0.3, 0.18], [0.43, 0.3], [0.44, 0.37], [0.5, 0.48], [0.51, 0.57], [0.47, 0.51], [0.48, 0.6], [0.52, 0.51]]}" + } + ], + "images": "76067_exterior_image_1_left#0", + "gt": "[[86, 27], [103, 30], [97, 32], [137, 54], [140, 67], [161, 87], [162, 103], [150, 92], [154, 108], [166, 91]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "12658_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'rotate the tap nozzle clockwise', and now you are performing subtask 'rotate the tap nozzle clockwise.'\nWith start location [0.45, 0.0], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.0], [0.54, 0.13], [0.59, 0.37], [0.55, 0.37], [0.52, 0.14], [0.53, 0.08], [0.46, 0.19], [0.38, 0.22], [0.42, 0.24], [0.39, 0.23]]}" + } + ], + "images": "12658_exterior_image_1_left#0", + "gt": "[[145, 0], [174, 23], [188, 66], [175, 67], [165, 25], [171, 15], [146, 34], [120, 40], [133, 43], [124, 42]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122061018#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a sliding window', and now you are performing subtask 'slide the right window.'\nThe trajectory starts at [0.48, 0.3]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.3], [0.49, 0.27], [0.48, 0.29], [0.46, 0.28], [0.46, 0.29], [0.45, 0.29], [0.44, 0.28], [0.45, 0.28], [0.48, 0.28], [0.5, 0.29]]}" + } + ], + "images": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122061018#407", + "gt": "[[307, 108], [314, 97], [307, 104], [295, 101], [297, 104], [287, 104], [284, 102], [287, 99], [309, 101], [317, 105]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0029_user_0001_scene_0004_cfg_0001_819112070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp the handle and open the drawer', and now you are performing subtask 'grasp the handle and open the drawer.'\nStarting at [0.62, 0.47], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.47], [0.6, 0.37], [0.55, 0.32], [0.5, 0.32], [0.46, 0.44], [0.49, 0.54], [0.42, 0.45], [0.43, 0.43], [0.48, 0.43], [0.51, 0.41]]}" + } + ], + "images": "RH20T_cfg1_task_0029_user_0001_scene_0004_cfg_0001_819112070044#0", + "gt": "[[400, 170], [383, 132], [352, 115], [320, 114], [292, 159], [311, 195], [270, 162], [276, 156], [308, 155], [325, 148]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "1571_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move all the plush toys from the basket to the table one by one', and now you are performing subtask 'pick up the third toy from the basket.'\nYour initial coordinate is [0.44, 0.48]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.48], [0.47, 0.49], [0.48, 0.52], [0.48, 0.54], [0.44, 0.46], [0.43, 0.45], [0.41, 0.5], [0.42, 0.59], [0.42, 0.46], [0.45, 0.41]]}" + } + ], + "images": "1571_exterior_image_1_left#168", + "gt": "[[140, 86], [149, 89], [155, 94], [155, 97], [142, 82], [137, 81], [131, 90], [134, 106], [134, 83], [143, 74]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0065_user_0007_scene_0003_cfg_0002_037522062165#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with small Lego blocks for the nth time with n blocks', and now you are performing subtask 'place the pink Lego block on the Lego base, pick up the black Lego block from the table, transfer the black Lego block from the table to the Lego base, place the black Lego block on the Lego base, pick up the green Lego block from the table, transfer the green Lego block from the table to the Lego base, place the green Lego block on the Lego base.'\nYour starting position is [0.33, 0.79]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.33, 0.79], [0.26, 0.79], [0.19, 0.86], [0.33, 0.75], [0.31, 0.79], [0.35, 0.91], [0.35, 1.0], [0.37, 1.0], [0.39, 0.89], [0.36, 0.84]]}" + } + ], + "images": "RH20T_cfg2_task_0065_user_0007_scene_0003_cfg_0002_037522062165#319", + "gt": "[[213, 283], [167, 283], [120, 308], [210, 271], [196, 284], [226, 327], [222, 360], [236, 360], [252, 319], [228, 303]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "27913_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pour the contents of the cup into the bowl', and now you are performing subtask 'pour the contents of the cup into the bowl.'\nFrom start point [0.42, 0.59], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.59], [0.45, 0.54], [0.48, 0.59], [0.5, 0.56], [0.5, 0.54], [0.49, 0.57], [0.49, 0.54], [0.49, 0.55], [0.49, 0.53], [0.49, 0.53]]}" + } + ], + "images": "27913_exterior_image_2_left#0", + "gt": "[[134, 107], [144, 97], [153, 107], [161, 101], [159, 98], [157, 102], [158, 97], [157, 99], [156, 96], [156, 96]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "40526_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the right door of the cabinet', and now you are performing subtask 'pull the handle of the right cabinet door to open the cabinet.'\nStart at [0.48, 0.29]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.29], [0.44, 0.28], [0.43, 0.29], [0.43, 0.29], [0.43, 0.28], [0.42, 0.29], [0.42, 0.3], [0.45, 0.29], [0.46, 0.28], [0.49, 0.21]]}" + } + ], + "images": "40526_exterior_image_1_left#167", + "gt": "[[153, 52], [140, 50], [138, 53], [136, 53], [136, 51], [133, 53], [133, 54], [143, 53], [147, 50], [156, 38]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0061_user_0007_scene_0001_cfg_0005_036422060215#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'place the first piece on the table, pick up the second piece from the gift box, transfer the second piece from the gift box to the table, place the second piece on the table.'\nBegin at [0.4, 0.43] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.43], [0.45, 0.39], [0.45, 0.46], [0.46, 0.43], [0.45, 0.4], [0.4, 0.43], [0.36, 0.39], [0.34, 0.41], [0.39, 0.41], [0.46, 0.35]]}" + } + ], + "images": "RH20T_cfg5_task_0061_user_0007_scene_0001_cfg_0005_036422060215#279", + "gt": "[[258, 155], [285, 142], [289, 165], [294, 153], [290, 145], [259, 153], [230, 141], [216, 146], [250, 149], [293, 125]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "35321_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the toaster', and now you are performing subtask 'turn on the toaster.'\nThe trajectory starts at [0.51, 0.01]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.01], [0.49, 0.09], [0.48, 0.23], [0.46, 0.28], [0.47, 0.31], [0.48, 0.43], [0.47, 0.38], [0.47, 0.31], [0.43, 0.28], [0.44, 0.22]]}" + } + ], + "images": "35321_exterior_image_1_left#0", + "gt": "[[163, 2], [156, 16], [154, 42], [148, 51], [151, 55], [154, 78], [149, 69], [152, 55], [138, 50], [140, 39]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0121_user_0010_scene_0002_cfg_0001_819112070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'taking out the light bulb by rotating it', and now you are performing subtask 'twist the light bulb counterclockwise to loosen it.'\nStart at [0.62, 0.46]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.46], [0.63, 0.4], [0.63, 0.39], [0.62, 0.32], [0.62, 0.32], [0.62, 0.33], [0.62, 0.34], [0.62, 0.34], [0.62, 0.37], [0.62, 0.37]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0121_user_0010_scene_0002_cfg_0001_819112070044#0.jpg", + "gt": "[[400, 167], [401, 144], [403, 139], [399, 117], [394, 115], [398, 119], [397, 123], [395, 124], [396, 132], [400, 132]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0008_user_0007_scene_0001_cfg_0005_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grab the block and place it at the designated location', and now you are performing subtask 'grab the block and place it at the designated location.'\nYour initial coordinate is [0.45, 0.29]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.29], [0.54, 0.29], [0.66, 0.34], [0.65, 0.41], [0.61, 0.47], [0.62, 0.41], [0.54, 0.37], [0.47, 0.38], [0.46, 0.4], [0.5, 0.29]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0008_user_0007_scene_0001_cfg_0005_036422060215#0.jpg", + "gt": "[[286, 106], [344, 104], [422, 124], [417, 148], [391, 168], [396, 149], [346, 133], [299, 135], [293, 144], [318, 106]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "19783_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the button on the adaptor and push the switch on the electric kettle', and now you are performing subtask 'press the switch on the electric kettle.'\nYou start from [0.54, 0.61]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.61], [0.54, 0.61], [0.55, 0.62], [0.55, 0.63], [0.55, 0.61], [0.55, 0.53], [0.56, 0.52], [0.57, 0.52], [0.56, 0.52], [0.56, 0.52]]}" + } + ], + "images": "19783_exterior_image_1_left#220", + "gt": "[[173, 109], [173, 110], [176, 111], [176, 113], [175, 110], [176, 96], [178, 93], [181, 93], [180, 93], [180, 94]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0094_user_0014_scene_0003_cfg_0004_104422070011#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'push the lid on the left side of the box.'\nThe trajectory starts at [0.4, 0.48]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.48], [0.44, 0.5], [0.47, 0.47], [0.5, 0.46], [0.52, 0.47], [0.52, 0.48], [0.5, 0.53], [0.5, 0.5], [0.49, 0.49], [0.42, 0.47]]}" + } + ], + "images": "RH20T_cfg4_task_0094_user_0014_scene_0003_cfg_0004_104422070011#203", + "gt": "[[254, 173], [283, 180], [301, 170], [320, 167], [330, 169], [331, 173], [320, 190], [319, 181], [313, 175], [271, 171]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "69016_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the second switch from the left', and now you are performing subtask 'turn on the second switch from the left.'\nYour starting position is [0.81, 0.01]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.81, 0.01], [0.73, 0.18], [0.77, 0.23], [0.74, 0.35], [0.75, 0.37], [0.77, 0.34], [0.77, 0.39], [0.81, 0.39], [0.78, 0.33], [0.78, 0.21]]}" + } + ], + "images": "69016_exterior_image_1_left#0", + "gt": "[[260, 2], [235, 33], [245, 42], [236, 63], [239, 66], [245, 61], [246, 70], [258, 70], [250, 59], [250, 38]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0130_user_0020_scene_0004_cfg_0006_104122064161#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb', and now you are performing subtask 'insert the bulb into the socket, press the switch on the socket to turn on the bulb, press the switch on the bulb.'\nStart at [0.59, 0.46]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.46], [0.6, 0.52], [0.59, 0.52], [0.61, 0.43], [0.59, 0.44], [0.61, 0.43], [0.6, 0.39], [0.55, 0.32], [0.54, 0.34], [0.55, 0.36]]}" + } + ], + "images": "RH20T_cfg6_task_0130_user_0020_scene_0004_cfg_0006_104122064161#338", + "gt": "[[380, 165], [383, 189], [378, 189], [388, 155], [379, 157], [390, 155], [387, 139], [350, 114], [343, 123], [351, 129]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122061018#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'transfer the object from the container to the surface, place the object on the surface, pick up the object in the container, transfer the object from the container to the surface, place the item on the table.'\nYour starting position is [0.41, 0.23]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.23], [0.4, 0.33], [0.4, 0.28], [0.41, 0.25], [0.46, 0.24], [0.5, 0.31], [0.51, 0.44], [0.52, 0.35], [0.48, 0.21], [0.47, 0.31]]}" + } + ], + "images": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122061018#300", + "gt": "[[264, 83], [254, 119], [254, 102], [264, 90], [294, 88], [317, 111], [324, 158], [333, 125], [308, 74], [298, 111]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "7930_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the towel to wipe the table', and now you are performing subtask 'push the fabric on the table to the left.'\nStart at [0.29, 0.0]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.29, 0.0], [0.41, 0.0], [0.54, 0.08], [0.6, 0.19], [0.67, 0.49], [0.67, 0.52], [0.62, 0.52], [0.55, 0.44], [0.54, 0.43], [0.54, 0.32]]}" + } + ], + "images": "7930_exterior_image_1_left#0", + "gt": "[[94, 0], [131, 0], [173, 15], [191, 34], [214, 89], [213, 93], [199, 93], [175, 80], [172, 77], [174, 57]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0044_user_0020_scene_0001_cfg_0007_104122061330#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up a bag of things', and now you are performing subtask 'pick up a bag of things.'\nBegin at [0.42, 0.29] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.29], [0.42, 0.37], [0.43, 0.46], [0.44, 0.48], [0.45, 0.56], [0.46, 0.58], [0.45, 0.59], [0.45, 0.5], [0.45, 0.39], [0.43, 0.36]]}" + } + ], + "images": "RH20T_cfg7_task_0044_user_0020_scene_0001_cfg_0007_104122061330#0", + "gt": "[[266, 103], [268, 134], [276, 166], [282, 173], [286, 200], [295, 210], [291, 211], [285, 181], [285, 139], [274, 128]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "25109_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the orange cube on top of the green cube', and now you are performing subtask 'put the orange cube on top of the green cube.'\nStart at [0.21, 0.54]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.21, 0.54], [0.23, 0.58], [0.25, 0.63], [0.25, 0.68], [0.26, 0.71], [0.27, 0.67], [0.31, 0.63], [0.33, 0.62], [0.33, 0.63], [0.34, 0.63]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/25109_exterior_image_2_left#0.jpg", + "gt": "[[67, 97], [74, 104], [80, 114], [81, 122], [84, 127], [87, 121], [98, 113], [105, 111], [107, 114], [109, 114]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0064_user_0020_scene_0009_cfg_0007_037522061512#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the roll of paper down from the shelf', and now you are performing subtask 'take the roll of paper down from the shelf.'\nYou begin at [0.41, 0.24]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.24], [0.37, 0.33], [0.35, 0.33], [0.25, 0.49], [0.25, 0.54], [0.26, 0.51], [0.33, 0.44], [0.38, 0.41], [0.37, 0.34], [0.39, 0.32]]}" + } + ], + "images": "RH20T_cfg7_task_0064_user_0020_scene_0009_cfg_0007_037522061512#0", + "gt": "[[261, 86], [239, 119], [226, 120], [160, 176], [163, 193], [165, 183], [210, 158], [244, 149], [235, 124], [250, 115]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0058_user_0010_scene_0010_cfg_0003_104122063550#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn off the power strips', and now you are performing subtask 'press the button on the power strip above, press the button on the power strip above.'\nYou begin at [0.62, 0.51]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.51], [0.61, 0.5], [0.62, 0.5], [0.62, 0.51], [0.62, 0.51], [0.62, 0.51], [0.61, 0.52], [0.61, 0.51], [0.63, 0.5], [0.65, 0.5]]}" + } + ], + "images": "RH20T_cfg3_task_0058_user_0010_scene_0010_cfg_0003_104122063550#88", + "gt": "[[395, 182], [393, 180], [396, 181], [394, 183], [394, 185], [394, 184], [391, 186], [390, 185], [403, 179], [418, 179]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "60705_exterior_image_2_left#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the table cloth', and now you are performing subtask 'place the unfolded tablecloth on the table.'\nFrom start point [0.61, 0.67], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.61, 0.67], [0.61, 0.68], [0.6, 0.67], [0.6, 0.68], [0.62, 0.68], [0.63, 0.71], [0.6, 0.77], [0.59, 0.73], [0.56, 0.73], [0.54, 0.58]]}" + } + ], + "images": "60705_exterior_image_2_left#389", + "gt": "[[195, 120], [194, 122], [192, 120], [193, 123], [198, 123], [201, 128], [193, 139], [190, 132], [180, 132], [174, 104]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0063_user_0020_scene_0008_cfg_0007_104122060811#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use a shovel to scoop up an object', and now you are performing subtask 'use a shovel to scoop up an object.'\nStarting at [0.6, 0.36], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.36], [0.5, 0.44], [0.54, 0.39], [0.56, 0.39], [0.54, 0.4], [0.58, 0.4], [0.55, 0.4], [0.54, 0.43], [0.57, 0.37], [0.6, 0.36]]}" + } + ], + "images": "RH20T_cfg7_task_0063_user_0020_scene_0008_cfg_0007_104122060811#0", + "gt": "[[387, 130], [320, 160], [344, 139], [359, 141], [345, 145], [372, 143], [353, 145], [346, 155], [368, 134], [387, 131]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "67326_exterior_image_2_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the T-shirt on the table', and now you are performing subtask 'unfold the T-shirt on the table.'\nStart at [0.43, 0.76]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.76], [0.56, 0.71], [0.56, 0.55], [0.6, 0.55], [0.91, 1.0], [0.88, 1.0], [0.87, 0.81], [0.8, 0.63], [0.57, 0.55], [0.57, 0.48]]}" + } + ], + "images": "67326_exterior_image_2_left#245", + "gt": "[[138, 137], [179, 128], [180, 99], [192, 99], [292, 180], [281, 180], [279, 145], [256, 114], [183, 99], [182, 87]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "82737_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move a sweet to the left', and now you are performing subtask 'move a sweet to the left.'\nBegin at [0.31, 0.3] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.31, 0.3], [0.41, 0.33], [0.51, 0.48], [0.51, 0.52], [0.47, 0.42], [0.47, 0.28], [0.47, 0.26], [0.48, 0.26], [0.52, 0.31], [0.45, 0.28]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/82737_exterior_image_1_left#0.jpg", + "gt": "[[98, 54], [130, 59], [162, 87], [164, 94], [152, 76], [149, 51], [151, 46], [155, 46], [166, 56], [145, 51]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg3_task_0057_user_0010_scene_0005_cfg_0003_104122062823#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn them on', and now you are performing subtask 'press the second button on the power board.'\nFrom start point [0.35, 0.14], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.14], [0.35, 0.13], [0.35, 0.13], [0.36, 0.13], [0.36, 0.14], [0.36, 0.14], [0.36, 0.15], [0.36, 0.16], [0.35, 0.13], [0.35, 0.11]]}" + } + ], + "images": "RH20T_cfg3_task_0057_user_0010_scene_0005_cfg_0003_104122062823#189", + "gt": "[[227, 49], [227, 47], [227, 48], [228, 48], [228, 49], [230, 52], [230, 55], [232, 57], [226, 46], [226, 40]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "3618_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the marker in the mug and put it on the table', and now you are performing subtask 'pick up the marker in the mug and put it on the table.'\nWith start location [0.85, 0.01], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.85, 0.01], [0.83, 0.16], [0.7, 0.4], [0.71, 0.53], [0.76, 0.45], [0.75, 0.53], [0.7, 0.26], [0.54, 0.42], [0.49, 0.74], [0.55, 0.49]]}" + } + ], + "images": "3618_exterior_image_2_left#0", + "gt": "[[271, 1], [266, 29], [224, 72], [226, 95], [243, 81], [241, 96], [223, 46], [173, 76], [158, 133], [176, 89]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_035622060973#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'place the third Megablock on the green base, pick up the nth Megablock from the left side of the table, place the nth Megablock on the green base.'\nYou start from [0.42, 0.41]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.41], [0.41, 0.43], [0.42, 0.42], [0.38, 0.38], [0.32, 0.42], [0.32, 0.45], [0.37, 0.41], [0.46, 0.45], [0.44, 0.5], [0.44, 0.47]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_035622060973#458.jpg", + "gt": "[[269, 147], [261, 156], [266, 150], [243, 136], [205, 150], [203, 163], [237, 146], [294, 162], [282, 181], [280, 171]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "18559_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the lid on the table', and now you are performing subtask 'place the lid on the table.'\nYou start from [0.34, 0.72]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.34, 0.72], [0.31, 0.75], [0.36, 0.7], [0.36, 0.7], [0.36, 0.71], [0.36, 0.71], [0.37, 0.71], [0.37, 0.71], [0.37, 0.71], [0.37, 0.71]]}" + } + ], + "images": "18559_exterior_image_1_left#153", + "gt": "[[108.0, 130.0], [100.0, 135.0], [114.0, 126.0], [116.0, 126.0], [115.0, 128.0], [116.0, 127.0], [118.0, 127.0], [117.0, 128.0], [117.0, 127.0], [117.0, 127.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0094_user_0007_scene_0009_cfg_0005_104122061850#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'press the cardboard box on the side, press the cardboard box on the side.'\nYou begin at [0.55, 0.36]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.36], [0.52, 0.36], [0.51, 0.36], [0.5, 0.34], [0.46, 0.32], [0.46, 0.33], [0.46, 0.32], [0.49, 0.33], [0.51, 0.35], [0.52, 0.38]]}" + } + ], + "images": "RH20T_cfg5_task_0094_user_0007_scene_0009_cfg_0005_104122061850#406", + "gt": "[[350, 129], [335, 128], [328, 131], [321, 122], [293, 116], [293, 120], [294, 116], [312, 119], [325, 127], [335, 138]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_819112070044#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'fold the third lid on the box on the table, fold the fourth lid on the box on the table.'\nYou begin at [0.7, 0.52]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.52], [0.71, 0.51], [0.71, 0.59], [0.71, 0.59], [0.71, 0.62], [0.71, 0.6], [0.72, 0.56], [0.72, 0.5], [0.7, 0.48], [0.7, 0.48]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_819112070044#154.jpg", + "gt": "[[449, 188], [454, 182], [457, 211], [453, 212], [455, 222], [454, 216], [461, 202], [461, 181], [446, 174], [446, 174]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0081_user_0007_scene_0003_cfg_0005_104422070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the leg of the glasses (with one hand)', and now you are performing subtask 'pick up the glasses on the table, fold the legs of the glasses to unfold them.'\nWith start location [0.58, 0.37], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.37], [0.58, 0.37], [0.56, 0.45], [0.59, 0.47], [0.6, 0.46], [0.6, 0.47], [0.59, 0.4], [0.6, 0.34], [0.6, 0.33], [0.6, 0.34]]}" + } + ], + "images": "RH20T_cfg5_task_0081_user_0007_scene_0003_cfg_0005_104422070044#0", + "gt": "[[374, 132], [370, 133], [360, 162], [376, 170], [384, 165], [387, 168], [379, 144], [386, 122], [385, 119], [382, 124]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_038522062547#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the second block on top of the first block, pick up the third block from the desktop, transfer the third block from the desktop to the second block above the second position, place the third block on top of the second block.'\nYour starting position is [0.71, 0.33]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.71, 0.33], [0.73, 0.28], [0.71, 0.22], [0.6, 0.17], [0.58, 0.2], [0.6, 0.28], [0.63, 0.14], [0.66, 0.21], [0.71, 0.29], [0.72, 0.27]]}" + } + ], + "images": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_038522062547#313", + "gt": "[[453, 120], [467, 101], [453, 80], [385, 61], [374, 72], [387, 101], [401, 49], [421, 77], [457, 104], [461, 96]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "47997_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the brush to sweep the candy on the table to the right', and now you are performing subtask 'use the brush to sweep the candy on the table to the right.'\nYour initial coordinate is [0.32, 0.51]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.51], [0.38, 0.39], [0.53, 0.36], [0.65, 0.49], [0.62, 0.52], [0.97, 0.76], [0.73, 0.51], [0.68, 0.49], [0.68, 0.52], [0.65, 0.53]]}" + } + ], + "images": "47997_exterior_image_1_left#0", + "gt": "[[103, 91], [122, 70], [171, 65], [207, 89], [199, 94], [311, 137], [233, 91], [216, 88], [218, 94], [209, 96]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "24041_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press down on the top of the silver flask', and now you are performing subtask 'press down on the top of the silver flask.'\nFrom start point [0.22, 0.41], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.22, 0.41], [0.32, 0.51], [0.32, 0.52], [0.31, 0.54], [0.32, 0.54], [0.31, 0.58], [0.32, 0.61], [0.32, 0.64], [0.32, 0.57], [0.32, 0.56]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/24041_exterior_image_2_left#0.jpg", + "gt": "[[71, 74], [104, 92], [104, 94], [99, 98], [101, 98], [100, 104], [104, 110], [104, 115], [104, 103], [101, 100]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_035622060973#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'pick up the first square from the right side of the table, transfer the first square from the right side of the table to the center of the cardboard, place the first square in the center of the cardboard.'\nYour starting position is [0.38, 0.36]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.36], [0.37, 0.39], [0.27, 0.39], [0.3, 0.39], [0.26, 0.34], [0.24, 0.37], [0.29, 0.43], [0.3, 0.39], [0.38, 0.39], [0.38, 0.41]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_035622060973#0.jpg", + "gt": "[[246, 130], [234, 140], [173, 140], [191, 139], [164, 124], [155, 132], [183, 156], [194, 141], [245, 141], [244, 149]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "46919_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the door of the washing machine and push the tray back in', and now you are performing subtask 'push the tray back into the washing machine.'\nYour initial coordinate is [0.48, 0.79]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.79], [0.52, 0.92], [0.58, 0.74], [0.57, 0.82], [0.58, 0.76], [0.55, 0.82], [0.47, 0.8], [0.44, 0.74], [0.45, 0.65], [0.43, 0.59]]}" + } + ], + "images": "46919_exterior_image_1_left#140", + "gt": "[[154, 142], [168, 166], [186, 133], [181, 148], [185, 136], [177, 147], [151, 144], [140, 134], [143, 117], [138, 106]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0073_user_0014_scene_0001_cfg_0007_037522061512#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the things in the drawer, close the drawer', and now you are performing subtask 'pick up the item from beside the drawer, transfer the item from the surface to the open drawer, place the item in the drawer.'\nFrom start point [0.41, 0.24], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.24], [0.37, 0.28], [0.32, 0.45], [0.33, 0.52], [0.37, 0.18], [0.41, 0.14], [0.4, 0.17], [0.38, 0.15], [0.37, 0.16], [0.35, 0.28]]}" + } + ], + "images": "RH20T_cfg7_task_0073_user_0014_scene_0001_cfg_0007_037522061512#0", + "gt": "[[261, 86], [235, 101], [204, 163], [212, 189], [236, 66], [261, 49], [254, 62], [243, 53], [235, 59], [224, 102]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0061_user_0001_scene_0006_cfg_0001_038522063145#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'pick up the large block from the gift box, transfer the large block from the gift box to the table, place the large block on the table, pick up the small block from the gift box.'\nThe trajectory starts at [0.69, 0.28]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.28], [0.7, 0.23], [0.72, 0.39], [0.71, 0.24], [0.75, 0.25], [0.8, 0.3], [0.83, 0.28], [0.82, 0.26], [0.71, 0.31], [0.69, 0.36]]}" + } + ], + "images": "RH20T_cfg1_task_0061_user_0001_scene_0006_cfg_0001_038522063145#0", + "gt": "[[444, 99], [449, 83], [458, 140], [452, 86], [477, 90], [513, 109], [529, 101], [524, 95], [457, 113], [444, 129]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "58558_exterior_image_2_left#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the clothes', and now you are performing subtask 'fold the garment once more on the table.'\nYour initial coordinate is [0.54, 0.63]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.63], [0.56, 0.72], [0.62, 0.63], [0.74, 0.61], [0.74, 0.58], [0.7, 0.52], [0.69, 0.45], [0.67, 0.36], [0.68, 0.37], [0.68, 0.39]]}" + } + ], + "images": "58558_exterior_image_2_left#513", + "gt": "[[173, 113], [178, 130], [199, 114], [237, 110], [237, 105], [223, 94], [222, 81], [213, 65], [219, 67], [219, 71]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_037522062165#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the block on the plane, transfer the block from the plane to the other blocks, place the block on the other blocks.'\nWith start location [0.32, 0.65], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.65], [0.29, 0.68], [0.28, 0.8], [0.29, 0.76], [0.38, 0.65], [0.41, 0.59], [0.42, 0.51], [0.47, 0.47], [0.51, 0.5], [0.56, 0.62]]}" + } + ], + "images": "RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_037522062165#677", + "gt": "[[207, 233], [184, 244], [178, 287], [186, 272], [246, 233], [264, 212], [271, 185], [300, 170], [329, 179], [359, 222]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "74032_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stir with a ladle in the bowl', and now you are performing subtask 'stir with a ladle in the bowl.'\nWith start location [0.8, 0.51], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.8, 0.51], [0.67, 0.43], [0.55, 0.49], [0.56, 0.59], [0.58, 0.48], [0.53, 0.5], [0.57, 0.42], [0.5, 0.54], [0.66, 0.52], [0.55, 0.42]]}" + } + ], + "images": "74032_exterior_image_1_left#0", + "gt": "[[255.0, 92.0], [213.0, 77.0], [176.0, 89.0], [178.0, 106.0], [185.0, 86.0], [169.0, 90.0], [184.0, 76.0], [159.0, 97.0], [211.0, 93.0], [177.0, 76.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "15888_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the cup to the left', and now you are performing subtask 'move the cup to the left.'\nYou start from [0.35, 0.03]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.03], [0.44, 0.08], [0.5, 0.27], [0.47, 0.26], [0.5, 0.31], [0.44, 0.41], [0.49, 0.44], [0.51, 0.54], [0.39, 0.48], [0.4, 0.4]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/15888_exterior_image_2_left#0.jpg", + "gt": "[[112, 5], [140, 14], [159, 49], [150, 47], [159, 56], [142, 74], [158, 80], [163, 98], [124, 86], [127, 72]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "39227_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'spread the towel on the countertop', and now you are performing subtask 'fold the towel.'\nYou start from [0.48, 0.67]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.67], [0.5, 0.67], [0.53, 0.68], [0.54, 0.7], [0.54, 0.73], [0.56, 0.74], [0.61, 0.73], [0.57, 0.67], [0.51, 0.67], [0.5, 0.68]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/39227_exterior_image_2_left#248.jpg", + "gt": "[[154, 121], [160, 120], [171, 123], [173, 126], [173, 131], [178, 133], [195, 132], [184, 121], [163, 121], [159, 123]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "57480_exterior_image_1_left#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the cup backwards, empty the bowl then sweep the table', and now you are performing subtask 'place the broom on the countertop.'\nYou start from [0.52, 0.63]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.63], [0.59, 0.69], [0.51, 0.76], [0.53, 0.61], [0.48, 0.81], [0.51, 0.61], [0.57, 0.67], [0.49, 0.74], [0.52, 0.62], [0.52, 0.62]]}" + } + ], + "images": "57480_exterior_image_1_left#483", + "gt": "[[166, 114], [190, 124], [163, 137], [171, 110], [154, 145], [164, 110], [181, 121], [156, 133], [167, 111], [167, 111]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "25857_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the second drawer from the right then put the left straw in the drawer', and now you are performing subtask 'throw the straw into the drawer.'\nBegin at [0.43, 0.35] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.35], [0.47, 0.38], [0.47, 0.4], [0.5, 0.38], [0.51, 0.42], [0.52, 0.42], [0.44, 0.45], [0.42, 0.54], [0.43, 0.61], [0.41, 0.61]]}" + } + ], + "images": "25857_exterior_image_1_left#264", + "gt": "[[138.5, 62.5], [149.5, 68.5], [149.5, 71.5], [161.5, 67.5], [163.5, 75.5], [167.5, 75.5], [141.5, 81.5], [133.5, 97.5], [138.5, 110.5], [131.5, 109.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "79072_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the second switch from the left', and now you are performing subtask 'press the second switch from the left on the power outlet.'\nStart at [0.84, 0.08]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.84, 0.08], [0.55, 0.26], [0.48, 0.23], [0.48, 0.26], [0.41, 0.31], [0.41, 0.39], [0.38, 0.38], [0.38, 0.36], [0.39, 0.38], [0.39, 0.36]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/79072_exterior_image_1_left#0.jpg", + "gt": "[[270, 15], [176, 47], [155, 41], [155, 46], [132, 56], [130, 71], [121, 69], [123, 65], [126, 69], [124, 64]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "51697_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker in the mug', and now you are performing subtask 'put the marker in the mug.'\nStart at [0.54, 0.12]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.12], [0.41, 0.52], [0.45, 0.63], [0.57, 0.6], [0.53, 0.68], [0.57, 0.42], [0.62, 0.39], [0.59, 0.46], [0.63, 0.49], [0.52, 0.28]]}" + } + ], + "images": "51697_exterior_image_2_left#0", + "gt": "[[174, 22], [132, 94], [144, 113], [184, 108], [169, 123], [183, 75], [200, 70], [190, 82], [201, 88], [165, 51]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "16307_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'line up the empty water bottles', and now you are performing subtask 'pick up the third empty water bottle on the table.'\nBegin at [0.69, 0.68] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.68], [0.6, 0.88], [0.56, 0.74], [0.43, 0.78], [0.43, 0.83], [0.52, 0.81], [0.54, 0.87], [0.57, 0.86], [0.52, 0.73], [0.42, 0.73]]}" + } + ], + "images": "16307_exterior_image_1_left#0", + "gt": "[[222, 122], [191, 158], [178, 134], [138, 140], [138, 150], [167, 146], [174, 156], [182, 155], [168, 132], [134, 132]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0051_user_0015_scene_0002_cfg_0001_819112070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the dish off the dish rack', and now you are performing subtask 'take the dish off the dish rack.'\nStart at [0.62, 0.46]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.46], [0.65, 0.38], [0.64, 0.44], [0.65, 0.41], [0.65, 0.42], [0.66, 0.37], [0.64, 0.37], [0.51, 0.44], [0.43, 0.55], [0.5, 0.49]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0051_user_0015_scene_0002_cfg_0001_819112070044#0.jpg", + "gt": "[[400, 167], [414, 136], [412, 160], [413, 148], [416, 152], [420, 134], [409, 133], [325, 157], [273, 198], [322, 175]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "36648_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the microwave door, close the rice cooker, push the lever on the toaster downwards', and now you are performing subtask 'push the microwave door to the closed position.'\nFrom start point [0.32, 0.54], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.54], [0.32, 0.42], [0.39, 0.22], [0.44, 0.17], [0.48, 0.18], [0.51, 0.16], [0.52, 0.19], [0.54, 0.19], [0.55, 0.2], [0.55, 0.22]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/36648_exterior_image_1_left#0.jpg", + "gt": "[[103, 97], [104, 75], [125, 39], [140, 30], [153, 32], [163, 28], [165, 35], [174, 34], [177, 36], [177, 40]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "79230_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the green bottle to the right of the silver bowl', and now you are performing subtask 'pick up the green bottle on the counter.'\nYou start from [0.53, 0.13]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.13], [0.49, 0.18], [0.47, 0.23], [0.47, 0.31], [0.46, 0.35], [0.5, 0.38], [0.5, 0.42], [0.5, 0.43], [0.49, 0.39], [0.48, 0.38]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/79230_exterior_image_2_left#0.jpg", + "gt": "[[170, 24], [157, 33], [151, 42], [150, 55], [146, 63], [159, 69], [159, 75], [161, 78], [157, 70], [155, 69]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_037522061512#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'pick up the lid on the container, transfer the lid from the container to the table, place the lid on the table.'\nWith start location [0.41, 0.24], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.24], [0.37, 0.3], [0.33, 0.43], [0.32, 0.38], [0.3, 0.39], [0.3, 0.46], [0.25, 0.55], [0.27, 0.63], [0.27, 0.5], [0.3, 0.44]]}" + } + ], + "images": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_037522061512#0", + "gt": "[[261, 86], [235, 109], [213, 155], [206, 137], [195, 141], [194, 167], [162, 199], [172, 228], [172, 179], [195, 157]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0082_user_0015_scene_0005_cfg_0004_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'scooping with a large spoon from a white bowl to a red bowl', and now you are performing subtask 'scooping with a large spoon from a white bowl to a red bowl.'\nStart at [0.45, 0.0]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.0], [0.3, 0.0], [0.55, 0.0], [0.54, 0.06], [0.55, 0.09], [0.54, 0.09], [0.54, 0.08], [0.57, 0.11], [0.56, 0.07], [0.55, 0.06]]}" + } + ], + "images": "RH20T_cfg4_task_0082_user_0015_scene_0005_cfg_0004_104122062823#0", + "gt": "[[290, 0], [190, 0], [349, 0], [345, 20], [352, 33], [348, 34], [345, 29], [368, 39], [360, 24], [353, 21]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_038522063145#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with small Lego blocks for the nth time with n blocks', and now you are performing subtask 'place the white small Lego block on the black small Lego block, press the white small Lego block, press the white small Lego block.'\nWith start location [0.7, 0.45], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.45], [0.7, 0.42], [0.7, 0.44], [0.7, 0.42], [0.7, 0.43], [0.69, 0.43], [0.69, 0.45], [0.69, 0.42], [0.7, 0.44], [0.71, 0.36]]}" + } + ], + "images": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_038522063145#1038", + "gt": "[[447, 161], [446, 152], [449, 160], [445, 151], [446, 155], [441, 154], [441, 163], [443, 152], [447, 158], [453, 128]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "49229_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn the spout to the right, then put the carrot on the plate, then put it back on the counter', and now you are performing subtask 'place the carrot toy on the counter.'\nFrom start point [0.43, 0.49], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.49], [0.43, 0.51], [0.39, 0.44], [0.41, 0.43], [0.44, 0.49], [0.43, 0.51], [0.36, 0.42], [0.35, 0.43], [0.38, 0.46], [0.36, 0.43]]}" + } + ], + "images": "49229_exterior_image_1_left#224", + "gt": "[[138, 88], [139, 91], [124, 79], [131, 78], [140, 89], [137, 91], [115, 76], [112, 78], [122, 83], [114, 77]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122061602#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a sliding window', and now you are performing subtask 'slide the right window.'\nYour initial coordinate is [0.71, 0.54]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.71, 0.54], [0.71, 0.55], [0.71, 0.57], [0.73, 0.57], [0.73, 0.56], [0.73, 0.57], [0.74, 0.57], [0.75, 0.57], [0.75, 0.56], [0.75, 0.56]]}" + } + ], + "images": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122061602#424", + "gt": "[[453, 193], [453, 199], [456, 205], [468, 205], [468, 203], [465, 204], [474, 205], [477, 204], [477, 202], [477, 202]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0108_user_0015_scene_0009_cfg_0004_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'sharpen the pencil with a pencil sharpener', and now you are performing subtask 'twist the handle of the pencil sharpener clockwise.'\nStarting at [0.45, 0.0], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.0], [0.43, 0.03], [0.37, 0.05], [0.44, 0.09], [0.47, 0.2], [0.56, 0.38], [0.55, 0.46], [0.51, 0.48], [0.48, 0.45], [0.48, 0.35]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0108_user_0015_scene_0009_cfg_0004_104122062823#0.jpg", + "gt": "[[289, 0], [274, 12], [237, 18], [281, 34], [298, 71], [357, 136], [352, 167], [329, 174], [310, 163], [306, 127]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "2876_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the black block on the green block', and now you are performing subtask 'put the black block on the green block.'\nStarting at [0.24, 0.52], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.24, 0.52], [0.31, 0.59], [0.31, 0.67], [0.3, 0.7], [0.29, 0.69], [0.29, 0.67], [0.31, 0.58], [0.33, 0.53], [0.33, 0.58], [0.32, 0.6]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/2876_exterior_image_2_left#0.jpg", + "gt": "[[78, 93], [99, 106], [100, 121], [96, 126], [94, 125], [93, 121], [99, 104], [107, 95], [106, 105], [101, 108]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0020_user_0016_scene_0002_cfg_0004_036422060909#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the desk lamp by pressing the button', and now you are performing subtask 'press the button on the desk lamp to turn on the light.'\nYou start from [0.53, 0.22]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.22], [0.47, 0.26], [0.49, 0.29], [0.46, 0.39], [0.48, 0.4], [0.47, 0.37], [0.45, 0.37], [0.46, 0.43], [0.46, 0.55], [0.49, 0.49]]}" + } + ], + "images": "RH20T_cfg4_task_0020_user_0016_scene_0002_cfg_0004_036422060909#0", + "gt": "[[337, 78], [303, 93], [311, 103], [293, 141], [306, 143], [298, 132], [290, 133], [296, 153], [292, 197], [315, 175]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "47247_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker in the cup', and now you are performing subtask 'put the marker in the cup.'\nYou start from [0.41, 0.01]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.01], [0.31, 0.33], [0.32, 0.53], [0.25, 0.44], [0.11, 0.0], [0.18, 0.07], [0.16, 0.18], [0.23, 0.18], [0.23, 0.09], [0.26, 0.12]]}" + } + ], + "images": "47247_exterior_image_2_left#0", + "gt": "[[132, 2], [98, 60], [104, 95], [79, 79], [36, 0], [59, 12], [50, 33], [75, 33], [74, 17], [84, 22]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0119_user_0014_scene_0004_cfg_0001_819112070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn off the water tap', and now you are performing subtask 'turn off the water tap.'\nYour starting position is [0.62, 0.47]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.47], [0.58, 0.44], [0.54, 0.46], [0.53, 0.48], [0.54, 0.47], [0.57, 0.48], [0.56, 0.46], [0.56, 0.43], [0.56, 0.43], [0.57, 0.43]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0119_user_0014_scene_0004_cfg_0001_819112070044#0.jpg", + "gt": "[[400, 168], [373, 159], [344, 166], [341, 172], [347, 169], [364, 173], [359, 165], [361, 154], [358, 154], [368, 155]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "14672_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the red marker out of the yellow cup and put it on the table', and now you are performing subtask 'take the red marker out of the yellow cup and put it on the table.'\nYour initial coordinate is [0.69, 0.02]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.02], [0.65, 0.33], [0.61, 0.37], [0.65, 0.34], [0.68, 0.19], [0.72, 0.14], [0.73, 0.37], [0.68, 0.41], [0.7, 0.24], [0.7, 0.24]]}" + } + ], + "images": "14672_exterior_image_1_left#0", + "gt": "[[220, 3], [208, 59], [196, 66], [207, 62], [217, 35], [231, 26], [234, 67], [218, 73], [225, 43], [225, 43]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0081_user_0015_scene_0002_cfg_0004_104422070011#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the leg of the glasses (with one hand)', and now you are performing subtask 'pick up the glasses on the table, place the glasses with the right leg unfolded on the table.'\nThe trajectory starts at [0.52, 0.29]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.29], [0.52, 0.33], [0.5, 0.4], [0.49, 0.46], [0.5, 0.45], [0.44, 0.52], [0.46, 0.51], [0.48, 0.41], [0.52, 0.38], [0.53, 0.42]]}" + } + ], + "images": "RH20T_cfg4_task_0081_user_0015_scene_0002_cfg_0004_104422070011#0", + "gt": "[[333, 103], [336, 120], [317, 144], [315, 164], [322, 163], [282, 187], [293, 185], [310, 147], [334, 136], [337, 150]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "65111_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pour the contents from the blue cup and the orange cup into the plate', and now you are performing subtask 'transfer the orange cup from the table to the plate.'\nYour initial coordinate is [0.61, 0.46]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.61, 0.46], [0.57, 0.47], [0.53, 0.56], [0.66, 0.51], [0.66, 0.36], [0.63, 0.45], [0.54, 0.52], [0.54, 0.57], [0.58, 0.41], [0.54, 0.38]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/65111_exterior_image_2_left#380.jpg", + "gt": "[[196, 82], [181, 84], [171, 100], [210, 91], [210, 65], [202, 81], [172, 93], [173, 103], [187, 73], [173, 69]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "30654_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the white mug on the right side of the counter, put some water in it and place it next to the white mug on the left', and now you are performing subtask 'take the white mug on the right side of the counter, put some water in it and place it next to the white mug on the left.'\nFrom start point [0.67, 0.26], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.67, 0.26], [0.6, 0.43], [0.46, 0.4], [0.43, 0.57], [0.43, 0.51], [0.4, 0.56], [0.23, 0.42], [0.25, 0.46], [0.0, 0.7], [0.21, 0.48]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/30654_exterior_image_1_left#0.jpg", + "gt": "[[214, 46], [191, 77], [147, 72], [137, 102], [139, 92], [129, 101], [75, 75], [80, 82], [0, 126], [68, 87]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0065_user_0016_scene_0005_cfg_0004_104122063550#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with small Lego blocks for the nth time with n blocks', and now you are performing subtask 'pick up the second block from the table, transfer the second block from the table to the Lego base, place the second block on the Lego base.'\nWith start location [0.6, 0.71], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.71], [0.61, 0.69], [0.6, 0.71], [0.61, 0.68], [0.56, 0.71], [0.57, 0.72], [0.57, 0.7], [0.59, 0.71], [0.58, 0.71], [0.59, 0.68]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0065_user_0016_scene_0005_cfg_0004_104122063550#138.jpg", + "gt": "[[382, 256], [389, 250], [386, 257], [391, 244], [360, 256], [362, 258], [364, 253], [375, 257], [372, 254], [376, 246]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "30544_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the marker out of the bowl and put it on the table', and now you are performing subtask 'take the marker out of the bowl and put it on the table.'\nThe trajectory starts at [0.49, 0.01]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.01], [0.48, 0.09], [0.52, 0.33], [0.51, 0.22], [0.53, 0.14], [0.57, 0.13], [0.6, 0.16], [0.6, 0.31], [0.58, 0.36], [0.57, 0.28]]}" + } + ], + "images": "30544_exterior_image_1_left#0", + "gt": "[[157.0, 1.5], [155.0, 15.5], [168.0, 59.5], [163.0, 40.5], [171.0, 25.5], [184.0, 23.5], [191.0, 29.5], [193.0, 56.5], [187.0, 65.5], [184.0, 49.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0029_user_0010_scene_0006_cfg_0005_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp the handle and open the drawer', and now you are performing subtask 'grab the handle of the drawer.'\nYou begin at [0.45, 0.3]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.3], [0.45, 0.29], [0.46, 0.26], [0.45, 0.28], [0.41, 0.31], [0.41, 0.3], [0.41, 0.29], [0.42, 0.31], [0.4, 0.34], [0.4, 0.34]]}" + } + ], + "images": "RH20T_cfg5_task_0029_user_0010_scene_0006_cfg_0005_036422060215#0", + "gt": "[[286, 108], [290, 106], [295, 93], [286, 100], [261, 112], [261, 108], [264, 104], [269, 113], [257, 121], [256, 124]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0037_user_0015_scene_0004_cfg_0001_819112070044#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the first block on top of the second block, pick up the third block on the table, transfer the third block from the table to above the first block, place the third block on top of the first block.'\nYour initial coordinate is [0.64, 0.57]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.64, 0.57], [0.63, 0.51], [0.57, 0.46], [0.51, 0.59], [0.53, 0.62], [0.53, 0.68], [0.55, 0.52], [0.61, 0.5], [0.62, 0.48], [0.65, 0.51]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0037_user_0015_scene_0004_cfg_0001_819112070044#254.jpg", + "gt": "[[408, 205], [404, 183], [367, 165], [326, 213], [337, 224], [337, 244], [353, 186], [391, 179], [400, 172], [413, 184]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0126_user_0014_scene_0006_cfg_0006_104122061018#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'plug in the power cord to the socket', and now you are performing subtask 'plug in the power cord to the socket.'\nStarting at [0.57, 0.08], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.08], [0.55, 0.22], [0.49, 0.32], [0.43, 0.36], [0.4, 0.43], [0.42, 0.36], [0.42, 0.27], [0.41, 0.3], [0.43, 0.27], [0.45, 0.29]]}" + } + ], + "images": "RH20T_cfg6_task_0126_user_0014_scene_0006_cfg_0006_104122061018#0", + "gt": "[[367, 28], [349, 78], [316, 115], [272, 128], [255, 156], [269, 129], [267, 96], [264, 109], [273, 98], [289, 105]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0003_user_0014_scene_0010_cfg_0006_037522061512#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the green button in the middle.'\nStart at [0.38, 0.22]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.22], [0.37, 0.23], [0.38, 0.26], [0.39, 0.26], [0.39, 0.27], [0.4, 0.28], [0.4, 0.29], [0.4, 0.28], [0.39, 0.26], [0.39, 0.26]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0003_user_0014_scene_0010_cfg_0006_037522061512#108.jpg", + "gt": "[[242, 80], [239, 84], [242, 92], [250, 95], [252, 98], [258, 99], [257, 104], [255, 99], [249, 95], [249, 95]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "18719_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the lid on the jar', and now you are performing subtask 'place the lid on the jar.'\nFrom start point [0.0, 0.57], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.0, 0.57], [0.04, 0.59], [0.0, 0.58], [0.19, 0.44], [0.31, 0.64], [0.38, 0.47], [0.54, 0.46], [0.51, 0.47], [0.52, 0.35], [0.49, 0.38]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/18719_exterior_image_2_left#0.jpg", + "gt": "[[0, 102], [13, 107], [0, 104], [61, 80], [100, 115], [122, 84], [172, 82], [164, 85], [166, 63], [156, 69]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "83937_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the rectangle block from the table and put it on top of the toy tower blocks', and now you are performing subtask 'pick up the rectangular block from the table.'\nYour starting position is [0.82, 0.13]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.82, 0.13], [0.58, 0.41], [0.49, 0.42], [0.47, 0.46], [0.45, 0.46], [0.52, 0.46], [0.57, 0.49], [0.55, 0.49], [0.52, 0.59], [0.51, 0.58]]}" + } + ], + "images": "83937_exterior_image_2_left#0", + "gt": "[[264, 24], [186, 73], [156, 75], [152, 82], [145, 82], [166, 82], [184, 88], [177, 88], [168, 107], [163, 105]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "36863_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the towel from the backrest to the seat of the chair', and now you are performing subtask 'move the towel from the backrest to the seat of the chair.'\nFrom start point [0.66, 0.35], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.66, 0.35], [0.56, 0.31], [0.52, 0.24], [0.46, 0.26], [0.42, 0.34], [0.4, 0.32], [0.43, 0.47], [0.5, 0.59], [0.48, 0.58], [0.54, 0.73]]}" + } + ], + "images": "36863_exterior_image_1_left#0", + "gt": "[[212, 63], [180, 55], [166, 43], [146, 46], [135, 62], [129, 57], [138, 84], [160, 106], [154, 105], [172, 132]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "59407_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the mug to the left', and now you are performing subtask 'move the mug to the left.'\nFrom start point [0.74, 0.29], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.74, 0.29], [0.63, 0.29], [0.58, 0.52], [0.59, 0.36], [0.53, 0.4], [0.52, 0.45], [0.55, 0.44], [0.55, 0.51], [0.56, 0.38], [0.56, 0.38]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/59407_exterior_image_2_left#0.jpg", + "gt": "[[236, 53], [201, 53], [187, 93], [190, 65], [171, 72], [167, 81], [175, 80], [175, 92], [179, 69], [179, 69]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "90712_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the soda can from the cabinet to the countertop', and now you are performing subtask 'move the soda can from the cabinet to the countertop.'\nFrom start point [0.64, 0.52], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.64, 0.52], [0.54, 0.16], [0.5, 0.14], [0.46, 0.17], [0.49, 0.12], [0.54, 0.16], [0.57, 0.23], [0.59, 0.64], [0.57, 0.74], [0.58, 0.56]]}" + } + ], + "images": "90712_exterior_image_2_left#0", + "gt": "[[206, 94], [174, 28], [161, 25], [146, 31], [157, 21], [172, 28], [184, 41], [188, 116], [182, 133], [187, 100]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "65299_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the tap nozzle to the left', and now you are performing subtask 'move the tap nozzle to the left.'\nFrom start point [0.59, 0.06], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.06], [0.62, 0.34], [0.71, 0.64], [0.75, 0.74], [0.75, 0.8], [0.73, 0.76], [0.71, 0.78], [0.67, 0.77], [0.6, 0.72], [0.54, 0.8]]}" + } + ], + "images": "65299_exterior_image_1_left#0", + "gt": "[[188.5, 10.0], [199.5, 61.0], [227.5, 116.0], [240.5, 133.0], [241.5, 144.0], [234.5, 137.0], [226.5, 141.0], [213.5, 139.0], [192.5, 129.0], [172.5, 144.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "20007_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer on the right, take out three mugs and place them on the countertop', and now you are performing subtask 'pick up the third cup from the right drawer.'\nYou begin at [0.66, 0.68]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.66, 0.68], [0.65, 0.67], [0.68, 0.63], [0.57, 0.58], [0.57, 0.64], [0.55, 0.56], [0.53, 0.62], [0.56, 0.55], [0.68, 0.64], [0.65, 0.67]]}" + } + ], + "images": "20007_exterior_image_1_left#758", + "gt": "[[210, 122], [208, 120], [217, 113], [184, 105], [183, 115], [175, 101], [171, 111], [179, 99], [216, 116], [209, 121]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "67967_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the object on top of the box to the right on top of the pink doll', and now you are performing subtask 'place the object on top of the box to the right on top of the pink doll.'\nStart at [0.62, 0.18]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.18], [0.65, 0.14], [0.68, 0.21], [0.68, 0.38], [0.73, 0.2], [0.72, 0.17], [0.68, 0.22], [0.67, 0.28], [0.7, 0.26], [0.77, 0.06]]}" + } + ], + "images": "67967_exterior_image_1_left#0", + "gt": "[[200.0, 33.0], [209.0, 25.0], [219.0, 38.0], [218.0, 68.0], [234.0, 36.0], [230.0, 31.0], [217.0, 40.0], [214.0, 51.0], [225.0, 47.0], [245.0, 10.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "951_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer, then put the pineapple and banana plush toys in the drawer, then close the drawer', and now you are performing subtask 'pick up the banana plush toy from the countertop.'\nWith start location [0.41, 0.26], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.26], [0.42, 0.31], [0.37, 0.33], [0.37, 0.37], [0.38, 0.3], [0.43, 0.28], [0.53, 0.41], [0.48, 0.32], [0.43, 0.27], [0.4, 0.31]]}" + } + ], + "images": "951_exterior_image_1_left#220", + "gt": "[[132.0, 47.5], [134.0, 55.5], [118.0, 59.5], [117.0, 66.5], [122.0, 54.5], [137.0, 50.5], [170.0, 74.5], [155.0, 57.5], [139.0, 48.5], [128.0, 55.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "4312_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the pen out of the cup', and now you are performing subtask 'take the pen out of the cup.'\nYou begin at [0.45, 0.13]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.13], [0.51, 0.15], [0.5, 0.17], [0.51, 0.19], [0.51, 0.29], [0.53, 0.32], [0.52, 0.17], [0.47, 0.14], [0.42, 0.23], [0.41, 0.34]]}" + } + ], + "images": "4312_exterior_image_1_left#0", + "gt": "[[143.0, 24.0], [164.0, 27.0], [160.0, 30.0], [164.0, 34.0], [164.0, 53.0], [171.0, 57.0], [168.0, 31.0], [151.0, 25.0], [135.0, 41.0], [131.0, 61.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0091_user_0007_scene_0006_cfg_0004_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move an object from one box to another', and now you are performing subtask 'move an object from one box to another.'\nStarting at [0.46, 0.0], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.0], [0.59, 0.11], [0.6, 0.36], [0.59, 0.18], [0.52, 0.0], [0.5, 0.0], [0.43, 0.13], [0.57, 0.0], [0.63, 0.0], [0.69, 0.09]]}" + } + ], + "images": "RH20T_cfg4_task_0091_user_0007_scene_0006_cfg_0004_104122062823#0", + "gt": "[[293, 0], [375, 41], [382, 129], [375, 65], [333, 0], [317, 0], [277, 46], [368, 0], [405, 1], [441, 32]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "56724_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick the marker off the table and put it in the cup', and now you are performing subtask 'pick the marker off the table and put it in the cup.'\nYou start from [0.78, 0.22]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.77, 0.22], [0.71, 0.37], [0.64, 0.43], [0.66, 0.54], [0.65, 0.42], [0.72, 0.39], [0.68, 0.42], [0.66, 0.41], [0.65, 0.44], [0.62, 0.43]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/56724_exterior_image_1_left#0.jpg", + "gt": "[[248, 39], [227, 66], [206, 77], [212, 97], [208, 75], [231, 70], [219, 76], [210, 73], [207, 80], [200, 77]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "15247_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push down the toaster lever', and now you are performing subtask 'push down the toaster lever.'\nYour starting position is [0.6, 0.53]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.53], [0.76, 0.58], [0.73, 0.49], [0.73, 0.55], [0.68, 0.54], [0.66, 0.56], [0.67, 0.64], [0.65, 0.7], [0.64, 0.57], [0.6, 0.54]]}" + } + ], + "images": "15247_exterior_image_1_left#0", + "gt": "[[193, 96], [243, 104], [234, 89], [232, 99], [218, 97], [210, 101], [213, 116], [209, 126], [205, 102], [192, 97]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "8488_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the green towel from right to left', and now you are performing subtask 'fold the green towel from right to left.'\nYour starting position is [0.4, 0.03]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.03], [0.43, 0.39], [0.48, 0.53], [0.46, 0.58], [0.47, 0.66], [0.43, 0.43], [0.44, 0.5], [0.43, 0.48], [0.43, 0.37], [0.44, 0.35]]}" + } + ], + "images": "8488_exterior_image_1_left#0", + "gt": "[[129, 5], [136, 71], [153, 96], [147, 104], [150, 118], [139, 78], [142, 90], [136, 87], [137, 66], [142, 63]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_038522063145#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'putting a portion of the Lego blocks on the Lego baseplate', and now you are performing subtask 'pick up the yellow Lego brick from the table, transfer the yellow Lego brick from the table to the Lego base, insert the yellow Lego brick into the Lego base.'\nStarting at [0.75, 0.39], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.75, 0.39], [0.78, 0.39], [0.77, 0.44], [0.79, 0.43], [0.78, 0.4], [0.75, 0.36], [0.74, 0.32], [0.71, 0.28], [0.68, 0.34], [0.67, 0.41]]}" + } + ], + "images": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_038522063145#260", + "gt": "[[482, 140], [500, 141], [495, 158], [504, 153], [502, 143], [478, 128], [475, 114], [457, 101], [435, 123], [426, 148]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "86000_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the bottle and put it on the counter', and now you are performing subtask 'pick up the bottle and put it on the counter.'\nWith start location [0.62, 0.74], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.74], [0.52, 0.52], [0.49, 0.52], [0.48, 0.59], [0.52, 0.42], [0.55, 0.39], [0.6, 0.62], [0.55, 0.82], [0.57, 0.79], [0.53, 0.59]]}" + } + ], + "images": "86000_exterior_image_2_left#0", + "gt": "[[198, 133], [167, 94], [158, 93], [155, 106], [167, 76], [177, 71], [192, 112], [177, 148], [182, 142], [170, 106]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "67867_exterior_image_1_left#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the pot from the right top plate to the right bottom plate of the stove', and now you are performing subtask 'transfer the pot from the right top plate of the stove to the right bottom plate of the stove.'\nYou begin at [0.65, 0.72]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.72], [0.65, 0.73], [0.65, 0.73], [0.65, 0.72], [0.63, 0.71], [0.63, 0.72], [0.64, 0.73], [0.64, 0.74], [0.64, 0.73], [0.64, 0.73]]}" + } + ], + "images": "67867_exterior_image_1_left#225", + "gt": "[[208, 130], [207, 131], [208, 131], [207, 129], [203, 127], [203, 130], [206, 131], [206, 133], [204, 131], [204, 131]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0103_user_0014_scene_0012_cfg_0004_036422060909#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'drag the plate to the goal post after holding it down', and now you are performing subtask 'push the plate towards the target bar.'\nStarting at [0.5, 0.44], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.44], [0.5, 0.44], [0.52, 0.46], [0.56, 0.5], [0.56, 0.49], [0.53, 0.37], [0.52, 0.32], [0.51, 0.24], [0.51, 0.23], [0.49, 0.24]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0103_user_0014_scene_0012_cfg_0004_036422060909#200.jpg", + "gt": "[[322, 157], [321, 158], [330, 166], [361, 181], [356, 175], [339, 132], [333, 114], [329, 88], [326, 84], [312, 88]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0073_user_0020_scene_0005_cfg_0006_037522061512#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the things in the drawer', and now you are performing subtask 'push the drawer to close the drawer.'\nYour starting position is [0.36, 0.24]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.24], [0.37, 0.26], [0.37, 0.3], [0.31, 0.36], [0.28, 0.42], [0.28, 0.34], [0.31, 0.26], [0.32, 0.19], [0.31, 0.15], [0.32, 0.16]]}" + } + ], + "images": "RH20T_cfg6_task_0073_user_0020_scene_0005_cfg_0006_037522061512#590", + "gt": "[[230, 88], [237, 95], [237, 108], [197, 128], [177, 150], [182, 123], [199, 94], [207, 70], [197, 54], [204, 56]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0081_user_0020_scene_0004_cfg_0006_104122064161#TrajectoryPrediction_Task#12", + "task": "TrajectoryPrediction_Task", + "p_index": 12, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the leg of the glasses (with one hand)', and now you are performing subtask 'fold the leg of the glasses out.'\nStarting at [0.59, 0.38], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.38], [0.6, 0.38], [0.55, 0.38], [0.54, 0.41], [0.56, 0.4], [0.59, 0.32], [0.59, 0.3], [0.58, 0.27], [0.59, 0.27], [0.6, 0.29]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0081_user_0020_scene_0004_cfg_0006_104122064161#714.jpg", + "gt": "[[377, 135], [386, 137], [355, 138], [348, 146], [358, 144], [380, 115], [380, 107], [371, 98], [380, 98], [382, 105]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "42384_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the pen from the table and put it in the bowl', and now you are performing subtask 'pick up the pen from the table and put it in the bowl.'\nStarting at [0.21, 0.37], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.21, 0.37], [0.29, 0.41], [0.39, 0.52], [0.41, 0.56], [0.43, 0.44], [0.45, 0.42], [0.49, 0.41], [0.5, 0.43], [0.55, 0.56], [0.55, 0.56]]}" + } + ], + "images": "42384_exterior_image_1_left#0", + "gt": "[[66, 67], [92, 74], [125, 93], [130, 101], [138, 80], [144, 75], [156, 74], [161, 77], [175, 101], [175, 101]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "81300_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the pen out of the bowl and put it on the table', and now you are performing subtask 'take the pen out of the bowl and put it on the table.'\nYou start from [0.12, 0.02]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.12, 0.02], [0.45, 0.41], [0.54, 0.66], [0.55, 0.48], [0.51, 0.33], [0.45, 0.27], [0.4, 0.28], [0.36, 0.37], [0.36, 0.46], [0.38, 0.43]]}" + } + ], + "images": "81300_exterior_image_1_left#0", + "gt": "[[37, 3], [144, 74], [172, 119], [177, 86], [163, 59], [143, 49], [127, 51], [115, 67], [114, 82], [121, 77]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0091_user_0020_scene_0003_cfg_0006_104122064161#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move an object from one box to another', and now you are performing subtask 'transfer the object from the left box to the right box, place the object in the right box, pick up the large square from the left, move the large square from the left gift box to above the pink gift box, place the large square in the pink gift box.'\nYour starting position is [0.58, 0.45]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.45], [0.57, 0.3], [0.55, 0.34], [0.59, 0.34], [0.59, 0.56], [0.59, 0.36], [0.62, 0.3], [0.6, 0.36], [0.6, 0.23], [0.6, 0.23]]}" + } + ], + "images": "RH20T_cfg6_task_0091_user_0020_scene_0003_cfg_0006_104122064161#181", + "gt": "[[371, 161], [364, 109], [353, 123], [377, 121], [380, 201], [378, 131], [395, 109], [383, 130], [384, 82], [384, 82]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0059_user_0010_scene_0002_cfg_0003_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn the knob to increase the volume of the speaker', and now you are performing subtask 'turn the knob clockwise to increase the volume.'\nThe trajectory starts at [0.3, 0.12]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.3, 0.12], [0.33, 0.11], [0.31, 0.17], [0.23, 0.17], [0.37, 0.16], [0.46, 0.23], [0.46, 0.33], [0.43, 0.35], [0.42, 0.42], [0.41, 0.39]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg3_task_0059_user_0010_scene_0002_cfg_0003_104122062823#0.jpg", + "gt": "[[193, 42], [209, 41], [200, 61], [149, 63], [234, 56], [297, 82], [297, 119], [278, 125], [270, 151], [265, 141]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0081_user_0020_scene_0004_cfg_0006_104122060811#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the leg of the glasses (with one hand)', and now you are performing subtask 'fold the leg of the glasses out.'\nYour initial coordinate is [0.56, 0.49]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.56, 0.49], [0.56, 0.49], [0.56, 0.48], [0.56, 0.49], [0.57, 0.48], [0.57, 0.46], [0.57, 0.46], [0.57, 0.46], [0.58, 0.47], [0.58, 0.47]]}" + } + ], + "images": "RH20T_cfg6_task_0081_user_0020_scene_0004_cfg_0006_104122060811#500", + "gt": "[[357, 175], [358, 175], [357, 174], [359, 175], [363, 172], [367, 167], [367, 165], [367, 167], [370, 171], [370, 171]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0034_user_0015_scene_0008_cfg_0001_038522062547#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'place the first block in front of the table, pick up the second block from the table, transfer the second block from the table to beside the first block, place the second block beside the first block.'\nStart at [0.6, 0.54]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.54], [0.61, 0.52], [0.61, 0.46], [0.59, 0.33], [0.56, 0.3], [0.54, 0.33], [0.6, 0.33], [0.61, 0.34], [0.61, 0.41], [0.6, 0.45]]}" + } + ], + "images": "RH20T_cfg1_task_0034_user_0015_scene_0008_cfg_0001_038522062547#208", + "gt": "[[384, 194], [390, 189], [391, 164], [375, 118], [360, 108], [345, 118], [383, 118], [391, 124], [393, 147], [383, 161]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0038_user_0020_scene_0006_cfg_0007_104122060811#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the cup', and now you are performing subtask 'pick up the cup.'\nFrom start point [0.6, 0.36], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.36], [0.59, 0.35], [0.56, 0.37], [0.55, 0.34], [0.54, 0.35], [0.55, 0.34], [0.56, 0.35], [0.56, 0.33], [0.58, 0.29], [0.57, 0.29]]}" + } + ], + "images": "RH20T_cfg7_task_0038_user_0020_scene_0006_cfg_0007_104122060811#0", + "gt": "[[387, 130], [376, 127], [357, 132], [350, 122], [346, 125], [354, 123], [356, 126], [356, 120], [369, 106], [367, 105]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122061602#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a sliding window', and now you are performing subtask 'slide the left window, slide the right window.'\nBegin at [0.73, 0.49] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.73, 0.49], [0.7, 0.5], [0.71, 0.5], [0.72, 0.51], [0.72, 0.51], [0.72, 0.49], [0.71, 0.5], [0.71, 0.52], [0.71, 0.54], [0.71, 0.54]]}" + } + ], + "images": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122061602#352", + "gt": "[[468, 177], [446, 179], [455, 180], [460, 183], [459, 182], [460, 177], [457, 179], [457, 189], [453, 193], [453, 193]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "46488_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the banana plush toy out of the box and put it on the table', and now you are performing subtask 'take the banana plush toy out of the box and put it on the table.'\nYour initial coordinate is [0.39, 0.49]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.49], [0.46, 0.45], [0.5, 0.44], [0.54, 0.48], [0.54, 0.53], [0.54, 0.47], [0.48, 0.38], [0.47, 0.38], [0.44, 0.49], [0.44, 0.46]]}" + } + ], + "images": "46488_exterior_image_2_left#0", + "gt": "[[126, 88], [147, 81], [160, 80], [172, 86], [173, 95], [172, 85], [155, 68], [149, 69], [140, 88], [141, 82]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0081_user_0015_scene_0002_cfg_0004_104422070011#TrajectoryPrediction_Task#10", + "task": "TrajectoryPrediction_Task", + "p_index": 10, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the leg of the glasses (with one hand)', and now you are performing subtask 'pick up the glasses on the table.'\nYour starting position is [0.45, 0.44]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.44], [0.45, 0.43], [0.46, 0.46], [0.44, 0.46], [0.46, 0.46], [0.47, 0.51], [0.44, 0.52], [0.41, 0.49], [0.46, 0.48], [0.43, 0.47]]}" + } + ], + "images": "RH20T_cfg4_task_0081_user_0015_scene_0002_cfg_0004_104422070011#325", + "gt": "[[287, 159], [291, 154], [292, 166], [284, 166], [296, 167], [304, 185], [279, 186], [263, 175], [295, 173], [275, 170]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "39560_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the topmost drawer, take out the towel and place it on the counter then close the drawer', and now you are performing subtask 'pick up the towel in the top drawer.'\nYour initial coordinate is [0.23, 0.81]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.22, 0.81], [0.22, 0.77], [0.25, 0.72], [0.3, 0.72], [0.32, 0.76], [0.33, 0.85], [0.36, 0.9], [0.37, 0.94], [0.39, 0.99], [0.38, 0.93]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/39560_exterior_image_1_left#163.jpg", + "gt": "[[72.0, 145.5], [71.0, 139.5], [79.0, 129.5], [95.0, 129.5], [101.0, 136.5], [107.0, 153.5], [116.0, 161.5], [118.0, 169.5], [124.0, 177.5], [123.0, 167.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "55054_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn off the switch on the kettle', and now you are performing subtask 'turn off the switch on the kettle.'\nStarting at [0.38, 0.01], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.01], [0.42, 0.13], [0.56, 0.24], [0.57, 0.33], [0.56, 0.23], [0.54, 0.34], [0.6, 0.45], [0.54, 0.48], [0.62, 0.52], [0.52, 0.53]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/55054_exterior_image_1_left#0.jpg", + "gt": "[[120, 1], [134, 23], [178, 44], [184, 60], [179, 41], [173, 62], [191, 81], [173, 86], [198, 93], [167, 95]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "19018_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the pillow and lean it against the handle of the couch, pick up the second pillow and lean it against the first pillow', and now you are performing subtask 'pick up the second pillow from the couch.'\nYou begin at [0.67, 0.42]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.67, 0.42], [0.56, 0.6], [0.56, 0.67], [0.58, 0.56], [0.67, 0.53], [0.59, 0.63], [0.69, 0.59], [0.63, 0.68], [0.72, 0.5], [0.44, 0.73]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/19018_exterior_image_1_left#0.jpg", + "gt": "[[215, 75], [178, 108], [180, 120], [185, 101], [213, 96], [188, 114], [222, 107], [203, 123], [231, 90], [142, 132]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "26153_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the salt cellar to the right of the table', and now you are performing subtask 'move the salt cellar to the right of the table.'\nFrom start point [0.44, 0.13], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.13], [0.5, 0.24], [0.47, 0.32], [0.52, 0.22], [0.65, 0.19], [0.78, 0.34], [0.79, 0.44], [0.77, 0.5], [0.78, 0.4], [0.68, 0.29]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/26153_exterior_image_1_left#0.jpg", + "gt": "[[140, 24], [161, 44], [152, 58], [166, 39], [208, 35], [249, 62], [253, 79], [247, 90], [249, 72], [217, 53]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "81019_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the fruits on the bowl and put them on the counter', and now you are performing subtask 'transfer the first fruit from the bowl to the countertop.'\nYour starting position is [0.41, 0.32]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.32], [0.5, 0.38], [0.56, 0.5], [0.56, 0.47], [0.57, 0.44], [0.6, 0.44], [0.61, 0.49], [0.59, 0.64], [0.6, 0.62], [0.6, 0.51]]}" + } + ], + "images": "81019_exterior_image_2_left#0", + "gt": "[[130, 57], [160, 68], [180, 90], [179, 85], [184, 80], [193, 80], [196, 88], [188, 115], [192, 112], [193, 92]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "32668_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the doors of the second shelf of the cabinet', and now you are performing subtask 'pull the handle of the right door of the second shelf to open the door.'\nBegin at [0.31, 0.08] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.31, 0.08], [0.28, 0.14], [0.32, 0.18], [0.42, 0.23], [0.5, 0.2], [0.51, 0.18], [0.49, 0.18], [0.45, 0.23], [0.47, 0.13], [0.4, 0.0]]}" + } + ], + "images": "32668_exterior_image_1_left#222", + "gt": "[[98, 14], [91, 26], [104, 33], [134, 41], [160, 36], [163, 32], [156, 33], [143, 42], [151, 23], [128, 0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0009_user_0014_scene_0007_cfg_0001_039422060546#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take out a Hanoi block and throw it aside', and now you are performing subtask 'take out a Hanoi block and throw it aside.'\nBegin at [0.42, 0.34] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.34], [0.51, 0.22], [0.49, 0.12], [0.5, 0.05], [0.48, 0.14], [0.52, 0.14], [0.52, 0.26], [0.51, 0.19], [0.5, 0.21], [0.51, 0.43]]}" + } + ], + "images": "RH20T_cfg1_task_0009_user_0014_scene_0007_cfg_0001_039422060546#0", + "gt": "[[267, 121], [328, 81], [313, 44], [318, 18], [310, 50], [334, 51], [336, 92], [326, 70], [322, 76], [327, 153]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "62801_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the lid on the black pot', and now you are performing subtask 'put the lid on the black pot.'\nStarting at [0.53, 0.45], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.45], [0.51, 0.34], [0.47, 0.31], [0.48, 0.35], [0.4, 0.49], [0.36, 0.49], [0.37, 0.54], [0.36, 0.44], [0.39, 0.55], [0.39, 0.5]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/62801_exterior_image_1_left#0.jpg", + "gt": "[[169.0, 81.5], [164.0, 61.5], [151.0, 56.5], [154.0, 62.5], [127.0, 87.5], [116.0, 88.5], [119.0, 96.5], [115.0, 78.5], [126.0, 98.5], [126.0, 89.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_038522062547#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn them on', and now you are performing subtask 'press the second button on the right side of the power strip.'\nYou start from [0.6, 0.82]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.82], [0.6, 0.82], [0.59, 0.83], [0.58, 0.83], [0.57, 0.83], [0.57, 0.84], [0.57, 0.83], [0.58, 0.83], [0.58, 0.82], [0.59, 0.84]]}" + } + ], + "images": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_038522062547#130", + "gt": "[[384, 296], [384, 294], [380, 299], [369, 298], [367, 299], [368, 301], [368, 298], [373, 300], [371, 295], [377, 302]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "50954_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push the orange chair back and forth', and now you are performing subtask 'push the chair back.'\nFrom start point [1.0, 0.53], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[1.0, 0.53], [0.69, 0.5], [0.59, 0.46], [0.61, 0.5], [0.54, 0.56], [0.62, 0.52], [0.76, 0.57], [0.7, 0.5], [0.69, 0.46], [0.71, 0.42]]}" + } + ], + "images": "50954_exterior_image_1_left#0", + "gt": "[[320, 96], [220, 90], [189, 82], [194, 90], [172, 101], [199, 94], [243, 102], [224, 90], [221, 83], [226, 75]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "3118_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'throw the grey cup into the black waste bin', and now you are performing subtask 'throw the grey cup into the black waste bin.'\nYou start from [0.97, 0.06]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.98, 0.06], [0.91, 0.21], [0.92, 0.31], [0.74, 0.48], [0.68, 0.5], [0.66, 0.42], [0.65, 0.48], [0.6, 0.22], [0.51, 0.09], [0.54, 0.23]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/3118_exterior_image_1_left#0.jpg", + "gt": "[[312, 10], [290, 37], [294, 55], [237, 87], [217, 90], [211, 76], [208, 86], [193, 40], [162, 16], [173, 41]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0127_user_0020_scene_0006_cfg_0006_104122061330#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'plug in the power cord of the desk lamp, turn on the socket, and light up the desk lamp', and now you are performing subtask 'press the switch on the desk lamp to turn it on.'\nYour initial coordinate is [0.65, 0.24]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.24], [0.67, 0.22], [0.69, 0.22], [0.7, 0.24], [0.69, 0.24], [0.65, 0.23], [0.59, 0.24], [0.58, 0.22], [0.59, 0.21], [0.59, 0.23]]}" + } + ], + "images": "RH20T_cfg6_task_0127_user_0020_scene_0006_cfg_0006_104122061330#597", + "gt": "[[416, 86], [429, 80], [439, 81], [445, 85], [444, 87], [417, 82], [378, 86], [374, 80], [375, 76], [379, 83]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0080_user_0015_scene_0005_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the paper with the back side facing up and put it down', and now you are performing subtask 'pick up the paper with the back side facing up and put it down.'\nYou begin at [0.63, 0.37]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.66, 0.24], [0.64, 0.28], [0.67, 0.06], [0.65, 0.0], [0.64, 0.0], [0.67, 0.14], [0.66, 0.24], [0.7, 0.42], [0.7, 0.56]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0080_user_0015_scene_0005_cfg_0001_038522062547#0.jpg", + "gt": "[[405, 132], [422, 85], [409, 99], [426, 20], [419, 0], [410, 0], [428, 49], [424, 85], [449, 151], [445, 201]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "28295_exterior_image_2_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the second switch from the right on the adapter', and now you are performing subtask 'press the second switch on the right side of the adapter.'\nThe trajectory starts at [0.38, 0.63]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.63], [0.38, 0.63], [0.38, 0.68], [0.38, 0.68], [0.38, 0.67], [0.39, 0.67], [0.39, 0.68], [0.39, 0.69], [0.39, 0.69], [0.39, 0.69]]}" + } + ], + "images": "28295_exterior_image_2_left#155", + "gt": "[[120, 113], [121, 113], [122, 122], [123, 122], [123, 121], [126, 121], [125, 123], [125, 124], [126, 124], [125, 124]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "58949_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the blue cup to the front left side of the table', and now you are performing subtask 'move the blue cup to the front left side of the table.'\nStarting at [0.64, 0.28], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.64, 0.28], [0.71, 0.33], [0.72, 0.37], [0.68, 0.42], [0.59, 0.22], [0.48, 0.31], [0.48, 0.37], [0.52, 0.4], [0.52, 0.31], [0.55, 0.24]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/58949_exterior_image_1_left#0.jpg", + "gt": "[[204, 51], [226, 59], [229, 66], [219, 75], [188, 39], [154, 55], [155, 66], [165, 72], [165, 55], [177, 44]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "64577_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn the tap nozzle to the right', and now you are performing subtask 'turn the tap nozzle to the right.'\nStart at [0.38, 0.0]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.53, 0.2], [0.57, 0.28], [0.58, 0.35], [0.62, 0.38], [0.63, 0.42], [0.57, 0.34], [0.51, 0.16], [0.55, 0.2], [0.52, 0.17]]}" + } + ], + "images": "64577_exterior_image_1_left#0", + "gt": "[[121, 0], [170, 36], [184, 51], [186, 63], [198, 68], [201, 76], [184, 61], [163, 29], [175, 36], [168, 31]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "60718_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the green grapes inside the pot in the sink', and now you are performing subtask 'put the green grapes inside the pot in the sink.'\nBegin at [0.28, 0.02] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.28, 0.02], [0.45, 0.24], [0.44, 0.35], [0.54, 0.41], [0.52, 0.46], [0.52, 0.57], [0.61, 0.56], [0.63, 0.66], [0.58, 0.65], [0.57, 0.55]]}" + } + ], + "images": "60718_exterior_image_1_left#0", + "gt": "[[91, 3], [145, 43], [140, 63], [174, 73], [165, 82], [168, 103], [196, 100], [202, 118], [185, 117], [182, 99]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_035622060973#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the third block from the desktop, transfer the third block from the desktop to the second block above the second position, place the third block on top of the second block.'\nThe trajectory starts at [0.33, 0.41]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.33, 0.41], [0.39, 0.42], [0.46, 0.51], [0.46, 0.54], [0.45, 0.59], [0.44, 0.58], [0.44, 0.51], [0.42, 0.42], [0.39, 0.39], [0.33, 0.41]]}" + } + ], + "images": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_035622060973#318", + "gt": "[[213, 149], [251, 152], [292, 182], [295, 195], [286, 212], [281, 209], [283, 184], [267, 151], [249, 140], [214, 147]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122063633#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the block on the other block, pick up the block on the surface, transfer the block from the surface to the other block, place the block on the other block.'\nThe trajectory starts at [0.34, 0.61]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.34, 0.61], [0.33, 0.66], [0.33, 0.64], [0.32, 0.67], [0.33, 0.71], [0.37, 0.53], [0.37, 0.55], [0.35, 0.51], [0.33, 0.5], [0.28, 0.52]]}" + } + ], + "images": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122063633#477", + "gt": "[[215, 219], [212, 239], [210, 231], [205, 241], [210, 254], [237, 190], [239, 197], [227, 184], [209, 180], [179, 189]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "86093_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the toy from the top of the stove and put it in the sink', and now you are performing subtask 'pick up the toy from the top of the stove and put it in the sink.'\nYou begin at [0.79, 0.04]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.79, 0.04], [0.67, 0.72], [0.53, 0.88], [0.42, 0.9], [0.39, 0.99], [0.56, 0.71], [0.68, 0.69], [0.71, 0.74], [0.67, 0.79], [0.71, 0.58]]}" + } + ], + "images": "86093_exterior_image_2_left#0", + "gt": "[[252, 7], [213, 130], [171, 158], [135, 162], [124, 179], [180, 127], [219, 124], [226, 133], [213, 143], [228, 104]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "35951_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the towel from right to left', and now you are performing subtask 'unfold the towel from right to left.'\nStart at [0.89, 0.01]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.89, 0.01], [0.86, 0.19], [0.87, 0.15], [0.88, 0.23], [0.91, 0.17], [0.98, 0.21], [0.84, 0.35], [0.78, 0.53], [0.7, 0.53], [0.69, 0.48]]}" + } + ], + "images": "35951_exterior_image_1_left#0", + "gt": "[[284, 1], [274, 34], [279, 27], [283, 41], [290, 31], [312, 38], [268, 63], [251, 95], [224, 96], [221, 86]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "79403_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the cauliflower, red ball, and light blue object into the silver bowl', and now you are performing subtask 'pick up the light blue object from the transparent box.'\nYour initial coordinate is [0.74, 0.56]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.74, 0.56], [0.74, 0.52], [0.73, 0.5], [0.71, 0.51], [0.64, 0.62], [0.64, 0.64], [0.65, 0.47], [0.65, 0.45], [0.64, 0.46], [0.64, 0.49]]}" + } + ], + "images": "79403_exterior_image_2_left#156", + "gt": "[[236, 101], [236, 93], [232, 90], [228, 91], [204, 112], [206, 116], [208, 84], [208, 81], [206, 82], [206, 88]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0080_user_0015_scene_0005_cfg_0001_035622060973#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the paper with the back side facing up and put it down', and now you are performing subtask 'pick up the paper with the back side facing up and put it down.'\nYou begin at [0.38, 0.36]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.36], [0.41, 0.65], [0.4, 0.36], [0.43, 0.31], [0.38, 0.32], [0.38, 0.29], [0.36, 0.33], [0.33, 0.31], [0.32, 0.35], [0.32, 0.35]]}" + } + ], + "images": "RH20T_cfg1_task_0080_user_0015_scene_0005_cfg_0001_035622060973#0", + "gt": "[[246, 130], [264, 233], [253, 131], [272, 110], [243, 116], [246, 106], [232, 120], [209, 112], [205, 126], [205, 126]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_104122063678#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the third block on the table, transfer the third block from the table to the position above the first block, place the third block on top of the first block, pick up the fourth block on the table.'\nStart at [0.6, 0.64]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.64], [0.62, 0.64], [0.61, 0.7], [0.63, 0.63], [0.63, 0.59], [0.63, 0.57], [0.6, 0.56], [0.57, 0.62], [0.6, 0.57], [0.6, 0.51]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_104122063678#414.jpg", + "gt": "[[382, 231], [395, 232], [388, 251], [404, 226], [405, 213], [404, 207], [386, 203], [365, 224], [382, 207], [384, 182]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "13344_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the marker from the cup', and now you are performing subtask 'remove the marker from the cup.'\nFrom start point [0.75, 0.0], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.75, 0.0], [0.58, 0.18], [0.56, 0.22], [0.57, 0.11], [0.62, 0.02], [0.58, 0.24], [0.52, 0.37], [0.53, 0.32], [0.57, 0.27], [0.57, 0.27]]}" + } + ], + "images": "13344_exterior_image_2_left#0", + "gt": "[[240, 0], [185, 32], [179, 39], [182, 20], [197, 3], [185, 43], [166, 67], [170, 58], [183, 48], [183, 48]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "65467_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the bottle from the table and put it inside the clear bowl, turn off the white extension cable, take out one bottle from the clear bowl and place it on the left side of the table', and now you are performing subtask 'place the bottle inside the clear bowl.'\nYour initial coordinate is [0.88, 0.13]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.88, 0.13], [0.83, 0.33], [0.85, 0.34], [0.79, 0.42], [0.74, 0.61], [0.75, 0.56], [0.73, 0.57], [0.73, 0.67], [0.66, 0.52], [0.66, 0.58]]}" + } + ], + "images": "65467_exterior_image_1_left#0", + "gt": "[[282, 23], [265, 59], [272, 61], [253, 76], [236, 110], [239, 100], [235, 103], [233, 121], [212, 94], [211, 105]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0004_user_0010_scene_0001_cfg_0005_104422070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the small block on the left and move it to the right', and now you are performing subtask 'pick up the small block on the left and move it to the right.'\nFrom start point [0.58, 0.37], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.37], [0.72, 0.42], [0.75, 0.48], [0.77, 0.59], [0.76, 0.63], [0.73, 0.56], [0.73, 0.52], [0.69, 0.49], [0.57, 0.48], [0.57, 0.48]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0004_user_0010_scene_0001_cfg_0005_104422070044#0.jpg", + "gt": "[[374, 132], [461, 152], [483, 173], [494, 213], [489, 228], [464, 201], [468, 188], [440, 178], [367, 173], [367, 173]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "44936_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the water bottle forward', and now you are performing subtask 'place the water bottle at the edge of the table.'\nStarting at [0.41, 0.81], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.81], [0.38, 0.82], [0.4, 0.83], [0.39, 0.89], [0.42, 0.79], [0.41, 0.79], [0.41, 0.86], [0.47, 0.88], [0.47, 0.77], [0.51, 0.7]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/44936_exterior_image_1_left#374.jpg", + "gt": "[[132, 146], [123, 147], [128, 150], [124, 160], [135, 143], [130, 142], [131, 154], [152, 158], [151, 138], [163, 126]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0061_user_0007_scene_0003_cfg_0005_104122063678#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'place the small cube on the desktop, pick up the large cube in the box, transfer the large cube from the box to the desktop, place the large square onto the table.'\nStart at [0.44, 0.3]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.3], [0.47, 0.36], [0.47, 0.48], [0.46, 0.25], [0.46, 0.3], [0.46, 0.26], [0.51, 0.38], [0.49, 0.31], [0.5, 0.29], [0.5, 0.29]]}" + } + ], + "images": "RH20T_cfg5_task_0061_user_0007_scene_0003_cfg_0005_104122063678#306", + "gt": "[[283, 109], [302, 131], [299, 173], [293, 91], [294, 109], [297, 95], [326, 135], [311, 111], [317, 106], [317, 106]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "75616_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the pen to stir the contents in the mug then empty the mug', and now you are performing subtask 'place the cup on the table.'\nYou start from [0.51, 0.59]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.59], [0.49, 0.69], [0.53, 0.54], [0.45, 0.62], [0.41, 0.71], [0.43, 0.72], [0.44, 0.65], [0.46, 0.6], [0.54, 0.51], [0.53, 0.59]]}" + } + ], + "images": "75616_exterior_image_2_left#333", + "gt": "[[162, 107], [156, 124], [169, 98], [145, 111], [132, 127], [139, 130], [140, 117], [148, 108], [173, 92], [171, 106]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0065_user_0014_scene_0002_cfg_0001_819112070044#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with small Lego blocks for the nth time with n blocks', and now you are performing subtask 'place the pink block in the building area, pick up the light green block on the desktop, transfer the light green block from the desktop to the building area, place the light green block on the pink block.'\nYou begin at [0.65, 0.53]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.53], [0.64, 0.54], [0.65, 0.53], [0.66, 0.55], [0.65, 0.56], [0.66, 0.57], [0.65, 0.59], [0.63, 0.52], [0.62, 0.52], [0.62, 0.53]]}" + } + ], + "images": "RH20T_cfg1_task_0065_user_0014_scene_0002_cfg_0001_819112070044#344", + "gt": "[[413, 191], [412, 193], [414, 191], [420, 197], [419, 202], [422, 207], [419, 211], [403, 188], [398, 188], [397, 190]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0044_user_0010_scene_0001_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up a bag of things', and now you are performing subtask 'pick up a bag of things.'\nStarting at [0.33, 0.33], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.33, 0.33], [0.53, 0.54], [0.62, 0.71], [0.56, 0.7], [0.52, 0.85], [0.55, 0.66], [0.59, 0.62], [0.59, 0.66], [0.47, 0.51], [0.38, 0.3]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0044_user_0010_scene_0001_cfg_0005_104122063678#0.jpg", + "gt": "[[213, 120], [337, 196], [396, 255], [356, 251], [335, 305], [354, 237], [379, 224], [375, 237], [303, 184], [245, 107]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "70814_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'rotate the spout from left to right', and now you are performing subtask 'rotate the spout from left to right.'\nYou begin at [0.27, 0.39]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.27, 0.39], [0.51, 0.22], [0.72, 0.42], [0.98, 0.41], [0.88, 0.41], [0.95, 0.4], [0.92, 0.4], [0.95, 0.33], [0.92, 0.25], [0.78, 0.24]]}" + } + ], + "images": "70814_exterior_image_1_left#0", + "gt": "[[87.0, 70.5], [162.0, 40.5], [229.0, 75.5], [315.0, 73.5], [280.0, 74.5], [304.0, 71.5], [293.0, 71.5], [305.0, 59.5], [295.0, 45.5], [250.0, 43.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0044_user_0010_scene_0007_cfg_0005_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up a bag of things', and now you are performing subtask 'pick up a bag of things.'\nStart at [0.49, 0.14]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.14], [0.5, 0.26], [0.55, 0.45], [0.55, 0.49], [0.52, 0.34], [0.5, 0.13], [0.48, 0.11], [0.47, 0.14], [0.45, 0.14], [0.45, 0.12]]}" + } + ], + "images": "RH20T_cfg5_task_0044_user_0010_scene_0007_cfg_0005_104122061850#0", + "gt": "[[311, 50], [319, 94], [352, 161], [351, 175], [331, 122], [319, 48], [305, 39], [299, 51], [291, 50], [289, 44]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0032_user_0020_scene_0009_cfg_0006_104122061602#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'slide the outer casing onto the gift box', and now you are performing subtask 'slide the outer casing onto the gift box.'\nWith start location [0.7, 0.29], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.29], [0.81, 0.49], [0.88, 0.69], [0.86, 0.74], [0.8, 0.65], [0.77, 0.52], [0.83, 0.49], [0.82, 0.46], [0.84, 0.49], [0.8, 0.27]]}" + } + ], + "images": "RH20T_cfg6_task_0032_user_0020_scene_0009_cfg_0006_104122061602#0", + "gt": "[[445, 106], [517, 177], [565, 248], [549, 265], [514, 233], [494, 187], [531, 175], [527, 166], [535, 177], [511, 97]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_038522063145#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn them on', and now you are performing subtask 'press the third button on the right side of the power strip, press the fourth button on the right side of the power strip.'\nYour starting position is [0.81, 0.59]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.81, 0.59], [0.79, 0.55], [0.79, 0.56], [0.78, 0.57], [0.79, 0.54], [0.8, 0.56], [0.77, 0.49], [0.76, 0.45], [0.75, 0.43], [0.74, 0.44]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_038522063145#272.jpg", + "gt": "[[517, 214], [507, 198], [506, 203], [501, 205], [505, 196], [512, 200], [492, 175], [486, 163], [478, 156], [476, 158]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "72764_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the switch on the breakfast machine', and now you are performing subtask 'press the switch on the breakfast machine.'\nYou begin at [0.49, 0.43]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.43], [0.5, 0.46], [0.6, 0.53], [0.59, 0.52], [0.59, 0.53], [0.62, 0.51], [0.66, 0.52], [0.73, 0.5], [0.67, 0.47], [0.64, 0.38]]}" + } + ], + "images": "72764_exterior_image_2_left#0", + "gt": "[[156, 77], [160, 82], [192, 96], [189, 93], [190, 95], [199, 91], [211, 93], [235, 90], [215, 84], [206, 68]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "10952_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick the wooden object on the table, put it on the case cover, take the container out of the case and put it on the box then place the dumbbell at the edge of the box', and now you are performing subtask 'place the dumbbell at the edge of the box.'\nYou start from [0.53, 0.57]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.57], [0.54, 0.58], [0.53, 0.61], [0.53, 0.58], [0.52, 0.59], [0.49, 0.57], [0.48, 0.59], [0.51, 0.53], [0.57, 0.5], [0.6, 0.51]]}" + } + ], + "images": "10952_exterior_image_1_left#550", + "gt": "[[170.5, 103.0], [172.5, 105.0], [168.5, 110.0], [168.5, 104.0], [166.5, 106.0], [155.5, 103.0], [152.5, 106.0], [162.5, 96.0], [182.5, 90.0], [192.5, 91.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "48888_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'after opening and then closing the flask lid, and then opening it again', and now you are performing subtask 'press the flask lid open.'\nYou start from [0.73, 0.47]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.73, 0.47], [0.72, 0.49], [0.7, 0.42], [0.68, 0.4], [0.67, 0.46], [0.7, 0.51], [0.66, 0.44], [0.64, 0.36], [0.65, 0.36], [0.71, 0.44]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/48888_exterior_image_1_left#185.jpg", + "gt": "[[234, 85], [231, 89], [225, 75], [217, 72], [214, 82], [224, 92], [211, 80], [204, 65], [208, 65], [226, 80]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg3_task_0018_user_0011_scene_0002_cfg_0003_036422060909#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'play jenga', and now you are performing subtask 'pull the blocks from the pile, place the blocks on the table.'\nYou begin at [0.53, 0.24]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.24], [0.51, 0.21], [0.5, 0.24], [0.5, 0.17], [0.46, 0.26], [0.47, 0.24], [0.5, 0.36], [0.48, 0.4], [0.54, 0.36], [0.54, 0.36]]}" + } + ], + "images": "RH20T_cfg3_task_0018_user_0011_scene_0002_cfg_0003_036422060909#0", + "gt": "[[340, 88], [325, 77], [323, 88], [322, 63], [292, 94], [301, 88], [320, 128], [309, 143], [347, 129], [347, 129]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0038_user_0010_scene_0003_cfg_0003_104422070011#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the cup to the top of the cup rack and then place it down', and now you are performing subtask 'move the cup to the top of the cup rack and then place it down.'\nBegin at [0.51, 0.3] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.3], [0.56, 0.43], [0.55, 0.51], [0.51, 0.49], [0.48, 0.38], [0.45, 0.33], [0.47, 0.35], [0.45, 0.28], [0.46, 0.29], [0.48, 0.23]]}" + } + ], + "images": "RH20T_cfg3_task_0038_user_0010_scene_0003_cfg_0003_104422070011#0", + "gt": "[[328, 107], [361, 154], [353, 183], [327, 176], [310, 136], [289, 120], [300, 127], [287, 101], [297, 105], [306, 83]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122063633#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'pick up the block on the flat surface, transfer the block from the flat surface to the block board, place the block on the block board.'\nYour initial coordinate is [0.37, 0.34]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.34], [0.25, 0.33], [0.24, 0.35], [0.26, 0.39], [0.26, 0.41], [0.26, 0.38], [0.29, 0.34], [0.34, 0.45], [0.31, 0.42], [0.27, 0.42]]}" + } + ], + "images": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122063633#0", + "gt": "[[234, 121], [158, 118], [155, 125], [169, 141], [167, 148], [168, 138], [186, 122], [217, 163], [198, 151], [172, 150]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0057_user_0007_scene_0003_cfg_0005_105422061350#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn on the power strips', and now you are performing subtask 'press the second button on the power socket, press the third button on the power socket.'\nYour initial coordinate is [0.41, 0.42]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.42], [0.42, 0.41], [0.42, 0.42], [0.42, 0.43], [0.43, 0.42], [0.42, 0.4], [0.41, 0.38], [0.42, 0.38], [0.46, 0.38], [0.45, 0.38]]}" + } + ], + "images": "RH20T_cfg5_task_0057_user_0007_scene_0003_cfg_0005_105422061350#397", + "gt": "[[265, 151], [268, 146], [266, 151], [271, 155], [272, 150], [267, 145], [264, 136], [270, 138], [292, 135], [291, 137]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0012_user_0010_scene_0001_cfg_0004_038522062288#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push the soccer ball into the goal', and now you are performing subtask 'push the soccer ball the goal.'\nThe trajectory starts at [0.42, 0.57]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.57], [0.41, 0.57], [0.47, 0.56], [0.48, 0.55], [0.48, 0.53], [0.48, 0.52], [0.45, 0.51], [0.35, 0.62], [0.34, 0.65], [0.32, 0.65]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0012_user_0010_scene_0001_cfg_0004_038522062288#151.jpg", + "gt": "[[266, 207], [261, 204], [302, 202], [310, 199], [309, 192], [306, 189], [291, 185], [225, 225], [216, 233], [208, 233]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "18373_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the purple plush toy', and now you are performing subtask 'move the purple plush toy.'\nYou begin at [0.65, 0.18]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.18], [0.54, 0.46], [0.49, 0.63], [0.45, 0.44], [0.46, 0.41], [0.47, 0.42], [0.48, 0.51], [0.49, 0.42], [0.47, 0.45], [0.48, 0.42]]}" + } + ], + "images": "18373_exterior_image_2_left#0", + "gt": "[[207, 33], [172, 82], [156, 113], [145, 79], [146, 74], [149, 76], [153, 91], [158, 75], [151, 81], [153, 76]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "70297_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker in the mug', and now you are performing subtask 'put the marker in the mug.'\nWith start location [0.53, 0.3], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.3], [0.46, 0.48], [0.49, 0.66], [0.43, 0.56], [0.38, 0.54], [0.3, 0.57], [0.27, 0.65], [0.28, 0.69], [0.29, 0.62], [0.35, 0.52]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/70297_exterior_image_2_left#0.jpg", + "gt": "[[168, 54], [148, 86], [158, 118], [136, 101], [120, 98], [97, 103], [85, 117], [89, 125], [94, 111], [111, 94]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0037_user_0014_scene_0010_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up a block from the table, transfer a block from the table to the top of a block, place the block on another block.'\nYou begin at [0.63, 0.37]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.65, 0.33], [0.66, 0.36], [0.65, 0.36], [0.66, 0.32], [0.7, 0.28], [0.72, 0.28], [0.73, 0.28], [0.73, 0.32], [0.72, 0.32]]}" + } + ], + "images": "RH20T_cfg1_task_0037_user_0014_scene_0010_cfg_0001_038522062547#0", + "gt": "[[406, 134], [418, 118], [420, 128], [414, 130], [424, 117], [447, 100], [461, 99], [468, 102], [465, 114], [461, 115]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "74617_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the fries plush toy and put it on the plate', and now you are performing subtask 'place the fries toy on the plate.'\nYour starting position is [0.49, 0.57]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.57], [0.49, 0.56], [0.48, 0.6], [0.47, 0.6], [0.47, 0.62], [0.47, 0.68], [0.47, 0.71], [0.48, 0.71], [0.48, 0.68], [0.5, 0.69]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/74617_exterior_image_1_left#183.jpg", + "gt": "[[157, 103], [156, 101], [154, 108], [152, 108], [150, 112], [150, 122], [152, 128], [153, 128], [154, 123], [160, 125]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "24973_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'switch off the stove light', and now you are performing subtask 'switch off the stove light.'\nStart at [0.54, 0.73]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.73], [0.62, 0.41], [0.59, 0.33], [0.56, 0.29], [0.57, 0.32], [0.47, 0.41], [0.4, 0.37], [0.42, 0.4], [0.46, 0.33], [0.48, 0.64]]}" + } + ], + "images": "24973_exterior_image_1_left#0", + "gt": "[[172, 131], [197, 74], [190, 60], [180, 53], [184, 57], [152, 73], [129, 67], [134, 72], [146, 60], [154, 115]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "1571_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move all the plush toys from the basket to the table one by one', and now you are performing subtask 'pick up the second toy from the basket.'\nYour initial coordinate is [0.73, 0.32]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.73, 0.32], [0.53, 0.0], [0.51, 0.09], [0.56, 0.05], [0.52, 0.13], [0.52, 0.38], [0.46, 0.48], [0.43, 0.4], [0.38, 0.57], [0.44, 0.48]]}" + } + ], + "images": "1571_exterior_image_1_left#0", + "gt": "[[234, 58], [169, 0], [162, 16], [179, 9], [165, 23], [168, 69], [148, 86], [137, 72], [121, 103], [140, 86]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "79766_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'straighten the cable', and now you are performing subtask 'place the line on the tabletop.'\nStart at [0.73, 0.48]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.73, 0.48], [0.63, 0.41], [0.64, 0.46], [0.62, 0.46], [0.62, 0.51], [0.63, 0.49], [0.45, 0.41], [0.3, 0.45], [0.19, 0.54], [0.25, 0.32]]}" + } + ], + "images": "79766_exterior_image_1_left#305", + "gt": "[[235, 87], [203, 74], [205, 82], [199, 82], [199, 91], [201, 88], [145, 74], [97, 81], [61, 98], [80, 57]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "86920_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'cover the pot and fold the cloth', and now you are performing subtask 'place the lid on the pot.'\nYour initial coordinate is [0.33, 0.0]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.33, 0.0], [0.43, 0.18], [0.46, 0.44], [0.51, 0.66], [0.49, 0.56], [0.54, 0.53], [0.58, 0.56], [0.63, 0.64], [0.63, 0.71], [0.57, 0.62]]}" + } + ], + "images": "86920_exterior_image_1_left#0", + "gt": "[[105, 0], [136, 33], [147, 80], [163, 119], [158, 101], [172, 96], [186, 101], [202, 115], [203, 128], [183, 111]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "16567_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn the tap nozzle to the left', and now you are performing subtask 'turn the tap nozzle to the left.'\nFrom start point [0.69, 0.01], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.01], [0.83, 0.63], [0.81, 0.64], [0.82, 0.67], [0.81, 0.75], [0.76, 0.71], [0.76, 0.66], [0.77, 0.63], [0.77, 0.61], [0.68, 0.65]]}" + } + ], + "images": "16567_exterior_image_1_left#0", + "gt": "[[222.0, 2.0], [265.0, 113.0], [259.0, 115.0], [262.0, 120.0], [259.0, 135.0], [244.0, 127.0], [242.0, 119.0], [248.0, 114.0], [248.0, 109.0], [216.0, 117.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "89114_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the plush toy in the bowl, wipe the counter with the blue towel, and move the faucet to the center', and now you are performing subtask 'place the plush toy in the bowl.'\nThe trajectory starts at [0.0, 0.03]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.0, 0.03], [0.2, 0.3], [0.28, 0.59], [0.27, 0.61], [0.28, 0.57], [0.31, 0.54], [0.33, 0.56], [0.35, 0.59], [0.35, 0.57], [0.28, 0.58]]}" + } + ], + "images": "89114_exterior_image_1_left#0", + "gt": "[[0, 6], [65, 54], [91, 106], [86, 109], [89, 102], [98, 97], [107, 100], [112, 107], [112, 103], [88, 104]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0018_user_0008_scene_0007_cfg_0001_035622060973#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'play jenga', and now you are performing subtask 'play jenga.'\nThe trajectory starts at [0.38, 0.36]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.36], [0.36, 0.49], [0.39, 0.61], [0.41, 0.6], [0.43, 0.65], [0.4, 0.6], [0.4, 0.52], [0.4, 0.49], [0.43, 0.49], [0.43, 0.47]]}" + } + ], + "images": "RH20T_cfg1_task_0018_user_0008_scene_0007_cfg_0001_035622060973#0", + "gt": "[[245, 130], [233, 177], [252, 218], [260, 217], [273, 234], [257, 216], [255, 186], [259, 177], [274, 178], [276, 171]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "40480_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the marker from the green cup', and now you are performing subtask 'remove the marker from the green cup.'\nStart at [0.26, 0.52]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.26, 0.52], [0.46, 0.63], [0.45, 0.62], [0.49, 0.68], [0.5, 0.73], [0.46, 0.76], [0.47, 0.76], [0.46, 0.78], [0.4, 0.66], [0.41, 0.71]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/40480_exterior_image_1_left#0.jpg", + "gt": "[[84, 93], [148, 113], [144, 112], [158, 123], [160, 131], [146, 137], [149, 136], [147, 141], [128, 119], [130, 128]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0064_user_0001_scene_0008_cfg_0001_039422060546#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the roll of paper down from the shelf', and now you are performing subtask 'take the roll of paper down from the shelf.'\nBegin at [0.41, 0.35] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.35], [0.38, 0.39], [0.38, 0.57], [0.34, 0.6], [0.38, 0.59], [0.39, 0.7], [0.52, 0.46], [0.52, 0.5], [0.49, 0.39], [0.4, 0.33]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0064_user_0001_scene_0008_cfg_0001_039422060546#0.jpg", + "gt": "[[261, 126], [242, 141], [242, 204], [216, 216], [241, 213], [247, 251], [330, 166], [334, 179], [316, 139], [254, 120]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0037_user_0007_scene_0004_cfg_0005_104122061850#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the second block from the surface, place the second block on top of the first block, pick up the third block from the surface.'\nWith start location [0.53, 0.31], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.31], [0.55, 0.31], [0.58, 0.33], [0.6, 0.41], [0.61, 0.52], [0.61, 0.47], [0.6, 0.41], [0.57, 0.34], [0.52, 0.26], [0.52, 0.27]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0037_user_0007_scene_0004_cfg_0005_104122061850#413.jpg", + "gt": "[[341, 113], [354, 111], [372, 118], [387, 148], [391, 187], [389, 168], [382, 146], [367, 124], [336, 93], [336, 96]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "24992_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the two cushions on the right side of the sofa', and now you are performing subtask 'pull the second cushion to the left side of the first cushion.'\nYou start from [0.6, 0.57]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.57], [0.53, 0.51], [0.48, 0.68], [0.55, 0.73], [0.57, 0.66], [0.66, 0.64], [0.6, 0.61], [0.65, 0.66], [0.67, 0.56], [0.6, 0.51]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/24992_exterior_image_1_left#303.jpg", + "gt": "[[192, 103], [171, 92], [155, 123], [175, 132], [183, 119], [211, 116], [193, 109], [209, 119], [214, 100], [192, 92]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0023_user_0014_scene_0002_cfg_0001_039422060546#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the power strip by pressing the button', and now you are performing subtask 'turn on the power strip by pressing the button.'\nYou begin at [0.41, 0.34]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.34], [0.44, 0.36], [0.43, 0.36], [0.43, 0.34], [0.41, 0.33], [0.41, 0.38], [0.43, 0.39], [0.43, 0.41], [0.42, 0.42], [0.41, 0.36]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0023_user_0014_scene_0002_cfg_0001_039422060546#0.jpg", + "gt": "[[265, 124], [280, 128], [277, 131], [275, 122], [263, 120], [265, 138], [273, 140], [275, 147], [267, 151], [263, 130]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0057_user_0010_scene_0010_cfg_0003_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn on the power strips', and now you are performing subtask 'press the first button on the power strip.'\nWith start location [0.38, 0.0], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.45, 0.0], [0.41, 0.01], [0.42, 0.09], [0.38, 0.06], [0.49, 0.15], [0.64, 0.37], [0.58, 0.43], [0.58, 0.49], [0.54, 0.51]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg3_task_0057_user_0010_scene_0010_cfg_0003_104122062823#0.jpg", + "gt": "[[246, 0], [288, 0], [260, 3], [267, 31], [243, 22], [312, 53], [412, 134], [369, 156], [373, 176], [344, 184]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0034_user_0007_scene_0005_cfg_0005_104122061850#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares in a row', and now you are performing subtask 'pick up the second square from the table, transfer the second square from the table to the table next to the first square.'\nFrom start point [0.62, 0.32], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.32], [0.64, 0.32], [0.64, 0.35], [0.63, 0.36], [0.64, 0.32], [0.64, 0.31], [0.58, 0.28], [0.54, 0.3], [0.53, 0.32], [0.54, 0.32]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0034_user_0007_scene_0005_cfg_0005_104122061850#303.jpg", + "gt": "[[395, 115], [408, 117], [411, 127], [406, 128], [410, 117], [407, 110], [374, 99], [348, 108], [339, 116], [348, 117]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122061018#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a sliding window', and now you are performing subtask 'slide the left window.'\nStart at [0.57, 0.08]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.08], [0.5, 0.28], [0.48, 0.33], [0.48, 0.34], [0.47, 0.32], [0.47, 0.34], [0.46, 0.33], [0.46, 0.36], [0.43, 0.35], [0.46, 0.34]]}" + } + ], + "images": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122061018#0", + "gt": "[[367, 28], [323, 102], [305, 120], [308, 123], [300, 116], [303, 122], [297, 118], [294, 131], [276, 127], [296, 123]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0003_user_0015_scene_0008_cfg_0001_038522062547#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the three buttons from left to right in sequence', and now you are performing subtask 'press the yellow button in the middle.'\nFrom start point [0.59, 0.14], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.14], [0.59, 0.14], [0.59, 0.12], [0.59, 0.11], [0.58, 0.09], [0.58, 0.11], [0.58, 0.12], [0.58, 0.12], [0.58, 0.12], [0.58, 0.12]]}" + } + ], + "images": "RH20T_cfg1_task_0003_user_0015_scene_0008_cfg_0001_038522062547#286", + "gt": "[[380, 50], [380, 49], [377, 44], [376, 40], [371, 33], [371, 40], [372, 42], [373, 42], [373, 44], [373, 44]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "12800_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the cloth on the wooden board, then remove the cloth and put it on the table', and now you are performing subtask 'place the cloth on the wooden board.'\nYour initial coordinate is [0.73, 0.33]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.73, 0.32], [0.47, 0.61], [0.4, 0.95], [0.38, 0.68], [0.43, 0.41], [0.53, 0.34], [0.55, 0.51], [0.63, 0.42], [0.7, 0.66], [0.64, 0.5]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/12800_exterior_image_1_left#0.jpg", + "gt": "[[234.5, 58.5], [150.5, 110.5], [129.5, 170.5], [121.5, 121.5], [138.5, 74.5], [168.5, 61.5], [175.5, 92.5], [202.5, 75.5], [225.5, 119.5], [205.5, 90.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0057_user_0007_scene_0002_cfg_0005_105422061350#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press a row of four power strip buttons to turn on the power strips', and now you are performing subtask 'press the first button on the power socket, press the second button on the power socket.'\nBegin at [0.43, 0.13] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.13], [0.44, 0.17], [0.42, 0.3], [0.39, 0.38], [0.41, 0.43], [0.43, 0.38], [0.41, 0.43], [0.42, 0.44], [0.41, 0.39], [0.41, 0.39]]}" + } + ], + "images": "RH20T_cfg5_task_0057_user_0007_scene_0002_cfg_0005_105422061350#0", + "gt": "[[276, 47], [280, 62], [271, 107], [249, 138], [264, 154], [275, 137], [260, 155], [266, 160], [263, 141], [263, 141]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_104422070044#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the third block on top of the first block, pick up the fourth block on the table, transfer the fourth block from the table to the position above the third block, place the fourth block on top of the third block.'\nStarting at [0.68, 0.44], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.44], [0.69, 0.41], [0.67, 0.37], [0.64, 0.37], [0.64, 0.43], [0.64, 0.44], [0.65, 0.36], [0.69, 0.38], [0.69, 0.4], [0.68, 0.39]]}" + } + ], + "images": "RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_104422070044#269", + "gt": "[[437, 160], [439, 147], [431, 133], [412, 134], [410, 155], [412, 158], [417, 129], [441, 136], [439, 145], [433, 141]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the first block from the table, transfer the first block from the table to the position above the second block, place the first block on top of the second block, pick up the third block from the table.'\nYou begin at [0.63, 0.37]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.65, 0.34], [0.67, 0.34], [0.65, 0.36], [0.65, 0.28], [0.62, 0.18], [0.6, 0.19], [0.6, 0.17], [0.63, 0.22], [0.65, 0.34]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_038522062547#0.jpg", + "gt": "[[406, 134], [413, 122], [429, 123], [419, 131], [415, 99], [397, 65], [387, 69], [385, 61], [405, 80], [413, 124]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0033_user_0010_scene_0010_cfg_0005_104122063678#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp a small block to sweep the other small block into the designated position', and now you are performing subtask 'push another block towards the mark.'\nThe trajectory starts at [0.62, 0.51]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.51], [0.62, 0.52], [0.61, 0.53], [0.6, 0.56], [0.62, 0.59], [0.63, 0.58], [0.63, 0.6], [0.61, 0.64], [0.64, 0.61], [0.62, 0.63]]}" + } + ], + "images": "RH20T_cfg5_task_0033_user_0010_scene_0010_cfg_0005_104122063678#361", + "gt": "[[397, 184], [399, 189], [389, 192], [387, 202], [394, 213], [401, 209], [406, 217], [390, 230], [409, 218], [397, 228]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0034_user_0007_scene_0005_cfg_0005_036422060215#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares in a row', and now you are performing subtask 'pick up the second square from the table, transfer the second square from the table to the table next to the first square, place the second square on the table next to the first square, pick up the third square from the table.'\nStarting at [0.39, 0.39], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.39], [0.35, 0.36], [0.34, 0.37], [0.34, 0.39], [0.34, 0.35], [0.35, 0.39], [0.4, 0.48], [0.39, 0.46], [0.4, 0.43], [0.4, 0.43]]}" + } + ], + "images": "RH20T_cfg5_task_0034_user_0007_scene_0005_cfg_0005_036422060215#334", + "gt": "[[247, 139], [225, 130], [215, 133], [219, 141], [218, 125], [226, 141], [254, 173], [250, 166], [253, 155], [259, 154]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "42703_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the glass lid on top of the black pot', and now you are performing subtask 'put the glass lid on top of the black pot.'\nYou start from [0.32, 0.09]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.09], [0.39, 0.16], [0.48, 0.47], [0.47, 0.61], [0.51, 0.41], [0.56, 0.33], [0.57, 0.38], [0.53, 0.4], [0.53, 0.34], [0.53, 0.34]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/42703_exterior_image_1_left#0.jpg", + "gt": "[[103, 16], [126, 29], [155, 85], [149, 109], [163, 73], [179, 60], [181, 69], [171, 72], [170, 62], [170, 62]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg3_task_0038_user_0010_scene_0003_cfg_0003_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the cup to the top of the cup rack and then place it down', and now you are performing subtask 'move the cup to the top of the cup rack and then place it down.'\nFrom start point [0.31, 0.11], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.31, 0.11], [0.34, 0.05], [0.37, 0.09], [0.36, 0.23], [0.3, 0.31], [0.31, 0.36], [0.26, 0.36], [0.25, 0.43], [0.29, 0.28], [0.26, 0.23]]}" + } + ], + "images": "RH20T_cfg3_task_0038_user_0010_scene_0003_cfg_0003_104122062823#0", + "gt": "[[197, 41], [218, 18], [235, 34], [233, 84], [195, 111], [197, 128], [169, 131], [158, 154], [187, 102], [164, 82]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "34037_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the marker from the mug and place it on the table', and now you are performing subtask 'remove the marker from the mug and place it on the table.'\nYou start from [0.47, 0.0]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.0], [0.53, 0.19], [0.52, 0.41], [0.56, 0.3], [0.58, 0.27], [0.62, 0.28], [0.64, 0.33], [0.65, 0.49], [0.66, 0.48], [0.65, 0.4]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/34037_exterior_image_1_left#0.jpg", + "gt": "[[151, 0], [170, 35], [168, 74], [179, 54], [187, 48], [198, 50], [205, 59], [208, 88], [211, 86], [207, 72]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "6997_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the peach plush toy in the box and remove the green plush from the box and put it on the tray', and now you are performing subtask 'push box to side of table.'\nThe trajectory starts at [0.53, 0.69]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.69], [0.56, 0.69], [0.56, 0.75], [0.51, 0.71], [0.49, 0.74], [0.49, 0.8], [0.53, 0.77], [0.49, 0.79], [0.47, 0.76], [0.48, 0.66]]}" + } + ], + "images": "6997_exterior_image_2_left#154", + "gt": "[[169, 124], [179, 125], [180, 135], [162, 127], [156, 133], [156, 144], [171, 138], [156, 142], [150, 137], [154, 119]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0066_user_0016_scene_0001_cfg_0004_036422060909#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large megabloks for the nth time with n blocks', and now you are performing subtask 'build with large megabloks for the nth time with n blocks.'\nStart at [0.53, 0.22]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.22], [0.52, 0.18], [0.59, 0.1], [0.57, 0.1], [0.59, 0.18], [0.59, 0.13], [0.57, 0.15], [0.6, 0.25], [0.6, 0.17], [0.57, 0.17]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0066_user_0016_scene_0001_cfg_0004_036422060909#0.jpg", + "gt": "[[341, 79], [335, 64], [376, 35], [365, 36], [376, 65], [379, 47], [365, 55], [381, 91], [385, 63], [366, 63]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0105_user_0020_scene_0001_cfg_0006_104122064161#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the object on the shelf', and now you are performing subtask 'pick up the cup on the table, transfer the cup from the table to the shelf, place the cup on the shelf.'\nYou begin at [0.64, 0.27]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.64, 0.27], [0.63, 0.38], [0.66, 0.39], [0.66, 0.37], [0.66, 0.41], [0.69, 0.24], [0.73, 0.17], [0.74, 0.22], [0.73, 0.32], [0.72, 0.26]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0105_user_0020_scene_0001_cfg_0006_104122064161#318.jpg", + "gt": "[[408, 96], [404, 135], [422, 140], [422, 132], [422, 146], [442, 86], [466, 63], [476, 80], [465, 114], [459, 93]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122060811#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the block on the surface, transfer the block from the surface to the other block, place the block on the other block.'\nBegin at [0.6, 0.35] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.35], [0.59, 0.39], [0.6, 0.44], [0.6, 0.43], [0.57, 0.42], [0.56, 0.46], [0.56, 0.43], [0.53, 0.43], [0.52, 0.45], [0.52, 0.44]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122060811#0.jpg", + "gt": "[[385, 127], [375, 142], [381, 157], [385, 156], [362, 152], [360, 166], [359, 154], [341, 154], [331, 161], [332, 157]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0020_user_0008_scene_0002_cfg_0001_750612070851#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the desk lamp by pressing the button', and now you are performing subtask 'turn on the desk lamp by pressing the button.'\nWith start location [0.52, 0.43], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.43], [0.25, 0.44], [0.23, 0.47], [0.22, 0.54], [0.25, 0.58], [0.25, 0.66], [0.27, 0.59], [0.33, 0.53], [0.42, 0.55], [0.46, 0.53]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0020_user_0008_scene_0002_cfg_0001_750612070851#0.jpg", + "gt": "[[330, 156], [157, 159], [145, 168], [141, 195], [160, 210], [162, 237], [173, 212], [210, 190], [269, 198], [292, 190]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0009_user_0010_scene_0006_cfg_0004_036422060909#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take out a Hanoi block and throw it aside', and now you are performing subtask 'pick up the Hanoi block from the surface of the wooden table.'\nFrom start point [0.53, 0.22], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.22], [0.53, 0.19], [0.52, 0.23], [0.52, 0.23], [0.52, 0.25], [0.5, 0.28], [0.49, 0.26], [0.5, 0.31], [0.5, 0.32], [0.51, 0.32]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0009_user_0010_scene_0006_cfg_0004_036422060909#0.jpg", + "gt": "[[337, 78], [339, 69], [335, 82], [331, 83], [333, 89], [322, 99], [316, 95], [323, 113], [320, 116], [327, 115]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0001_user_0014_scene_0002_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the button from top to bottom', and now you are performing subtask 'press the button from top to bottom.'\nYour initial coordinate is [0.5, 0.37]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.53, 0.3], [0.6, 0.28], [0.61, 0.33], [0.62, 0.36], [0.6, 0.39], [0.6, 0.4], [0.57, 0.56], [0.56, 0.48], [0.56, 0.48]]}" + } + ], + "images": "RH20T_cfg5_task_0001_user_0014_scene_0002_cfg_0005_104122063678#0", + "gt": "[[323, 132], [341, 109], [382, 99], [393, 118], [394, 130], [384, 139], [386, 143], [362, 201], [358, 174], [358, 174]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "4759_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the blue block on the orange plate', and now you are performing subtask 'pick up the block from the table.'\nYou start from [0.28, 0.01]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.28, 0.01], [0.38, 0.28], [0.33, 0.24], [0.34, 0.28], [0.45, 0.5], [0.46, 0.56], [0.44, 0.52], [0.46, 0.54], [0.44, 0.56], [0.44, 0.56]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/4759_exterior_image_1_left#0.jpg", + "gt": "[[89, 2], [122, 51], [106, 43], [109, 51], [143, 90], [146, 101], [141, 94], [146, 98], [141, 101], [141, 101]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "66357_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the spoon out of the bowl and put it into the cup, then slightly move the two cups to the left side of the bowl', and now you are performing subtask 'pick up the spoon from the bowl.'\nStarting at [0.67, 0.56], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.67, 0.56], [0.55, 0.57], [0.52, 0.63], [0.57, 0.72], [0.52, 0.77], [0.5, 0.74], [0.55, 0.71], [0.53, 0.68], [0.52, 0.7], [0.52, 0.7]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/66357_exterior_image_1_left#0.jpg", + "gt": "[[213, 101], [175, 103], [166, 114], [181, 129], [165, 139], [161, 134], [177, 128], [169, 122], [165, 126], [165, 126]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "68303_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the carrot plush toy backwards, move the blue towel backwards', and now you are performing subtask 'place the blue object on the back side of the table.'\nWith start location [0.35, 0.56], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.56], [0.34, 0.56], [0.35, 0.59], [0.38, 0.6], [0.36, 0.56], [0.36, 0.49], [0.37, 0.47], [0.37, 0.47], [0.38, 0.47], [0.38, 0.5]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/68303_exterior_image_1_left#78.jpg", + "gt": "[[110.5, 100.5], [109.5, 101.5], [112.5, 105.5], [120.5, 107.5], [115.5, 100.5], [115.5, 87.5], [117.5, 84.5], [119.5, 84.5], [121.5, 85.5], [122.5, 89.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0020_user_0010_scene_0007_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the desk lamp by pressing the button', and now you are performing subtask 'turn on the desk lamp by pressing the button.'\nYour starting position is [0.5, 0.37]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.51, 0.45], [0.49, 0.49], [0.5, 0.44], [0.53, 0.48], [0.55, 0.43], [0.54, 0.51], [0.56, 0.51], [0.55, 0.47], [0.6, 0.39]]}" + } + ], + "images": "RH20T_cfg5_task_0020_user_0010_scene_0007_cfg_0005_104122063678#0", + "gt": "[[323, 132], [326, 161], [311, 175], [321, 160], [338, 173], [349, 154], [345, 185], [361, 184], [353, 168], [381, 141]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "72204_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the green cylindrical object on the left side of the green triangular block on the table', and now you are performing subtask 'place the green cylindrical object on the left side of the green triangular block on the table.'\nYou begin at [0.0, 0.13]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.0, 0.13], [0.35, 0.29], [0.49, 0.39], [0.53, 0.46], [0.5, 0.53], [0.5, 0.51], [0.56, 0.44], [0.64, 0.51], [0.62, 0.44], [0.51, 0.43]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/72204_exterior_image_1_left#0.jpg", + "gt": "[[0, 24], [111, 52], [156, 70], [171, 83], [161, 96], [159, 91], [178, 79], [205, 92], [197, 80], [162, 78]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "47071_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the wooden spatula and mix the contents in the silver bowl', and now you are performing subtask 'place the spoon in the pot.'\nYou start from [0.33, 0.42]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.33, 0.42], [0.37, 0.18], [0.41, 0.18], [0.45, 0.32], [0.41, 0.37], [0.44, 0.36], [0.35, 0.23], [0.38, 0.16], [0.31, 0.33], [0.42, 0.38]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/47071_exterior_image_2_left#79.jpg", + "gt": "[[107, 76], [119, 33], [132, 32], [145, 58], [132, 67], [141, 65], [112, 41], [120, 29], [100, 59], [134, 69]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0075_user_0007_scene_0001_cfg_0005_104422070044#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the black chess pieces on the chessboard', and now you are performing subtask 'place the black piece on the chessboard, pick up the white piece on the table, transfer the white piece from the table to the chessboard.'\nStarting at [0.62, 0.41], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.41], [0.61, 0.35], [0.61, 0.33], [0.67, 0.34], [0.72, 0.48], [0.71, 0.5], [0.7, 0.46], [0.67, 0.44], [0.66, 0.44], [0.66, 0.47]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0075_user_0007_scene_0001_cfg_0005_104422070044#197.jpg", + "gt": "[[395, 148], [388, 125], [393, 119], [430, 121], [458, 172], [456, 179], [446, 166], [428, 157], [425, 157], [422, 170]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0092_user_0010_scene_0007_cfg_0003_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn the hands of a clock', and now you are performing subtask 'turn the hands of a clock.'\nBegin at [0.38, 0.0] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.37, 0.03], [0.32, 0.08], [0.34, 0.09], [0.29, 0.12], [0.26, 0.22], [0.32, 0.18], [0.39, 0.2], [0.37, 0.14], [0.38, 0.09]]}" + } + ], + "images": "RH20T_cfg3_task_0092_user_0010_scene_0007_cfg_0003_104122062823#0", + "gt": "[[242, 0], [239, 10], [202, 30], [216, 31], [187, 44], [165, 79], [205, 64], [251, 71], [234, 52], [243, 33]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "56943_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the fork', and now you are performing subtask 'move the fork.'\nYour initial coordinate is [0.37, 0.25]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.25], [0.44, 0.39], [0.47, 0.48], [0.46, 0.48], [0.5, 0.52], [0.52, 0.61], [0.51, 0.65], [0.52, 0.53], [0.52, 0.63], [0.54, 0.52]]}" + } + ], + "images": "56943_exterior_image_1_left#0", + "gt": "[[118, 45], [140, 70], [150, 86], [147, 87], [159, 94], [166, 109], [164, 117], [165, 95], [168, 114], [172, 93]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0028_user_0020_scene_0003_cfg_0007_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp the handle and close the drawer', and now you are performing subtask 'grasp the handle and close the drawer.'\nYour starting position is [0.53, 0.32]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.32], [0.55, 0.4], [0.58, 0.4], [0.59, 0.42], [0.62, 0.42], [0.57, 0.31], [0.59, 0.31], [0.58, 0.26], [0.57, 0.26], [0.56, 0.3]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0028_user_0020_scene_0003_cfg_0007_104122064161#0.jpg", + "gt": "[[337, 115], [355, 145], [370, 144], [380, 152], [399, 150], [366, 111], [377, 113], [373, 95], [368, 93], [361, 107]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0003_user_0010_scene_0010_cfg_0004_038522062288#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the three buttons in sequence', and now you are performing subtask 'press the red button in the middle, do not press the dark green button.'\nBegin at [0.25, 0.38] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.25, 0.38], [0.28, 0.4], [0.31, 0.47], [0.3, 0.41], [0.25, 0.4], [0.26, 0.42], [0.23, 0.46], [0.22, 0.43], [0.41, 0.73], [0.43, 0.82]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0003_user_0010_scene_0010_cfg_0004_038522062288#123.jpg", + "gt": "[[157, 136], [182, 144], [197, 170], [195, 146], [161, 144], [165, 152], [150, 165], [141, 156], [261, 262], [273, 295]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0070_user_0014_scene_0008_cfg_0001_039422060546#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a sliding window', and now you are performing subtask 'open a sliding window.'\nThe trajectory starts at [0.42, 0.34]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.34], [0.48, 0.44], [0.5, 0.5], [0.51, 0.48], [0.52, 0.51], [0.52, 0.56], [0.51, 0.59], [0.5, 0.6], [0.51, 0.36], [0.46, 0.28]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0070_user_0014_scene_0008_cfg_0001_039422060546#0.jpg", + "gt": "[[266, 122], [310, 159], [322, 179], [325, 173], [336, 183], [336, 202], [328, 213], [322, 216], [325, 128], [297, 100]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "51700_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the towel on the rack', and now you are performing subtask 'put the towel on the rack.'\nFrom start point [0.8, 0.13], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.8, 0.13], [0.66, 0.34], [0.54, 0.4], [0.54, 0.33], [0.45, 0.56], [0.22, 0.04], [0.39, 0.07], [0.44, 0.16], [0.55, 0.21], [0.58, 0.36]]}" + } + ], + "images": "51700_exterior_image_1_left#0", + "gt": "[[255, 24], [210, 61], [174, 72], [173, 59], [145, 101], [72, 8], [126, 12], [141, 29], [175, 38], [185, 65]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0033_user_0010_scene_0010_cfg_0005_104122063678#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp a small block to sweep the other small block into the designated position', and now you are performing subtask 'push another block towards the mark.'\nStarting at [0.62, 0.63], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.63], [0.62, 0.66], [0.61, 0.66], [0.6, 0.61], [0.6, 0.7], [0.66, 0.64], [0.65, 0.7], [0.63, 0.7], [0.63, 0.78], [0.58, 0.76]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0033_user_0010_scene_0010_cfg_0005_104122063678#536.jpg", + "gt": "[[397, 228], [397, 238], [388, 237], [386, 221], [387, 251], [420, 231], [419, 252], [405, 251], [403, 281], [373, 274]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "52523_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the cloth on the table', and now you are performing subtask 'fold the cloth on the table.'\nFrom start point [0.42, 0.61], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.61], [0.39, 0.64], [0.35, 0.61], [0.36, 0.71], [0.37, 0.54], [0.45, 0.42], [0.54, 0.47], [0.54, 0.54], [0.46, 0.31], [0.53, 0.3]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/52523_exterior_image_1_left#124.jpg", + "gt": "[[136, 110], [125, 115], [112, 110], [116, 127], [119, 98], [143, 75], [173, 85], [174, 98], [147, 56], [169, 54]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "63647_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer and then put the three plush toys inside', and now you are performing subtask 'transfer a plush toy from the top of the table to the inside of the drawer.'\nThe trajectory starts at [0.75, 0.86]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.75, 0.86], [0.7, 0.84], [0.66, 0.85], [0.63, 0.82], [0.59, 0.85], [0.59, 0.88], [0.61, 0.88], [0.64, 0.85], [0.69, 0.84], [0.72, 0.81]]}" + } + ], + "images": "63647_exterior_image_1_left#158", + "gt": "[[241.5, 155.5], [224.5, 150.5], [210.5, 152.5], [202.5, 148.5], [188.5, 153.5], [189.5, 157.5], [194.5, 158.5], [203.5, 153.5], [220.5, 150.5], [231.5, 145.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0057_user_0016_scene_0006_cfg_0002_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn on the power strips', and now you are performing subtask 'press the button on the first socket.'\nWith start location [0.44, 0.49], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.49], [0.45, 0.48], [0.43, 0.54], [0.4, 0.54], [0.37, 0.62], [0.36, 0.61], [0.39, 0.58], [0.37, 0.66], [0.39, 0.68], [0.39, 0.68]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg2_task_0057_user_0016_scene_0006_cfg_0002_037522062165#0.jpg", + "gt": "[[279, 175], [287, 174], [275, 194], [253, 195], [237, 223], [230, 218], [250, 209], [239, 236], [247, 246], [247, 246]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "31502_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the green object into the silver pot in the sink', and now you are performing subtask 'put the green object into the silver pot in the sink.'\nFrom start point [0.7, 0.28], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.28], [0.61, 0.32], [0.46, 0.58], [0.5, 0.66], [0.47, 0.75], [0.5, 0.68], [0.46, 0.67], [0.46, 0.73], [0.52, 0.61], [0.57, 0.64]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/31502_exterior_image_2_left#0.jpg", + "gt": "[[225, 51], [195, 57], [148, 104], [161, 118], [150, 135], [160, 122], [147, 121], [146, 132], [168, 110], [182, 115]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "89978_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'arrange the water bottles in a straight line', and now you are performing subtask 'place the third bottle of water in the specified position in a straight line.'\nBegin at [0.56, 0.83] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.56, 0.83], [0.49, 0.73], [0.46, 0.87], [0.46, 0.78], [0.53, 0.77], [0.5, 0.74], [0.52, 0.81], [0.56, 0.86], [0.55, 0.89], [0.53, 0.86]]}" + } + ], + "images": "89978_exterior_image_1_left#156", + "gt": "[[180, 150], [156, 131], [148, 156], [148, 140], [169, 138], [161, 133], [168, 146], [179, 154], [176, 160], [171, 155]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0034_user_0003_scene_0006_cfg_0002_104422070044#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'place the second square to the left of the first square, pick up the third square from the large square, transfer the third square from the large square to the right side of the first square, place the third square to the right of the first square.'\nYou begin at [0.59, 0.64]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.64], [0.6, 0.6], [0.57, 0.53], [0.52, 0.52], [0.5, 0.56], [0.49, 0.62], [0.49, 0.61], [0.5, 0.62], [0.64, 0.63], [0.62, 0.69]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg2_task_0034_user_0003_scene_0006_cfg_0002_104422070044#253.jpg", + "gt": "[[376, 230], [383, 216], [366, 190], [330, 188], [317, 202], [315, 225], [316, 218], [319, 225], [408, 228], [400, 247]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0125_user_0007_scene_0004_cfg_0004_036422060909#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull up the lever', and now you are performing subtask 'pull up the rod on the fixture.'\nStarting at [0.53, 0.29], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.29], [0.56, 0.3], [0.53, 0.34], [0.53, 0.31], [0.48, 0.32], [0.56, 0.29], [0.67, 0.36], [0.57, 0.32], [0.51, 0.25], [0.51, 0.28]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0125_user_0007_scene_0004_cfg_0004_036422060909#165.jpg", + "gt": "[[339, 105], [360, 107], [338, 121], [338, 110], [305, 115], [356, 106], [428, 128], [368, 114], [326, 90], [325, 99]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "2362_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the wallet from the table and put it in the black bag', and now you are performing subtask 'pick up the wallet from the table.'\nYou start from [0.54, 0.7]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.7], [0.62, 0.72], [0.56, 0.83], [0.56, 0.93], [0.52, 0.86], [0.57, 0.93], [0.53, 0.76], [0.58, 0.82], [0.6, 0.72], [0.6, 0.72]]}" + } + ], + "images": "2362_exterior_image_2_left#240", + "gt": "[[174, 126], [199, 130], [179, 149], [180, 167], [168, 154], [183, 168], [171, 137], [186, 148], [191, 130], [191, 130]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0029_user_0014_scene_0005_cfg_0006_104122060811#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer', and now you are performing subtask 'open the drawer.'\nYou begin at [0.6, 0.35]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.35], [0.6, 0.34], [0.6, 0.34], [0.61, 0.37], [0.64, 0.4], [0.6, 0.42], [0.55, 0.4], [0.54, 0.41], [0.55, 0.4], [0.58, 0.41]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0029_user_0014_scene_0005_cfg_0006_104122060811#0.jpg", + "gt": "[[385, 127], [384, 121], [387, 122], [393, 133], [411, 143], [386, 150], [349, 145], [348, 148], [350, 145], [370, 148]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0123_user_0007_scene_0007_cfg_0005_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the knife on the knife rack', and now you are performing subtask 'put the knife on the knife rack.'\nWith start location [0.47, 0.4], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.4], [0.42, 0.82], [0.43, 0.84], [0.43, 0.73], [0.41, 0.59], [0.33, 0.46], [0.34, 0.58], [0.35, 0.49], [0.42, 0.45], [0.46, 0.4]]}" + } + ], + "images": "RH20T_cfg5_task_0123_user_0007_scene_0007_cfg_0005_037522062165#0", + "gt": "[[298, 143], [267, 297], [273, 304], [277, 262], [264, 211], [213, 164], [220, 209], [225, 175], [268, 162], [292, 144]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "30862_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the grey bin to the right', and now you are performing subtask 'move the grey bin to the right.'\nYour initial coordinate is [0.17, 0.87]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.17, 0.87], [0.63, 0.65], [0.66, 0.7], [0.64, 0.69], [0.68, 0.68], [0.73, 0.78], [0.72, 0.92], [0.66, 0.56], [0.57, 0.5], [0.47, 0.56]]}" + } + ], + "images": "30862_exterior_image_1_left#0", + "gt": "[[54, 156], [201, 117], [212, 126], [204, 125], [216, 123], [233, 140], [229, 165], [211, 101], [184, 90], [150, 100]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0003_user_0015_scene_0008_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the three buttons from left to right in sequence', and now you are performing subtask 'press the green button on the right.'\nYou begin at [0.64, 0.38]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.64, 0.38], [0.65, 0.34], [0.66, 0.42], [0.6, 0.52], [0.61, 0.54], [0.63, 0.51], [0.62, 0.54], [0.62, 0.49], [0.67, 0.39], [0.65, 0.33]]}" + } + ], + "images": "RH20T_cfg1_task_0003_user_0015_scene_0008_cfg_0001_038522062547#0", + "gt": "[[407, 136], [416, 123], [423, 151], [386, 187], [391, 195], [401, 184], [399, 194], [399, 177], [431, 142], [419, 119]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "7475_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the mug backwards then forwards', and now you are performing subtask 'slide the mug forward on the table.'\nWith start location [0.59, 0.04], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.04], [0.59, 0.08], [0.6, 0.11], [0.59, 0.28], [0.57, 0.34], [0.57, 0.38], [0.56, 0.41], [0.53, 0.4], [0.47, 0.33], [0.5, 0.34]]}" + } + ], + "images": "7475_exterior_image_1_left#0", + "gt": "[[190, 7], [188, 14], [193, 20], [189, 51], [181, 62], [184, 69], [179, 74], [169, 72], [152, 59], [160, 61]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0066_user_0014_scene_0007_cfg_0007_104122061330#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'place the building blocks on the pile of building blocks on the building board, pick up the building blocks on the table, transfer the building block from the building board to the stack of building blocks, place the building blocks on the pile of building blocks on the building board.'\nThe trajectory starts at [0.45, 0.45]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.45], [0.48, 0.49], [0.52, 0.46], [0.6, 0.52], [0.61, 0.57], [0.53, 0.49], [0.53, 0.46], [0.47, 0.46], [0.47, 0.49], [0.5, 0.43]]}" + } + ], + "images": "RH20T_cfg7_task_0066_user_0014_scene_0007_cfg_0007_104122061330#512", + "gt": "[[289, 163], [306, 176], [334, 166], [385, 189], [393, 206], [338, 176], [339, 164], [303, 165], [304, 176], [319, 154]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "17460_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the tape from the shelf and put it on the table', and now you are performing subtask 'take the tape from the shelf and put it on the table.'\nStart at [0.99, 0.34]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.99, 0.34], [0.75, 0.42], [0.69, 0.54], [0.63, 0.54], [0.65, 0.49], [0.7, 0.54], [0.71, 0.63], [0.67, 0.74], [0.71, 0.82], [0.75, 0.49]]}" + } + ], + "images": "17460_exterior_image_1_left#0", + "gt": "[[317.0, 61.0], [241.0, 76.0], [220.0, 97.0], [201.0, 97.0], [209.0, 88.0], [225.0, 98.0], [227.0, 114.0], [213.0, 134.0], [228.0, 148.0], [239.0, 89.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0121_user_0010_scene_0002_cfg_0001_819112070044#TrajectoryPrediction_Task#10", + "task": "TrajectoryPrediction_Task", + "p_index": 10, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'taking out the light bulb by rotating it', and now you are performing subtask 'twist the light bulb counterclockwise to loosen it, pick up the light bulb, transfer the light bulb from the table to the table, place the light bulb on the table.'\nYour initial coordinate is [0.62, 0.37]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.37], [0.6, 0.36], [0.61, 0.34], [0.62, 0.37], [0.61, 0.34], [0.52, 0.43], [0.53, 0.5], [0.56, 0.4], [0.61, 0.35], [0.62, 0.41]]}" + } + ], + "images": "RH20T_cfg1_task_0121_user_0010_scene_0002_cfg_0001_819112070044#464", + "gt": "[[395, 132], [387, 129], [392, 123], [394, 133], [393, 124], [336, 155], [339, 179], [360, 145], [389, 126], [399, 146]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "70082_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the silver spoon in the box', and now you are performing subtask 'place the silver spoon in the box.'\nYou begin at [0.63, 0.91]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.91], [0.64, 0.93], [0.63, 0.96], [0.63, 0.94], [0.61, 0.85], [0.62, 0.84], [0.63, 0.87], [0.63, 0.89], [0.64, 0.87], [0.64, 0.84]]}" + } + ], + "images": "70082_exterior_image_1_left#382", + "gt": "[[202, 163], [204, 167], [202, 173], [203, 169], [195, 153], [199, 152], [203, 156], [202, 161], [205, 156], [204, 151]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "59016_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the towel from the bar and put it on the table', and now you are performing subtask 'remove the towel from the bar and put it on the table.'\nWith start location [0.63, 0.02], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.02], [0.53, 0.17], [0.45, 0.18], [0.44, 0.22], [0.51, 0.11], [0.56, 0.1], [0.59, 0.14], [0.6, 0.24], [0.61, 0.3], [0.62, 0.28]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/59016_exterior_image_1_left#0.jpg", + "gt": "[[202, 4], [171, 31], [143, 32], [142, 40], [163, 19], [180, 18], [190, 25], [191, 44], [194, 54], [197, 51]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0037_user_0014_scene_0009_cfg_0007_104122064161#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the third block from the desktop, transfer the block from the desktop to the first block, place the block on the first block.'\nYou begin at [0.56, 0.31]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.56, 0.31], [0.55, 0.32], [0.55, 0.33], [0.55, 0.34], [0.55, 0.35], [0.55, 0.33], [0.55, 0.3], [0.56, 0.29], [0.57, 0.29], [0.59, 0.33]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0037_user_0014_scene_0009_cfg_0007_104122064161#162.jpg", + "gt": "[[357, 111], [354, 114], [353, 119], [351, 121], [349, 127], [353, 120], [354, 109], [358, 104], [363, 106], [376, 119]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "90712_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the soda can from the cabinet to the countertop', and now you are performing subtask 'move the soda can from the cabinet to the countertop.'\nWith start location [0.34, 0.52], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.34, 0.52], [0.33, 0.17], [0.35, 0.15], [0.37, 0.18], [0.36, 0.13], [0.35, 0.39], [0.41, 0.71], [0.41, 0.73], [0.39, 0.68], [0.37, 0.54]]}" + } + ], + "images": "90712_exterior_image_1_left#0", + "gt": "[[110, 94], [106, 30], [113, 27], [117, 33], [114, 23], [113, 71], [131, 128], [130, 132], [124, 123], [118, 98]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0034_user_0007_scene_0008_cfg_0005_104122063678#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'place the first square on the second square, pick up the third square from the table, transfer the third square from the table to the position above the first square, place the third square on the first square, pick up the fourth square from the table.'\nThe trajectory starts at [0.6, 0.54]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.54], [0.65, 0.53], [0.66, 0.56], [0.66, 0.59], [0.64, 0.49], [0.63, 0.48], [0.57, 0.48], [0.55, 0.5], [0.56, 0.47], [0.63, 0.55]]}" + } + ], + "images": "RH20T_cfg5_task_0034_user_0007_scene_0008_cfg_0005_104122063678#593", + "gt": "[[387, 194], [418, 192], [424, 200], [421, 214], [411, 178], [404, 172], [363, 172], [353, 181], [358, 168], [402, 198]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "70075_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the brown toy from the front end of the table to the back end of the table', and now you are performing subtask 'move the brown toy from the front end of the table to the back end of the table.'\nFrom start point [0.55, 0.02], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.02], [0.55, 0.17], [0.52, 0.32], [0.54, 0.43], [0.54, 0.49], [0.53, 0.37], [0.55, 0.28], [0.55, 0.32], [0.53, 0.25], [0.54, 0.23]]}" + } + ], + "images": "70075_exterior_image_1_left#0", + "gt": "[[177, 3], [176, 31], [166, 57], [174, 78], [172, 89], [169, 66], [176, 50], [177, 58], [170, 45], [173, 41]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0046_user_0014_scene_0001_cfg_0007_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the cup on the cup rack', and now you are performing subtask 'put the cup on the cup rack.'\nStart at [0.53, 0.32]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.32], [0.58, 0.34], [0.58, 0.37], [0.55, 0.25], [0.49, 0.1], [0.49, 0.06], [0.47, 0.07], [0.5, 0.09], [0.51, 0.13], [0.51, 0.19]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0046_user_0014_scene_0001_cfg_0007_104122064161#0.jpg", + "gt": "[[337, 115], [371, 122], [371, 134], [349, 90], [311, 37], [311, 23], [300, 25], [317, 32], [326, 47], [325, 69]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "21063_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the pen inside the mug cup', and now you are performing subtask 'place the pen inside the cup.'\nBegin at [0.41, 0.89] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.89], [0.43, 0.87], [0.43, 0.9], [0.43, 0.88], [0.43, 0.86], [0.41, 0.89], [0.38, 0.88], [0.34, 0.9], [0.34, 0.89], [0.37, 0.85]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/21063_exterior_image_1_left#69.jpg", + "gt": "[[130.0, 159.5], [137.0, 156.5], [139.0, 161.5], [139.0, 157.5], [136.0, 155.5], [131.0, 159.5], [121.0, 157.5], [110.0, 162.5], [109.0, 159.5], [118.0, 153.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0010_user_0010_scene_0005_cfg_0005_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the handset of the telephone on the corresponding phone cradle', and now you are performing subtask 'place the handset of the telephone on the corresponding phone cradle.'\nFrom start point [0.45, 0.3], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.3], [0.44, 0.3], [0.47, 0.41], [0.51, 0.43], [0.5, 0.48], [0.5, 0.42], [0.48, 0.39], [0.37, 0.4], [0.37, 0.44], [0.42, 0.35]]}" + } + ], + "images": "RH20T_cfg5_task_0010_user_0010_scene_0005_cfg_0005_036422060215#0", + "gt": "[[286, 108], [279, 109], [299, 147], [325, 156], [323, 172], [321, 150], [306, 140], [234, 143], [238, 157], [269, 125]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "80082_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker in the mug', and now you are performing subtask 'put the marker in the mug.'\nThe trajectory starts at [0.5, 0.04]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.04], [0.52, 0.08], [0.52, 0.17], [0.49, 0.29], [0.42, 0.44], [0.42, 0.51], [0.4, 0.44], [0.42, 0.38], [0.43, 0.46], [0.44, 0.42]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/80082_exterior_image_2_left#0.jpg", + "gt": "[[159, 8], [166, 14], [167, 30], [158, 53], [134, 80], [134, 92], [128, 80], [133, 69], [139, 82], [142, 76]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0132_user_0020_scene_0002_cfg_0006_104122061330#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks into a cross shape', and now you are performing subtask 'pick up the first block from the left side, transfer the first block from the left side to the designated position.'\nYour starting position is [0.42, 0.27]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.27], [0.43, 0.35], [0.49, 0.44], [0.5, 0.64], [0.51, 0.58], [0.5, 0.64], [0.53, 0.65], [0.53, 0.72], [0.5, 0.75], [0.5, 0.7]]}" + } + ], + "images": "RH20T_cfg6_task_0132_user_0020_scene_0002_cfg_0006_104122061330#0", + "gt": "[[266, 97], [274, 125], [313, 159], [322, 229], [328, 210], [321, 229], [341, 233], [341, 258], [323, 270], [322, 253]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "79932_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the smaller mug on the counter, fill it with water and place it on the left side of the counter', and now you are performing subtask 'take the smaller mug on the counter, fill it with water and place it on the left side of the counter.'\nWith start location [0.54, 0.31], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.31], [0.61, 0.34], [0.59, 0.45], [0.47, 0.44], [0.43, 0.61], [0.38, 0.54], [0.37, 0.61], [0.28, 0.59], [0.11, 0.78], [0.22, 0.61]]}" + } + ], + "images": "79932_exterior_image_1_left#0", + "gt": "[[172, 55], [195, 62], [188, 81], [150, 80], [137, 110], [121, 97], [118, 109], [88, 106], [34, 141], [69, 110]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "24404_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the blue can's bottom drawer and remove the green object', and now you are performing subtask 'place the green object in the blue basket.'\nStarting at [0.63, 0.39], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.39], [0.58, 0.49], [0.59, 0.48], [0.57, 0.34], [0.52, 0.27], [0.47, 0.24], [0.35, 0.34], [0.38, 0.32], [0.4, 0.26], [0.44, 0.23]]}" + } + ], + "images": "24404_exterior_image_1_left#202", + "gt": "[[201, 70], [186, 88], [190, 86], [183, 62], [168, 48], [151, 43], [113, 61], [122, 57], [129, 47], [142, 42]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0006_user_0010_scene_0008_cfg_0004_104122062823#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push the small block from the right side of the table to the left side', and now you are performing subtask 'push the small block from the right side of the table to the left side.'\nWith start location [0.59, 0.26], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.26], [0.59, 0.24], [0.6, 0.26], [0.59, 0.26], [0.55, 0.21], [0.47, 0.16], [0.28, 0.06], [0.3, 0.07], [0.37, 0.0], [0.5, 0.0]]}" + } + ], + "images": "RH20T_cfg4_task_0006_user_0010_scene_0008_cfg_0004_104122062823#336", + "gt": "[[376, 94], [380, 87], [382, 94], [376, 93], [353, 76], [302, 57], [180, 22], [195, 26], [236, 0], [320, 0]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0031_user_0020_scene_0007_cfg_0007_037522061512#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'cover the box', and now you are performing subtask 'cover the box.'\nFrom start point [0.41, 0.24], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.24], [0.4, 0.31], [0.4, 0.28], [0.42, 0.32], [0.41, 0.26], [0.38, 0.24], [0.32, 0.35], [0.3, 0.45], [0.32, 0.33], [0.35, 0.28]]}" + } + ], + "images": "RH20T_cfg7_task_0031_user_0020_scene_0007_cfg_0007_037522061512#0", + "gt": "[[261, 86], [255, 112], [257, 101], [266, 117], [260, 95], [243, 88], [208, 127], [193, 163], [207, 120], [223, 99]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0103_user_0007_scene_0002_cfg_0005_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'slide the plate into the goal post', and now you are performing subtask 'slide the plate into the goal post.'\nYou start from [0.45, 0.3]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.3], [0.42, 0.28], [0.38, 0.36], [0.38, 0.46], [0.45, 0.39], [0.42, 0.39], [0.44, 0.38], [0.43, 0.36], [0.48, 0.33], [0.48, 0.33]]}" + } + ], + "images": "RH20T_cfg5_task_0103_user_0007_scene_0002_cfg_0005_036422060215#0", + "gt": "[[286, 108], [270, 102], [243, 130], [244, 166], [291, 141], [268, 141], [281, 136], [278, 130], [309, 119], [309, 119]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "49916_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the lemon from the white plate and put it on the nightstand', and now you are performing subtask 'transfer the lemon from the white plate to the nightstand.'\nYou start from [0.35, 0.51]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.51], [0.36, 0.51], [0.36, 0.5], [0.36, 0.51], [0.35, 0.47], [0.33, 0.42], [0.3, 0.42], [0.3, 0.46], [0.31, 0.46], [0.31, 0.46]]}" + } + ], + "images": "49916_exterior_image_2_left#102", + "gt": "[[113, 91], [116, 91], [115, 90], [115, 92], [113, 84], [106, 76], [97, 76], [97, 83], [100, 83], [100, 83]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "32490_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the black marker on the mat, then remove the marker from the mat', and now you are performing subtask 'put the black marker on the mat, then remove the marker from the mat.'\nThe trajectory starts at [0.71, 0.01]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.71, 0.01], [0.73, 0.04], [0.63, 0.28], [0.64, 0.38], [0.62, 0.45], [0.65, 0.51], [0.62, 0.48], [0.65, 0.37], [0.71, 0.34], [0.7, 0.31]]}" + } + ], + "images": "32490_exterior_image_1_left#0", + "gt": "[[226, 1], [233, 7], [203, 51], [206, 69], [199, 81], [207, 91], [198, 87], [207, 66], [228, 61], [223, 55]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "82446_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker in the cup', and now you are performing subtask 'put the marker in the cup.'\nYour starting position is [0.28, 0.22]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.28, 0.22], [0.31, 0.43], [0.34, 0.47], [0.43, 0.29], [0.52, 0.27], [0.65, 0.31], [0.65, 0.39], [0.72, 0.24], [0.75, 0.27], [0.75, 0.24]]}" + } + ], + "images": "82446_exterior_image_2_left#0", + "gt": "[[91, 39], [99, 78], [108, 85], [139, 53], [167, 49], [209, 55], [209, 71], [230, 44], [239, 49], [239, 44]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0037_user_0016_scene_0002_cfg_0004_036422060909#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the first block on the second block.'\nStarting at [0.45, 0.37], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.37], [0.44, 0.37], [0.44, 0.37], [0.44, 0.34], [0.44, 0.34], [0.44, 0.32], [0.44, 0.33], [0.45, 0.35], [0.46, 0.34], [0.46, 0.33]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0037_user_0016_scene_0002_cfg_0004_036422060909#81.jpg", + "gt": "[[285, 132], [282, 134], [280, 133], [280, 123], [279, 121], [280, 116], [280, 120], [291, 126], [294, 123], [293, 120]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "62491_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the switch on the socket, open the box', and now you are performing subtask 'pull the lid on the box.'\nThe trajectory starts at [0.51, 0.36]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.36], [0.45, 0.41], [0.46, 0.5], [0.43, 0.37], [0.35, 0.46], [0.32, 0.29], [0.26, 0.31], [0.29, 0.38], [0.34, 0.39], [0.43, 0.31]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/62491_exterior_image_1_left#105.jpg", + "gt": "[[163, 65], [145, 73], [146, 90], [138, 67], [112, 83], [102, 52], [82, 56], [92, 68], [110, 71], [136, 55]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "88693_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the white lid from the food container and put it on the table', and now you are performing subtask 'remove the white lid from the food container and put it on the table.'\nThe trajectory starts at [0.67, 0.05]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.67, 0.05], [0.65, 0.01], [0.61, 0.28], [0.64, 0.38], [0.67, 0.23], [0.62, 0.31], [0.61, 0.19], [0.65, 0.16], [0.65, 0.4], [0.63, 0.21]]}" + } + ], + "images": "88693_exterior_image_1_left#0", + "gt": "[[215, 9], [208, 2], [194, 51], [205, 69], [215, 41], [198, 56], [194, 35], [208, 28], [207, 72], [201, 38]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "7026_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the first drawer from the right and then put the pen inside it', and now you are performing subtask 'place the pen in the first drawer on the right.'\nBegin at [0.31, 0.73] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.31, 0.73], [0.35, 0.66], [0.38, 0.65], [0.38, 0.67], [0.36, 0.68], [0.32, 0.76], [0.33, 0.8], [0.37, 0.79], [0.38, 0.81], [0.31, 0.74]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/7026_exterior_image_1_left#150.jpg", + "gt": "[[100.0, 131.0], [111.0, 119.0], [120.0, 117.0], [121.0, 121.0], [115.0, 122.0], [104.0, 136.0], [105.0, 144.0], [118.0, 143.0], [120.0, 145.0], [98.0, 134.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0094_user_0014_scene_0003_cfg_0004_038522062288#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'push the lid on the right side of the box, push the lid on the left side of the box.'\nYou begin at [0.37, 0.33]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.33], [0.42, 0.28], [0.34, 0.2], [0.32, 0.24], [0.3, 0.24], [0.3, 0.27], [0.24, 0.32], [0.21, 0.36], [0.17, 0.24], [0.2, 0.33]]}" + } + ], + "images": "RH20T_cfg4_task_0094_user_0014_scene_0003_cfg_0004_038522062288#0", + "gt": "[[235, 120], [269, 101], [217, 71], [206, 86], [189, 85], [190, 98], [152, 114], [137, 130], [109, 88], [130, 119]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "45177_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the marker from the cup and put it on the counter', and now you are performing subtask 'remove the marker from the cup and put it on the counter.'\nStart at [0.36, 0.62]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.62], [0.43, 0.67], [0.45, 0.71], [0.44, 0.75], [0.44, 0.7], [0.48, 0.72], [0.46, 0.69], [0.38, 0.61], [0.35, 0.66], [0.39, 0.67]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/45177_exterior_image_1_left#0.jpg", + "gt": "[[114.5, 112.5], [136.5, 120.5], [143.5, 128.5], [139.5, 134.5], [140.5, 125.5], [152.5, 129.5], [146.5, 123.5], [121.5, 110.5], [111.5, 118.5], [125.5, 120.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0069_user_0015_scene_0001_cfg_0004_104122062823#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'assemble one piece of a puzzle', and now you are performing subtask 'pick up the puzzle piece from the table.'\nFrom start point [0.61, 0.22], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.61, 0.22], [0.63, 0.22], [0.63, 0.21], [0.63, 0.21], [0.63, 0.22], [0.62, 0.22], [0.62, 0.2], [0.61, 0.19], [0.6, 0.19], [0.61, 0.2]]}" + } + ], + "images": "RH20T_cfg4_task_0069_user_0015_scene_0001_cfg_0004_104122062823#423", + "gt": "[[392, 80], [403, 80], [405, 77], [403, 75], [403, 79], [400, 78], [398, 73], [388, 68], [386, 70], [388, 72]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_750612070851#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'putting a portion of the Lego blocks on the Lego baseplate', and now you are performing subtask 'insert the yellow Lego brick into the Lego base, pick up the orange Lego brick from the table, transfer the orange Lego brick from the table to the Lego base, insert the orange Lego brick into the Lego base.'\nYou begin at [0.47, 0.59]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.59], [0.44, 0.59], [0.41, 0.67], [0.41, 0.85], [0.45, 0.65], [0.53, 0.58], [0.53, 0.47], [0.55, 0.48], [0.53, 0.53], [0.53, 0.53]]}" + } + ], + "images": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_750612070851#371", + "gt": "[[301, 214], [281, 213], [260, 241], [262, 307], [289, 235], [338, 208], [340, 169], [349, 174], [341, 192], [341, 192]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "40510_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the glue stick from the cup, place it on the box, then on the table, back into the cup, and lastly on the table', and now you are performing subtask 'place glue stick on the table.'\nThe trajectory starts at [0.68, 0.52]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.52], [0.68, 0.47], [0.68, 0.49], [0.68, 0.48], [0.67, 0.54], [0.69, 0.53], [0.7, 0.53], [0.72, 0.57], [0.74, 0.62], [0.76, 0.73]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/40510_exterior_image_2_left#119.jpg", + "gt": "[[216, 94], [218, 85], [219, 88], [218, 87], [213, 97], [221, 95], [224, 96], [230, 102], [237, 111], [244, 131]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0029_user_0010_scene_0006_cfg_0005_036422060215#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp the handle and open the drawer', and now you are performing subtask 'pull the handle to open the drawer.'\nWith start location [0.48, 0.35], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.35], [0.47, 0.36], [0.48, 0.39], [0.51, 0.38], [0.51, 0.39], [0.53, 0.4], [0.53, 0.29], [0.5, 0.23], [0.45, 0.23], [0.45, 0.22]]}" + } + ], + "images": "RH20T_cfg5_task_0029_user_0010_scene_0006_cfg_0005_036422060215#271", + "gt": "[[306, 125], [302, 131], [305, 139], [326, 135], [325, 140], [341, 145], [342, 104], [321, 83], [287, 83], [286, 79]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "28024_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'swap the places of the cup and the cutlery', and now you are performing subtask 'place the cutlery on the other side of the table.'\nYour initial coordinate is [0.44, 0.69]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.69], [0.44, 0.64], [0.57, 0.74], [0.61, 0.87], [0.59, 0.97], [0.57, 0.78], [0.43, 0.64], [0.38, 0.66], [0.42, 0.71], [0.42, 0.67]]}" + } + ], + "images": "28024_exterior_image_1_left#401", + "gt": "[[140, 124], [140, 116], [183, 134], [194, 156], [188, 174], [184, 140], [138, 115], [120, 119], [133, 128], [135, 120]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "54621_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'rotate the faucet counterclockwise, press the button on the faucet', and now you are performing subtask 'press the button on the faucet.'\nYou start from [0.42, 0.45]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.45], [0.32, 0.58], [0.34, 0.62], [0.39, 0.62], [0.37, 0.49], [0.41, 0.36], [0.52, 0.34], [0.6, 0.38], [0.58, 0.4], [0.58, 0.4]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/54621_exterior_image_1_left#236.jpg", + "gt": "[[134.0, 80.5], [104.0, 104.5], [108.0, 111.5], [125.0, 111.5], [118.0, 87.5], [132.0, 64.5], [167.0, 60.5], [193.0, 67.5], [187.0, 72.5], [187.0, 72.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the first block from the desktop, transfer the first block from the desktop to the second position above the first block, place the first block on top of the second position.'\nFrom start point [0.63, 0.37], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.66, 0.38], [0.69, 0.48], [0.73, 0.51], [0.73, 0.54], [0.75, 0.56], [0.72, 0.57], [0.72, 0.41], [0.7, 0.39], [0.69, 0.39]]}" + } + ], + "images": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_038522062547#0", + "gt": "[[405, 133], [423, 136], [440, 172], [467, 183], [469, 195], [477, 200], [461, 207], [458, 147], [451, 139], [442, 140]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "24957_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the purple object in the bowl', and now you are performing subtask 'put the purple object in the bowl.'\nYou start from [0.62, 0.0]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.0], [0.61, 0.1], [0.57, 0.21], [0.58, 0.31], [0.54, 0.14], [0.52, 0.12], [0.52, 0.17], [0.52, 0.24], [0.54, 0.29], [0.53, 0.2]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/24957_exterior_image_2_left#0.jpg", + "gt": "[[199, 0], [195, 18], [183, 38], [186, 56], [174, 25], [168, 22], [165, 31], [166, 43], [172, 53], [170, 36]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "17916_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push the toaster lever down', and now you are performing subtask 'push the toaster lever down.'\nYou begin at [0.33, 0.26]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.33, 0.26], [0.41, 0.24], [0.54, 0.31], [0.63, 0.42], [0.63, 0.44], [0.65, 0.45], [0.67, 0.56], [0.66, 0.52], [0.61, 0.41], [0.52, 0.32]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/17916_exterior_image_1_left#0.jpg", + "gt": "[[107, 47], [132, 43], [173, 56], [203, 75], [201, 79], [209, 81], [213, 101], [212, 94], [195, 74], [166, 57]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0090_user_0007_scene_0001_cfg_0001_819112071037#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the wooden block', and now you are performing subtask 'press the wooden block.'\nWith start location [0.52, 0.0], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.0], [0.39, 0.0], [0.3, 0.12], [0.27, 0.13], [0.26, 0.1], [0.25, 0.15], [0.16, 0.23], [0.11, 0.33], [0.1, 0.29], [0.28, 0.19]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0090_user_0007_scene_0001_cfg_0001_819112071037#0.jpg", + "gt": "[[331, 0], [250, 0], [190, 42], [173, 47], [168, 36], [162, 55], [102, 82], [73, 119], [65, 103], [177, 67]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "88032_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'switch on the adapter', and now you are performing subtask 'switch on the adapter.'\nYour initial coordinate is [0.66, 0.16]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.66, 0.16], [0.71, 0.16], [0.76, 0.31], [0.79, 0.42], [0.79, 0.48], [0.76, 0.49], [0.75, 0.54], [0.77, 0.58], [0.75, 0.47], [0.77, 0.31]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/88032_exterior_image_1_left#0.jpg", + "gt": "[[211, 29], [228, 28], [244, 56], [252, 76], [252, 86], [244, 89], [241, 98], [246, 105], [241, 84], [246, 56]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "81300_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the pen out of the bowl and put it on the table', and now you are performing subtask 'take the pen out of the bowl and put it on the table.'\nThe trajectory starts at [0.68, 0.0]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.0], [0.61, 0.11], [0.56, 0.28], [0.55, 0.12], [0.54, 0.09], [0.52, 0.12], [0.49, 0.3], [0.5, 0.46], [0.53, 0.56], [0.52, 0.48]]}" + } + ], + "images": "81300_exterior_image_2_left#0", + "gt": "[[219, 0], [196, 20], [180, 51], [177, 21], [172, 16], [166, 22], [156, 54], [159, 82], [169, 100], [165, 87]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0096_user_0015_scene_0005_cfg_0001_038522063145#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the block out of the drawer and put it on the table', and now you are performing subtask 'pull the drawer handle to open the drawer, pick up the block from the drawer, transfer the block from the drawer to the tabletop, place the block on the tabletop.'\nWith start location [0.7, 0.1], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.1], [0.69, 0.08], [0.68, 0.11], [0.67, 0.19], [0.7, 0.04], [0.75, 0.08], [0.78, 0.14], [0.8, 0.29], [0.78, 0.41], [0.79, 0.34]]}" + } + ], + "images": "RH20T_cfg1_task_0096_user_0015_scene_0005_cfg_0001_038522063145#236", + "gt": "[[446, 37], [444, 28], [436, 38], [429, 68], [449, 16], [479, 28], [500, 49], [514, 106], [499, 147], [504, 124]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "47471_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the marker from the cup and put it on the book', and now you are performing subtask 'pick up the marker from the cup and put it on the book.'\nFrom start point [0.47, 0.24], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.24], [0.51, 0.34], [0.59, 0.41], [0.6, 0.49], [0.55, 0.56], [0.54, 0.44], [0.62, 0.44], [0.6, 0.58], [0.58, 0.6], [0.57, 0.56]]}" + } + ], + "images": "47471_exterior_image_2_left#0", + "gt": "[[149, 44], [162, 61], [188, 74], [193, 89], [177, 100], [173, 80], [198, 80], [192, 105], [185, 108], [181, 100]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_819112070044#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn them on', and now you are performing subtask 'press the second button on the right side of the power strip.'\nBegin at [0.76, 0.53] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.76, 0.53], [0.75, 0.53], [0.75, 0.55], [0.75, 0.54], [0.75, 0.55], [0.76, 0.52], [0.77, 0.52], [0.76, 0.52], [0.76, 0.52], [0.76, 0.52]]}" + } + ], + "images": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_819112070044#177", + "gt": "[[484, 192], [482, 190], [480, 197], [483, 195], [481, 199], [485, 189], [490, 186], [488, 187], [487, 186], [487, 186]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0083_user_0015_scene_0004_cfg_0004_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'swat with a flyswatter', and now you are performing subtask 'pick up the flyswatter from the surface.'\nYour initial coordinate is [0.45, 0.0]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.0], [0.52, 0.0], [0.48, 0.0], [0.51, 0.0], [0.54, 0.13], [0.52, 0.17], [0.53, 0.17], [0.53, 0.2], [0.53, 0.21], [0.55, 0.18]]}" + } + ], + "images": "RH20T_cfg4_task_0083_user_0015_scene_0004_cfg_0004_104122062823#0", + "gt": "[[290, 0], [333, 0], [306, 0], [326, 0], [344, 46], [335, 62], [339, 62], [337, 72], [342, 75], [349, 65]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "43885_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the white towel', and now you are performing subtask 'fold the white towel on the table.'\nBegin at [0.72, 0.08] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.72, 0.08], [0.76, 0.23], [0.71, 0.41], [0.72, 0.49], [0.68, 0.55], [0.68, 0.49], [0.73, 0.4], [0.6, 0.4], [0.52, 0.5], [0.52, 0.56]]}" + } + ], + "images": "43885_exterior_image_2_left#0", + "gt": "[[229, 14], [242, 42], [228, 73], [229, 89], [216, 99], [217, 89], [234, 72], [192, 72], [166, 90], [167, 101]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0063_user_0007_scene_0007_cfg_0005_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use a shovel to scoop up corn', and now you are performing subtask 'use a shovel to scoop up corn.'\nYou begin at [0.45, 0.3]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.3], [0.46, 0.29], [0.47, 0.34], [0.43, 0.41], [0.41, 0.4], [0.42, 0.32], [0.4, 0.32], [0.39, 0.26], [0.4, 0.42], [0.39, 0.32]]}" + } + ], + "images": "RH20T_cfg5_task_0063_user_0007_scene_0007_cfg_0005_036422060215#0", + "gt": "[[286, 108], [297, 103], [304, 122], [276, 147], [263, 143], [271, 116], [259, 117], [249, 95], [257, 150], [251, 117]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0008_user_0016_scene_0005_cfg_0003_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grab the block and place it at the designated location', and now you are performing subtask 'push the block from right to left, pick up the block from the right side of the cardboard.'\nStarting at [0.65, 0.32], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.32], [0.64, 0.32], [0.64, 0.29], [0.62, 0.28], [0.61, 0.28], [0.58, 0.26], [0.59, 0.25], [0.59, 0.21], [0.6, 0.19], [0.6, 0.19]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg3_task_0008_user_0016_scene_0005_cfg_0003_104122062823#0.jpg", + "gt": "[[419, 115], [411, 115], [407, 103], [396, 99], [392, 101], [369, 93], [377, 91], [379, 75], [383, 67], [383, 67]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "63555_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the lid and close the pot', and now you are performing subtask 'pick up the lid and close the pot.'\nStarting at [0.81, 0.01], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.81, 0.01], [0.73, 0.23], [0.69, 0.27], [0.65, 0.39], [0.66, 0.32], [0.66, 0.4], [0.62, 0.31], [0.61, 0.4], [0.63, 0.41], [0.63, 0.33]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/63555_exterior_image_2_left#0.jpg", + "gt": "[[259, 1], [235, 41], [221, 49], [209, 71], [210, 58], [212, 72], [200, 55], [195, 72], [203, 73], [202, 60]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "74907_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the right yellow object out of the case and put it on the box', and now you are performing subtask 'take the right yellow object out of the case and put it on the box.'\nYou start from [0.49, 0.0]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.0], [0.58, 0.17], [0.57, 0.29], [0.58, 0.22], [0.56, 0.16], [0.47, 0.11], [0.38, 0.14], [0.38, 0.25], [0.42, 0.29], [0.41, 0.11]]}" + } + ], + "images": "74907_exterior_image_2_left#0", + "gt": "[[158, 0], [185, 30], [183, 53], [186, 40], [180, 29], [151, 19], [120, 26], [123, 45], [134, 53], [131, 19]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "63709_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the blue marker from the blue bowl and place it on the table', and now you are performing subtask 'remove the blue marker from the blue bowl and place it on the table.'\nYour starting position is [0.55, 0.63]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.63], [0.49, 0.63], [0.51, 0.64], [0.49, 0.72], [0.5, 0.71], [0.49, 0.68], [0.51, 0.59], [0.53, 0.56], [0.57, 0.56], [0.54, 0.52]]}" + } + ], + "images": "63709_exterior_image_1_left#0", + "gt": "[[177, 113], [158, 113], [162, 115], [158, 130], [160, 127], [157, 122], [163, 106], [169, 101], [181, 101], [173, 93]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg3_task_0058_user_0010_scene_0006_cfg_0003_104122062823#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn off the power strips', and now you are performing subtask 'press the third button of the power strip, press the fourth button of the power strip.'\nYou start from [0.2, 0.14]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.2, 0.14], [0.18, 0.1], [0.2, 0.11], [0.19, 0.12], [0.18, 0.08], [0.19, 0.04], [0.28, 0.0], [0.29, 0.0], [0.28, 0.03], [0.28, 0.02]]}" + } + ], + "images": "RH20T_cfg3_task_0058_user_0010_scene_0006_cfg_0003_104122062823#221", + "gt": "[[129, 50], [113, 36], [129, 41], [123, 44], [115, 27], [121, 15], [176, 0], [187, 0], [182, 9], [177, 8]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0006_user_0016_scene_0003_cfg_0003_036422060909#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the gripper to push the small block from left to right', and now you are performing subtask 'use the gripper to push the small block from left to right.'\nWith start location [0.53, 0.24], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.24], [0.48, 0.24], [0.5, 0.18], [0.54, 0.16], [0.6, 0.21], [0.56, 0.21], [0.44, 0.41], [0.49, 0.35], [0.5, 0.29], [0.5, 0.29]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg3_task_0006_user_0016_scene_0003_cfg_0003_036422060909#0.jpg", + "gt": "[[336, 88], [309, 87], [323, 65], [343, 58], [384, 77], [359, 75], [281, 147], [316, 127], [318, 104], [318, 104]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0013_user_0001_scene_0005_cfg_0001_039422060546#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the block on the scale', and now you are performing subtask 'place the block on the scale.'\nYour starting position is [0.4, 0.35]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.35], [0.43, 0.36], [0.52, 0.28], [0.61, 0.33], [0.6, 0.27], [0.59, 0.32], [0.59, 0.27], [0.49, 0.46], [0.51, 0.58], [0.49, 0.42]]}" + } + ], + "images": "RH20T_cfg1_task_0013_user_0001_scene_0005_cfg_0001_039422060546#0", + "gt": "[[255, 127], [275, 131], [332, 101], [393, 118], [385, 96], [380, 114], [375, 98], [314, 164], [328, 208], [316, 152]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "9787_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the pencil to the left side of the table', and now you are performing subtask 'move the pencil to the left side of the table.'\nYour starting position is [0.48, 0.01]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.01], [0.51, 0.06], [0.54, 0.17], [0.57, 0.49], [0.54, 0.39], [0.51, 0.36], [0.47, 0.4], [0.47, 0.44], [0.47, 0.35], [0.47, 0.37]]}" + } + ], + "images": "9787_exterior_image_1_left#0", + "gt": "[[155.0, 1.5], [163.0, 10.5], [173.0, 31.5], [183.0, 88.5], [172.0, 69.5], [164.0, 65.5], [151.0, 72.5], [152.0, 79.5], [150.0, 63.5], [152.0, 66.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "29599_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the towel on the table, pull the towel to the right side of the table', and now you are performing subtask 'unfold the towel on the table.'\nStarting at [0.4, 0.03], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.03], [0.55, 0.38], [0.52, 0.48], [0.54, 0.58], [0.48, 0.69], [0.54, 0.67], [0.5, 0.77], [0.59, 0.7], [0.7, 0.78], [0.72, 0.74]]}" + } + ], + "images": "29599_exterior_image_1_left#0", + "gt": "[[129, 6], [175, 69], [166, 87], [172, 105], [155, 124], [174, 120], [161, 139], [189, 126], [223, 141], [231, 134]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0047_user_0014_scene_0002_cfg_0006_104122063633#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the cup off the cup rack', and now you are performing subtask 'take the cup off the cup rack.'\nYour starting position is [0.36, 0.33]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.33], [0.28, 0.19], [0.29, 0.13], [0.32, 0.14], [0.34, 0.24], [0.33, 0.14], [0.32, 0.11], [0.31, 0.13], [0.32, 0.38], [0.35, 0.22]]}" + } + ], + "images": "RH20T_cfg6_task_0047_user_0014_scene_0002_cfg_0006_104122063633#0", + "gt": "[[233, 118], [177, 69], [188, 46], [203, 50], [217, 85], [212, 52], [206, 41], [201, 46], [208, 135], [221, 81]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0074_user_0015_scene_0006_cfg_0004_104122062295#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the button horizontally', and now you are performing subtask 'press the button horizontally.'\nYour starting position is [0.41, 0.17]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.17], [0.32, 0.19], [0.26, 0.31], [0.21, 0.24], [0.15, 0.26], [0.14, 0.21], [0.19, 0.24], [0.22, 0.22], [0.22, 0.27], [0.36, 0.46]]}" + } + ], + "images": "RH20T_cfg4_task_0074_user_0015_scene_0006_cfg_0004_104122062295#0", + "gt": "[[262, 63], [202, 70], [165, 113], [132, 86], [98, 93], [87, 77], [119, 87], [138, 80], [144, 96], [233, 165]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "18719_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the lid on the jar', and now you are performing subtask 'place the lid on the jar.'\nYour starting position is [0.99, 0.28]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.99, 0.28], [0.84, 0.58], [0.68, 0.41], [0.59, 0.43], [0.47, 0.72], [0.48, 0.34], [0.53, 0.32], [0.47, 0.31], [0.53, 0.24], [0.53, 0.24]]}" + } + ], + "images": "18719_exterior_image_1_left#0", + "gt": "[[318, 50], [268, 104], [216, 74], [189, 78], [152, 129], [153, 62], [170, 57], [151, 55], [171, 43], [171, 43]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0066_user_0016_scene_0003_cfg_0004_036422060909#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'building blocks', and now you are performing subtask 'pick up the green block from the table, transfer the green block from the table to the block board, place the green block on the block board, pick up the red block from the table.'\nThe trajectory starts at [0.53, 0.22]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.22], [0.48, 0.19], [0.51, 0.18], [0.51, 0.12], [0.49, 0.22], [0.54, 0.14], [0.57, 0.19], [0.56, 0.24], [0.55, 0.18], [0.51, 0.16]]}" + } + ], + "images": "RH20T_cfg4_task_0066_user_0016_scene_0003_cfg_0004_036422060909#0", + "gt": "[[341, 79], [308, 69], [327, 66], [326, 44], [315, 79], [343, 51], [368, 68], [361, 86], [352, 66], [326, 59]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0020_user_0016_scene_0002_cfg_0004_104122062823#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the desk lamp by pressing the button', and now you are performing subtask 'press the button on the desk lamp to turn on the light.'\nFrom start point [0.25, 0.18], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.25, 0.18], [0.25, 0.21], [0.24, 0.22], [0.24, 0.21], [0.25, 0.19], [0.24, 0.21], [0.25, 0.23], [0.26, 0.19], [0.33, 0.16], [0.36, 0.16]]}" + } + ], + "images": "RH20T_cfg4_task_0020_user_0016_scene_0002_cfg_0004_104122062823#142", + "gt": "[[160, 66], [162, 75], [155, 81], [153, 77], [162, 69], [154, 77], [159, 82], [168, 68], [213, 56], [229, 58]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "47660_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the blue cutlery box forwards, take out a butter knife and put it on the table', and now you are performing subtask 'pull the blue cutlery box.'\nBegin at [0.74, 0.43] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.74, 0.43], [0.68, 0.4], [0.61, 0.42], [0.62, 0.44], [0.61, 0.45], [0.62, 0.54], [0.6, 0.55], [0.6, 0.58], [0.62, 0.59], [0.6, 0.59]]}" + } + ], + "images": "47660_exterior_image_1_left#0", + "gt": "[[238, 78], [216, 72], [195, 76], [198, 79], [195, 81], [197, 98], [192, 99], [192, 104], [198, 107], [193, 106]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "69880_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the masking tape from the rack', and now you are performing subtask 'remove the masking tape from the rack.'\nYour initial coordinate is [0.45, 0.43]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.43], [0.67, 0.39], [0.58, 0.43], [0.66, 0.46], [0.63, 0.42], [0.62, 0.46], [0.62, 0.71], [0.65, 0.7], [0.63, 0.74], [0.6, 0.42]]}" + } + ], + "images": "69880_exterior_image_2_left#0", + "gt": "[[145, 78], [215, 70], [185, 78], [210, 83], [202, 76], [198, 82], [198, 127], [209, 126], [203, 134], [193, 75]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "30828_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the bottle from the table and place it inside the clear container', and now you are performing subtask 'pick up the bottle from the table and place it inside the clear container.'\nYour starting position is [0.64, 0.01]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.64, 0.01], [0.61, 0.16], [0.55, 0.23], [0.57, 0.24], [0.51, 0.34], [0.47, 0.52], [0.47, 0.57], [0.47, 0.47], [0.49, 0.53], [0.47, 0.45]]}" + } + ], + "images": "30828_exterior_image_1_left#0", + "gt": "[[205, 1], [195, 29], [176, 42], [183, 44], [162, 62], [150, 93], [152, 103], [149, 85], [157, 95], [149, 81]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "777_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the clear plastic from the box and place it on the left side of the counter, remove the packaging from the clear plastic', and now you are performing subtask 'place the object on the tabletop.'\nStarting at [0.34, 0.83], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.34, 0.83], [0.3, 0.83], [0.32, 0.87], [0.41, 0.89], [0.38, 0.7], [0.32, 0.71], [0.34, 0.76], [0.43, 0.71], [0.37, 0.71], [0.37, 0.71]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/777_exterior_image_1_left#292.jpg", + "gt": "[[110, 149], [97, 150], [103, 157], [131, 160], [123, 126], [103, 128], [109, 136], [139, 128], [119, 128], [119, 128]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "2836_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unhang the towel from the object and put it on the table', and now you are performing subtask 'unhang the towel from the object and put it on the table.'\nWith start location [0.88, 0.02], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.88, 0.02], [0.69, 0.44], [0.71, 0.21], [0.73, 0.01], [0.73, 0.12], [0.73, 0.37], [0.61, 0.43], [0.64, 0.38], [0.69, 0.36], [0.68, 0.39]]}" + } + ], + "images": "2836_exterior_image_2_left#0", + "gt": "[[281, 4], [221, 79], [226, 37], [235, 2], [232, 22], [235, 66], [196, 78], [206, 69], [221, 65], [216, 71]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "42894_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull the white bowl to the left, take out the fork and the spoon from the bowl, then place them on the table, move the white cup to the left', and now you are performing subtask 'pick up the spoon from the white bowl.'\nYour starting position is [0.38, 0.5]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.5], [0.39, 0.66], [0.41, 0.61], [0.4, 0.66], [0.42, 0.58], [0.36, 0.64], [0.42, 0.69], [0.42, 0.77], [0.43, 0.64], [0.41, 0.65]]}" + } + ], + "images": "42894_exterior_image_1_left#141", + "gt": "[[121, 90], [125, 118], [132, 110], [129, 119], [133, 105], [116, 115], [135, 125], [134, 139], [138, 115], [130, 117]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0061_user_0007_scene_0003_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'pick up the small cube inside the gift box, transfer the small cube from inside the gift box to the desktop, place the small cube on the desktop.'\nYour starting position is [0.5, 0.37]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.5, 0.34], [0.48, 0.32], [0.47, 0.42], [0.46, 0.46], [0.44, 0.36], [0.44, 0.36], [0.43, 0.37], [0.43, 0.31], [0.44, 0.3]]}" + } + ], + "images": "RH20T_cfg5_task_0061_user_0007_scene_0003_cfg_0005_104122063678#0", + "gt": "[[323, 132], [318, 122], [306, 114], [304, 151], [295, 167], [283, 131], [279, 128], [275, 134], [276, 113], [283, 109]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "77779_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the glue stick from the green cup and put it on the black box', and now you are performing subtask 'take the glue stick from the green cup and put it on the black box.'\nBegin at [0.98, 0.31] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.98, 0.31], [0.78, 0.29], [0.8, 0.31], [0.68, 0.36], [0.61, 0.44], [0.59, 0.62], [0.64, 0.38], [0.69, 0.4], [0.69, 0.56], [0.71, 0.45]]}" + } + ], + "images": "77779_exterior_image_1_left#0", + "gt": "[[314.0, 55.5], [250.0, 52.5], [256.0, 56.5], [216.0, 65.5], [195.0, 79.5], [188.0, 111.5], [205.0, 67.5], [221.0, 71.5], [222.0, 100.5], [227.0, 81.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0103_user_0010_scene_0005_cfg_0001_038522063145#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the plate to the goal post', and now you are performing subtask 'move the plate to the goal post.'\nWith start location [0.69, 0.28], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.28], [0.7, 0.51], [0.71, 0.48], [0.7, 0.46], [0.57, 0.24], [0.53, 0.24], [0.6, 0.22], [0.67, 0.27], [0.69, 0.25], [0.69, 0.25]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0103_user_0010_scene_0005_cfg_0001_038522063145#0.jpg", + "gt": "[[440, 99], [445, 182], [453, 173], [448, 164], [368, 88], [340, 85], [384, 80], [430, 98], [444, 90], [444, 90]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0064_user_0016_scene_0003_cfg_0004_036422060909#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the roll of paper down from the shelf', and now you are performing subtask 'pick up the roll of paper from the shelf.'\nWith start location [0.53, 0.22], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.22], [0.51, 0.22], [0.56, 0.16], [0.57, 0.21], [0.62, 0.15], [0.6, 0.12], [0.57, 0.14], [0.57, 0.21], [0.53, 0.23], [0.53, 0.23]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0064_user_0016_scene_0003_cfg_0004_036422060909#0.jpg", + "gt": "[[341, 79], [329, 78], [356, 59], [364, 77], [396, 55], [385, 42], [367, 52], [364, 74], [339, 83], [339, 83]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0038_user_0014_scene_0006_cfg_0006_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the cup', and now you are performing subtask 'pick up the cup.'\nBegin at [0.53, 0.31] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.31], [0.52, 0.25], [0.52, 0.26], [0.53, 0.25], [0.55, 0.27], [0.56, 0.3], [0.58, 0.42], [0.58, 0.49], [0.57, 0.39], [0.57, 0.28]]}" + } + ], + "images": "RH20T_cfg6_task_0038_user_0014_scene_0006_cfg_0006_104122064161#0", + "gt": "[[338, 110], [336, 89], [336, 92], [340, 91], [349, 96], [356, 107], [374, 152], [373, 177], [368, 139], [367, 99]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "52922_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put some detergent in the detergent compartment', and now you are performing subtask 'place the bottle on the top of the device.'\nYour starting position is [0.4, 0.44]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.44], [0.51, 0.33], [0.47, 0.26], [0.48, 0.17], [0.45, 0.17], [0.46, 0.01], [0.58, 0.11], [0.64, 0.43], [0.57, 0.08], [0.52, 0.02]]}" + } + ], + "images": "52922_exterior_image_1_left#0", + "gt": "[[129.0, 79.0], [163.0, 59.0], [149.0, 47.0], [154.0, 30.0], [144.0, 30.0], [147.0, 2.0], [187.0, 20.0], [206.0, 78.0], [182.0, 15.0], [168.0, 4.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_750612070851#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn them on', and now you are performing subtask 'press the fourth button on the right side of the power strip.'\nYou begin at [0.25, 0.63]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.25, 0.63], [0.26, 0.64], [0.26, 0.66], [0.27, 0.65], [0.27, 0.66], [0.27, 0.63], [0.26, 0.62], [0.25, 0.61], [0.36, 0.71], [0.38, 0.7]]}" + } + ], + "images": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_750612070851#273", + "gt": "[[162, 228], [169, 232], [169, 237], [172, 233], [174, 237], [172, 227], [168, 223], [157, 220], [232, 257], [245, 253]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "69300_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the glue stick inside the open drawer', and now you are performing subtask 'put the glue stick inside the open drawer.'\nYou begin at [0.47, 0.35]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.35], [0.4, 0.31], [0.39, 0.33], [0.4, 0.32], [0.45, 0.38], [0.41, 0.38], [0.4, 0.45], [0.42, 0.44], [0.45, 0.51], [0.4, 0.47]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/69300_exterior_image_1_left#0.jpg", + "gt": "[[150.5, 63.0], [128.5, 56.0], [125.5, 60.0], [129.5, 57.0], [144.5, 68.0], [132.5, 69.0], [127.5, 81.0], [133.5, 80.0], [144.5, 92.0], [126.5, 84.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "38414_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the plush toy and put it on the plate, then move the tap to the right', and now you are performing subtask 'twist the faucet counterclockwise.'\nStart at [0.17, 0.01]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.17, 0.01], [0.29, 0.28], [0.32, 0.42], [0.36, 0.5], [0.39, 0.48], [0.42, 0.5], [0.42, 0.52], [0.38, 0.51], [0.39, 0.48], [0.5, 0.34]]}" + } + ], + "images": "38414_exterior_image_1_left#0", + "gt": "[[55.5, 2.0], [92.5, 51.0], [103.5, 75.0], [115.5, 90.0], [124.5, 86.0], [133.5, 90.0], [133.5, 94.0], [122.5, 92.0], [125.5, 87.0], [160.5, 62.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_035622060973#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn them on', and now you are performing subtask 'press the second button on the right side of the power strip, press the third button on the right side of the power strip, press the fourth button on the right side of the power strip.'\nYou begin at [0.35, 0.22]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.22], [0.34, 0.26], [0.33, 0.26], [0.35, 0.27], [0.35, 0.28], [0.35, 0.28], [0.35, 0.24], [0.34, 0.28], [0.34, 0.3], [0.34, 0.3]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_035622060973#216.jpg", + "gt": "[[225, 79], [218, 92], [214, 94], [222, 97], [223, 99], [222, 99], [222, 87], [215, 102], [215, 108], [218, 109]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "36386_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the second last switch of the adapter', and now you are performing subtask 'turn on the second last switch of the adapter.'\nWith start location [0.15, 0.0], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.15, 0.0], [0.29, 0.03], [0.44, 0.14], [0.43, 0.26], [0.48, 0.29], [0.46, 0.36], [0.47, 0.43], [0.44, 0.45], [0.43, 0.39], [0.45, 0.45]]}" + } + ], + "images": "36386_exterior_image_2_left#0", + "gt": "[[49, 0], [94, 5], [142, 26], [139, 47], [155, 52], [148, 65], [150, 78], [140, 81], [139, 70], [144, 81]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "76361_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the lid on the counter', and now you are performing subtask 'put the lid on the counter.'\nYou begin at [0.79, 0.76]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.79, 0.76], [0.57, 0.54], [0.51, 0.61], [0.6, 0.59], [0.7, 0.75], [0.75, 0.96], [0.75, 0.82], [0.78, 0.99], [0.78, 0.84], [0.66, 0.72]]}" + } + ], + "images": "76361_exterior_image_2_left#0", + "gt": "[[254, 136], [183, 97], [162, 109], [192, 106], [225, 135], [240, 173], [240, 147], [250, 178], [250, 152], [211, 129]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122061330#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'pick up the lid on the container, transfer the lid from the container to the table, place the lid on the table.'\nYou begin at [0.42, 0.29]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.29], [0.42, 0.28], [0.43, 0.29], [0.52, 0.32], [0.54, 0.37], [0.54, 0.19], [0.57, 0.12], [0.58, 0.15], [0.58, 0.19], [0.58, 0.19]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122061330#0.jpg", + "gt": "[[266, 103], [266, 99], [276, 106], [332, 114], [347, 132], [343, 68], [364, 44], [370, 53], [369, 68], [369, 68]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "37064_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the marker and put it above the table', and now you are performing subtask 'pick up the marker and put it above the table.'\nStarting at [0.99, 0.29], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.99, 0.29], [0.7, 0.72], [0.66, 0.63], [0.62, 0.57], [0.51, 0.48], [0.41, 0.45], [0.38, 0.48], [0.38, 0.55], [0.39, 0.48], [0.41, 0.46]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/37064_exterior_image_2_left#0.jpg", + "gt": "[[318, 52], [224, 130], [211, 113], [198, 103], [162, 87], [131, 81], [123, 86], [122, 99], [124, 86], [130, 83]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0130_user_0014_scene_0004_cfg_0006_037522061512#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb', and now you are performing subtask 'pick up the lamp holder and the light bulb from the table, transfer the lamp holder from the table to the socket, insert the lamp holder into the socket.'\nStart at [0.4, 0.23]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.23], [0.38, 0.31], [0.31, 0.39], [0.31, 0.42], [0.32, 0.44], [0.36, 0.35], [0.39, 0.34], [0.39, 0.37], [0.37, 0.34], [0.39, 0.31]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0130_user_0014_scene_0004_cfg_0006_037522061512#0.jpg", + "gt": "[[256, 84], [240, 112], [200, 141], [199, 152], [207, 157], [233, 126], [247, 121], [247, 133], [237, 121], [248, 110]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0003_user_0010_scene_0001_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the yellow button on the right, press the red button in the middle.'\nWith start location [0.5, 0.37], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.52, 0.36], [0.53, 0.54], [0.55, 0.54], [0.55, 0.56], [0.56, 0.59], [0.56, 0.56], [0.55, 0.67], [0.57, 0.61], [0.53, 0.53]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0003_user_0010_scene_0001_cfg_0005_104122063678#0.jpg", + "gt": "[[323, 132], [334, 129], [340, 196], [355, 193], [351, 203], [358, 211], [360, 203], [353, 240], [366, 219], [342, 192]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "47649_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the candy bar from the sink', and now you are performing subtask 'remove the candy bar from the sink.'\nWith start location [0.37, 0.48], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.48], [0.39, 0.67], [0.48, 0.88], [0.47, 0.78], [0.39, 0.71], [0.32, 0.69], [0.27, 0.73], [0.3, 0.7], [0.38, 0.72], [0.38, 0.72]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/47649_exterior_image_1_left#0.jpg", + "gt": "[[118, 86], [125, 120], [155, 158], [150, 140], [124, 127], [101, 125], [86, 132], [96, 126], [121, 129], [121, 129]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "91403_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the banana and the pear plush toys from the box to the plate then move the box to the right', and now you are performing subtask 'place the banana plush toy on the plate.'\nStarting at [0.64, 0.29], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.64, 0.29], [0.42, 0.37], [0.36, 0.57], [0.32, 0.56], [0.36, 0.53], [0.38, 0.61], [0.34, 0.52], [0.28, 0.54], [0.25, 0.59], [0.37, 0.51]]}" + } + ], + "images": "91403_exterior_image_1_left#0", + "gt": "[[204, 53], [135, 66], [115, 102], [104, 100], [114, 96], [122, 110], [109, 94], [89, 97], [81, 107], [117, 91]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "7338_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the door', and now you are performing subtask 'close the door.'\nThe trajectory starts at [0.01, 0.62]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.01, 0.62], [0.08, 0.55], [0.09, 0.52], [0.07, 0.53], [0.08, 0.52], [0.18, 0.46], [0.25, 0.47], [0.49, 0.46], [0.22, 0.57], [0.22, 0.57]]}" + } + ], + "images": "7338_exterior_image_2_left#0", + "gt": "[[4, 111], [25, 99], [29, 94], [23, 96], [25, 93], [59, 82], [80, 85], [156, 82], [71, 102], [71, 102]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0010_user_0001_scene_0005_cfg_0001_750612070851#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the handset of the telephone on the corresponding phone cradle', and now you are performing subtask 'place the handset of the telephone on the corresponding phone cradle.'\nStart at [0.42, 0.17]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.17], [0.44, 0.33], [0.52, 0.57], [0.53, 0.87], [0.56, 0.88], [0.55, 0.64], [0.52, 0.56], [0.53, 0.48], [0.51, 0.48], [0.55, 0.5]]}" + } + ], + "images": "RH20T_cfg1_task_0010_user_0001_scene_0005_cfg_0001_750612070851#0", + "gt": "[[272, 61], [281, 119], [335, 205], [342, 313], [358, 315], [350, 229], [331, 203], [337, 174], [325, 173], [350, 181]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0065_user_0007_scene_0012_cfg_0002_104122063678#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with small Lego blocks', and now you are performing subtask 'press the black block on the block base, pick up the red block on the table, transfer the red block from the table to the block base, place the red block on the block base, press the red block, push the black block to adjust its position, push the black block to adjust its position, press the black block.'\nWith start location [0.72, 0.68], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.72, 0.68], [0.8, 0.72], [0.79, 0.76], [0.72, 0.7], [0.72, 0.73], [0.69, 0.58], [0.71, 0.69], [0.73, 0.68], [0.7, 0.68], [0.72, 0.62]]}" + } + ], + "images": "RH20T_cfg2_task_0065_user_0007_scene_0012_cfg_0002_104122063678#291", + "gt": "[[463, 243], [511, 260], [507, 272], [458, 251], [458, 264], [443, 209], [453, 249], [468, 244], [450, 246], [461, 225]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "51525_exterior_image_2_left#TrajectoryPrediction_Task#10", + "task": "TrajectoryPrediction_Task", + "p_index": 10, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the microwave door, move the pink jar on the counter', and now you are performing subtask 'transfer the pink jar from the microwave to the countertop.'\nFrom start point [0.36, 0.4], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.4], [0.36, 0.39], [0.36, 0.39], [0.36, 0.38], [0.36, 0.35], [0.38, 0.34], [0.38, 0.35], [0.38, 0.37], [0.38, 0.38], [0.38, 0.37]]}" + } + ], + "images": "51525_exterior_image_2_left#268", + "gt": "[[116, 72], [114, 71], [116, 71], [114, 69], [115, 63], [121, 61], [123, 63], [122, 67], [120, 68], [121, 66]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "72531_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull the box under the cabinet', and now you are performing subtask 'pull the box under the cabinet.'\nYou begin at [0.32, 0.58]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.58], [0.44, 0.62], [0.53, 0.59], [0.51, 0.61], [0.54, 0.64], [0.58, 0.57], [0.61, 0.57], [0.54, 0.62], [0.52, 0.54], [0.5, 0.58]]}" + } + ], + "images": "72531_exterior_image_2_left#0", + "gt": "[[101, 105], [141, 111], [170, 107], [164, 109], [173, 116], [187, 102], [194, 103], [172, 111], [166, 98], [159, 105]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_037522062165#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the fifth block on the table, transfer the fifth block from the table to the position of the fourth block, place the fifth block on top of the fourth block.'\nStarting at [0.4, 0.41], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.41], [0.45, 0.62], [0.44, 0.56], [0.41, 0.53], [0.39, 0.3], [0.36, 0.27], [0.3, 0.31], [0.28, 0.36], [0.28, 0.26], [0.29, 0.26]]}" + } + ], + "images": "RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_037522062165#819", + "gt": "[[255, 148], [286, 225], [279, 202], [263, 190], [252, 107], [228, 98], [195, 113], [182, 128], [179, 95], [184, 93]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "31185_exterior_image_2_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the lid off the pot and put it on the table, take the toy tomato out of the pot and put it on the table, then use the towel to wipe the table', and now you are performing subtask 'wipe the table with the towel.'\nYou begin at [0.59, 0.63]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.63], [0.77, 0.52], [0.79, 0.41], [0.75, 0.45], [0.83, 0.51], [0.77, 0.48], [0.8, 0.54], [0.83, 0.52], [0.77, 0.53], [0.79, 0.58]]}" + } + ], + "images": "31185_exterior_image_2_left#370", + "gt": "[[190, 114], [247, 93], [254, 74], [241, 81], [267, 91], [248, 86], [255, 98], [266, 93], [245, 96], [252, 104]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the first button on the desk.'\nStarting at [0.63, 0.38], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.38], [0.63, 0.44], [0.63, 0.49], [0.67, 0.66], [0.66, 0.72], [0.68, 0.77], [0.67, 0.79], [0.66, 0.76], [0.66, 0.77], [0.66, 0.78]]}" + } + ], + "images": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_038522062547#0", + "gt": "[[405, 135], [403, 157], [406, 177], [426, 238], [423, 258], [433, 278], [426, 285], [421, 275], [420, 279], [425, 282]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0023_user_0010_scene_0003_cfg_0003_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the power strip by pressing the button', and now you are performing subtask 'turn on the power strip by pressing the button.'\nYour initial coordinate is [0.38, 0.0]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.37, 0.03], [0.39, 0.02], [0.4, 0.07], [0.46, 0.14], [0.48, 0.22], [0.46, 0.24], [0.45, 0.27], [0.45, 0.13], [0.43, 0.08]]}" + } + ], + "images": "RH20T_cfg3_task_0023_user_0010_scene_0003_cfg_0003_104122062823#0", + "gt": "[[246, 0], [239, 10], [250, 7], [253, 24], [292, 52], [310, 78], [292, 86], [289, 97], [291, 46], [272, 28]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "41728_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the cup from the countertop and pour its contents into the bowl', and now you are performing subtask 'pick up the cup from the countertop and pour its contents into the bowl.'\nStarting at [0.45, 0.63], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.63], [0.46, 0.63], [0.46, 0.57], [0.47, 0.55], [0.5, 0.56], [0.51, 0.59], [0.51, 0.62], [0.5, 0.59], [0.43, 0.58], [0.42, 0.51]]}" + } + ], + "images": "41728_exterior_image_2_left#0", + "gt": "[[143, 113], [147, 114], [148, 103], [151, 99], [161, 101], [164, 106], [164, 111], [159, 107], [139, 104], [134, 91]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "14884_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the microwave light, turn off the microwave light', and now you are performing subtask 'press a button on the microwave.'\nYou begin at [0.34, 0.58]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.34, 0.58], [0.39, 0.55], [0.44, 0.56], [0.46, 0.53], [0.49, 0.57], [0.55, 0.58], [0.55, 0.56], [0.52, 0.59], [0.49, 0.58], [0.49, 0.56]]}" + } + ], + "images": "14884_exterior_image_2_left#0", + "gt": "[[109, 105], [124, 99], [140, 101], [148, 96], [157, 103], [176, 104], [176, 101], [168, 106], [158, 104], [158, 101]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "68188_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the carrot plush toy in the pot', and now you are performing subtask 'put the carrot plush toy in the pot.'\nYou start from [0.14, 0.02]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.14, 0.02], [0.38, 0.4], [0.48, 0.41], [0.45, 0.43], [0.48, 0.47], [0.58, 0.49], [0.51, 0.56], [0.6, 0.52], [0.66, 0.69], [0.66, 0.58]]}" + } + ], + "images": "68188_exterior_image_1_left#0", + "gt": "[[46, 4], [120, 72], [154, 74], [144, 77], [154, 85], [186, 89], [163, 101], [193, 94], [210, 124], [211, 104]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "57878_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the red-white book and the purple object in the black box', and now you are performing subtask 'place the purple object in the black box.'\nYou begin at [0.47, 0.51]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.51], [0.46, 0.54], [0.49, 0.53], [0.47, 0.26], [0.43, 0.24], [0.4, 0.31], [0.44, 0.34], [0.44, 0.38], [0.46, 0.32], [0.5, 0.33]]}" + } + ], + "images": "57878_exterior_image_2_left#276", + "gt": "[[150, 92], [147, 97], [156, 96], [150, 47], [139, 43], [128, 56], [140, 62], [140, 69], [148, 57], [159, 59]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0058_user_0015_scene_0003_cfg_0001_035622060973#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn off the power strips', and now you are performing subtask 'press the second button on the power socket, press the third button on the power socket.'\nStart at [0.36, 0.24]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.24], [0.36, 0.24], [0.36, 0.24], [0.36, 0.24], [0.36, 0.22], [0.35, 0.25], [0.35, 0.26], [0.36, 0.27], [0.36, 0.28], [0.35, 0.27]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0058_user_0015_scene_0003_cfg_0001_035622060973#195.jpg", + "gt": "[[230, 86], [230, 88], [231, 88], [229, 87], [230, 81], [226, 90], [226, 95], [228, 98], [228, 99], [227, 98]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "83621_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the brown toy and place it on the stove', and now you are performing subtask 'pick up the brown toy and place it on the stove.'\nWith start location [0.03, 0.03], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.03, 0.03], [0.07, 0.03], [0.21, 0.32], [0.35, 0.46], [0.34, 0.53], [0.34, 0.46], [0.32, 0.49], [0.4, 0.57], [0.65, 0.58], [0.68, 0.69]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/83621_exterior_image_1_left#0.jpg", + "gt": "[[11, 6], [23, 5], [67, 58], [111, 83], [109, 95], [109, 83], [104, 89], [129, 103], [207, 105], [217, 125]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "25286_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the boxes inside the case', and now you are performing subtask 'place the box inside the box.'\nBegin at [0.68, 0.79] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.79], [0.74, 0.96], [0.7, 0.94], [0.68, 0.99], [0.67, 0.69], [0.66, 0.76], [0.68, 0.78], [0.67, 0.71], [0.57, 0.58], [0.54, 0.61]]}" + } + ], + "images": "25286_exterior_image_1_left#209", + "gt": "[[217, 143], [236, 172], [224, 170], [219, 179], [214, 124], [210, 137], [218, 141], [214, 127], [183, 104], [174, 110]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "11415_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the red cup from the sink and put it in the dish rack', and now you are performing subtask 'pick up the red cup from the sink and put it in the dish rack.'\nStarting at [0.1, 0.24], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.1, 0.24], [0.62, 0.89], [0.53, 0.62], [0.4, 0.52], [0.37, 0.24], [0.33, 0.41], [0.38, 0.28], [0.44, 0.28], [0.43, 0.49], [0.43, 0.49]]}" + } + ], + "images": "11415_exterior_image_2_left#0", + "gt": "[[32, 44], [199, 160], [170, 112], [127, 93], [118, 44], [106, 74], [122, 50], [141, 50], [138, 88], [138, 88]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "67544_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the left cabinet's drawer', and now you are performing subtask 'close the left cabinet's drawer.'\nBegin at [0.43, 0.63] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.63], [0.43, 0.62], [0.43, 0.52], [0.43, 0.51], [0.44, 0.41], [0.45, 0.42], [0.46, 0.39], [0.47, 0.39], [0.47, 0.39], [0.48, 0.39]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/67544_exterior_image_2_left#0.jpg", + "gt": "[[137, 113], [138, 112], [138, 94], [139, 92], [141, 74], [144, 75], [148, 71], [150, 71], [150, 70], [154, 70]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0010_user_0010_scene_0007_cfg_0005_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the handset of the telephone on the corresponding phone cradle', and now you are performing subtask 'place the handset of the telephone on the corresponding phone cradle.'\nYour initial coordinate is [0.45, 0.42]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.42], [0.44, 0.46], [0.41, 0.45], [0.4, 0.5], [0.36, 0.5], [0.38, 0.44], [0.37, 0.47], [0.37, 0.41], [0.41, 0.39], [0.45, 0.5]]}" + } + ], + "images": "RH20T_cfg5_task_0010_user_0010_scene_0007_cfg_0005_037522062165#0", + "gt": "[[289, 152], [280, 165], [263, 161], [258, 179], [232, 180], [246, 158], [235, 168], [238, 147], [262, 142], [289, 180]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0007_user_0014_scene_0007_cfg_0006_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'sweeping a small block with the gripper from left to right on the table', and now you are performing subtask 'sweeping a small block with the gripper from left to right on the table.'\nYou begin at [0.53, 0.31]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.31], [0.55, 0.37], [0.57, 0.46], [0.57, 0.63], [0.55, 0.73], [0.53, 0.5], [0.55, 0.45], [0.53, 0.29], [0.56, 0.27], [0.54, 0.27]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0007_user_0014_scene_0007_cfg_0006_104122064161#0.jpg", + "gt": "[[338, 110], [353, 134], [362, 165], [362, 227], [355, 262], [340, 179], [349, 161], [340, 103], [356, 98], [348, 98]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0105_user_0020_scene_0007_cfg_0006_037522061512#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the object on the shelf', and now you are performing subtask 'pick up the object from the tabletop, transfer the object from the tabletop to the specified location, place the object at the specified location, pick up the object from the tabletop.'\nYou begin at [0.34, 0.33]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.34, 0.33], [0.37, 0.32], [0.39, 0.26], [0.4, 0.35], [0.37, 0.32], [0.32, 0.36], [0.31, 0.46], [0.32, 0.53], [0.3, 0.41], [0.33, 0.34]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0105_user_0020_scene_0007_cfg_0006_037522061512#438.jpg", + "gt": "[[218, 119], [235, 117], [247, 94], [257, 126], [234, 115], [208, 130], [199, 165], [203, 191], [189, 149], [210, 124]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0002_user_0014_scene_0004_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull out a napkin', and now you are performing subtask 'pull out a napkin.'\nBegin at [0.63, 0.37] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.66, 0.27], [0.65, 0.34], [0.68, 0.48], [0.67, 0.49], [0.66, 0.4], [0.66, 0.24], [0.68, 0.22], [0.66, 0.04], [0.66, 0.04]]}" + } + ], + "images": "RH20T_cfg1_task_0002_user_0014_scene_0004_cfg_0001_038522062547#0", + "gt": "[[406, 133], [421, 98], [415, 122], [433, 174], [428, 176], [420, 144], [424, 86], [438, 78], [424, 13], [424, 13]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_038522062547#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the first block on top of the second position, pick up the second block from the desktop, transfer the second block from the desktop to the second position above the first block, place the second block on top of the first block.'\nThe trajectory starts at [0.69, 0.39]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.39], [0.62, 0.46], [0.64, 0.54], [0.64, 0.52], [0.63, 0.54], [0.63, 0.51], [0.62, 0.5], [0.69, 0.43], [0.72, 0.34], [0.71, 0.33]]}" + } + ], + "images": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_038522062547#171", + "gt": "[[442, 140], [400, 165], [408, 193], [410, 188], [402, 194], [403, 184], [396, 181], [443, 154], [458, 121], [453, 120]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "6997_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the peach plush toy in the box and remove the green plush from the box and put it on the tray', and now you are performing subtask 'place peach plush toy in box.'\nStarting at [0.43, 0.56], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.56], [0.47, 0.69], [0.46, 0.72], [0.47, 0.8], [0.45, 0.84], [0.45, 0.79], [0.48, 0.69], [0.52, 0.64], [0.54, 0.64], [0.53, 0.69]]}" + } + ], + "images": "6997_exterior_image_2_left#0", + "gt": "[[138, 100], [151, 124], [148, 129], [152, 144], [144, 152], [143, 143], [153, 124], [165, 115], [172, 116], [169, 124]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "6963_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the stopper in the drain of the sink', and now you are performing subtask 'put the stopper in the drain of the sink.'\nYou begin at [0.45, 0.32]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.32], [0.56, 0.36], [0.74, 0.53], [0.76, 0.48], [0.78, 0.48], [0.77, 0.51], [0.78, 0.5], [0.75, 0.44], [0.7, 0.47], [0.68, 0.44]]}" + } + ], + "images": "6963_exterior_image_1_left#0", + "gt": "[[144, 57], [179, 64], [236, 96], [242, 86], [249, 86], [245, 91], [249, 90], [240, 80], [225, 85], [216, 80]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "47675_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn off the switch on the extension cord then close the clear container', and now you are performing subtask 'push the clear container.'\nFrom start point [0.57, 0.38], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.38], [0.36, 0.61], [0.33, 0.79], [0.26, 0.74], [0.24, 0.62], [0.33, 0.51], [0.33, 0.36], [0.45, 0.36], [0.49, 0.41], [0.48, 0.37]]}" + } + ], + "images": "47675_exterior_image_1_left#104", + "gt": "[[182, 69], [116, 110], [106, 143], [84, 133], [77, 112], [106, 92], [105, 65], [145, 64], [157, 74], [155, 66]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0044_user_0014_scene_0005_cfg_0007_104122061330#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up a bag of things', and now you are performing subtask 'pick up a bag of things.'\nWith start location [0.42, 0.29], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.29], [0.43, 0.33], [0.45, 0.49], [0.44, 0.55], [0.43, 0.63], [0.43, 0.64], [0.43, 0.63], [0.43, 0.58], [0.41, 0.36], [0.38, 0.24]]}" + } + ], + "images": "RH20T_cfg7_task_0044_user_0014_scene_0005_cfg_0007_104122061330#0", + "gt": "[[266, 103], [277, 118], [287, 178], [283, 199], [272, 227], [274, 230], [277, 228], [277, 208], [264, 131], [243, 88]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "511_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the box slightly to the center of the table', and now you are performing subtask 'move the box slightly to the center of the table.'\nWith start location [0.72, 0.11], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.72, 0.11], [0.68, 0.26], [0.67, 0.53], [0.61, 0.59], [0.6, 0.66], [0.56, 0.63], [0.53, 0.57], [0.55, 0.54], [0.53, 0.42], [0.56, 0.36]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/511_exterior_image_1_left#0.jpg", + "gt": "[[230, 20], [217, 47], [215, 95], [196, 107], [193, 118], [180, 114], [169, 102], [175, 98], [171, 75], [178, 64]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0103_user_0014_scene_0012_cfg_0004_104122062823#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'drag the plate to the goal post after holding it down', and now you are performing subtask 'push the plate towards the target bar, push the plate towards the target bar.'\nBegin at [0.41, 0.41] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.41], [0.43, 0.41], [0.44, 0.31], [0.45, 0.32], [0.43, 0.43], [0.45, 0.35], [0.48, 0.15], [0.47, 0.05], [0.47, 0.03], [0.44, 0.01]]}" + } + ], + "images": "RH20T_cfg4_task_0103_user_0014_scene_0012_cfg_0004_104122062823#157", + "gt": "[[265, 148], [276, 149], [279, 111], [286, 116], [272, 156], [289, 126], [307, 55], [302, 17], [298, 9], [284, 3]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0091_user_0010_scene_0004_cfg_0003_104122062823#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the green button, the basketball toy, and the football toy from the black box to the pink box', and now you are performing subtask 'place the green button with the yellow base in the pink box, pick up the basketball toy from the black box, transfer the basketball toy from the black box to above the pink box.'\nYour starting position is [0.25, 0.07]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.25, 0.07], [0.34, 0.0], [0.49, 0.07], [0.54, 0.15], [0.5, 0.22], [0.52, 0.29], [0.52, 0.17], [0.49, 0.09], [0.37, 0.05], [0.3, 0.07]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg3_task_0091_user_0010_scene_0004_cfg_0003_104122062823#277.jpg", + "gt": "[[157, 24], [220, 0], [312, 24], [343, 53], [320, 80], [332, 106], [336, 61], [311, 33], [237, 18], [194, 26]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "18085_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'folded shirt', and now you are performing subtask 'fold the shirt on the table.'\nYour starting position is [0.43, 0.55]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.55], [0.52, 0.67], [0.44, 0.51], [0.44, 0.44], [0.36, 0.55], [0.3, 0.57], [0.32, 0.62], [0.28, 0.66], [0.33, 0.65], [0.37, 0.72]]}" + } + ], + "images": "18085_exterior_image_1_left#0", + "gt": "[[138, 99], [165, 120], [142, 92], [142, 80], [115, 99], [96, 102], [101, 111], [91, 119], [107, 117], [118, 129]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg3_task_0057_user_0010_scene_0005_cfg_0003_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn them on', and now you are performing subtask 'press the first button on the power board.'\nYou begin at [0.38, 0.0]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.34, 0.01], [0.34, 0.06], [0.3, 0.05], [0.36, 0.04], [0.38, 0.11], [0.36, 0.16], [0.38, 0.15], [0.36, 0.19], [0.35, 0.14]]}" + } + ], + "images": "RH20T_cfg3_task_0057_user_0010_scene_0005_cfg_0003_104122062823#0", + "gt": "[[246, 0], [220, 2], [220, 22], [192, 18], [230, 14], [245, 40], [229, 59], [242, 55], [232, 70], [227, 49]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "38730_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the towel to wipe on the table', and now you are performing subtask 'slide the towel to wipe the surface on the table.'\nStarting at [0.68, 0.63], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.63], [0.66, 0.64], [0.61, 0.64], [0.6, 0.6], [0.56, 0.53], [0.56, 0.5], [0.54, 0.45], [0.53, 0.45], [0.55, 0.46], [0.54, 0.46]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/38730_exterior_image_1_left#166.jpg", + "gt": "[[217, 114], [210, 116], [194, 115], [193, 108], [179, 95], [178, 90], [173, 81], [170, 81], [176, 83], [173, 83]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122061330#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'place the block on the block board, pick up the block on the flat surface, transfer the block from the flat surface to the block board, place the block on the block board.'\nBegin at [0.5, 0.49] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.49], [0.56, 0.54], [0.59, 0.59], [0.52, 0.49], [0.51, 0.4], [0.5, 0.34], [0.49, 0.35], [0.49, 0.37], [0.51, 0.36], [0.53, 0.43]]}" + } + ], + "images": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122061330#332", + "gt": "[[320, 178], [356, 193], [375, 212], [335, 177], [329, 143], [317, 123], [314, 125], [313, 132], [324, 128], [342, 154]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0010_user_0010_scene_0005_cfg_0005_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the handset of the telephone on the corresponding phone cradle', and now you are performing subtask 'place the handset of the telephone on the corresponding phone cradle.'\nYour starting position is [0.48, 0.14]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.14], [0.49, 0.43], [0.46, 0.5], [0.46, 0.56], [0.5, 0.32], [0.54, 0.24], [0.56, 0.23], [0.57, 0.26], [0.56, 0.28], [0.52, 0.22]]}" + } + ], + "images": "RH20T_cfg5_task_0010_user_0010_scene_0005_cfg_0005_104122061850#0", + "gt": "[[305, 50], [312, 153], [293, 179], [293, 202], [323, 117], [344, 86], [360, 82], [366, 92], [360, 102], [330, 79]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "18044_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the packet on the box', and now you are performing subtask 'throw the second packet into the box.'\nYou start from [0.66, 0.73]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.66, 0.73], [0.67, 0.76], [0.64, 0.8], [0.64, 0.83], [0.61, 0.82], [0.62, 0.77], [0.65, 0.73], [0.69, 0.69], [0.69, 0.65], [0.7, 0.63]]}" + } + ], + "images": "18044_exterior_image_2_left#134", + "gt": "[[211, 132], [213, 137], [206, 144], [204, 149], [194, 148], [200, 139], [207, 132], [220, 124], [221, 117], [224, 114]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "13437_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the door', and now you are performing subtask 'open the door.'\nWith start location [1.0, 0.37], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[1.0, 0.37], [0.68, 0.49], [0.5, 0.49], [0.47, 0.53], [0.55, 0.67], [0.73, 0.62], [0.86, 0.53], [0.89, 0.46], [0.88, 0.42], [0.88, 0.42]]}" + } + ], + "images": "13437_exterior_image_1_left#0", + "gt": "[[319, 66], [216, 88], [160, 89], [152, 96], [176, 121], [235, 112], [276, 96], [286, 83], [283, 75], [283, 75]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "6777_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'the contents in the flask', and now you are performing subtask 'clean the cup with the brush.'\nThe trajectory starts at [0.17, 0.33]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.17, 0.33], [0.28, 0.41], [0.28, 0.57], [0.3, 0.53], [0.32, 0.55], [0.31, 0.32], [0.31, 0.51], [0.32, 0.31], [0.31, 0.52], [0.29, 0.24]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/6777_exterior_image_2_left#0.jpg", + "gt": "[[54, 60], [91, 74], [89, 102], [97, 96], [103, 99], [99, 57], [98, 92], [101, 56], [100, 94], [94, 44]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122063633#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'pick up the object in the container, transfer the object from the container to the surface, place the item on the table.'\nYour initial coordinate is [0.3, 0.56]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.3, 0.56], [0.3, 0.57], [0.3, 0.45], [0.32, 0.4], [0.32, 0.47], [0.32, 0.44], [0.29, 0.41], [0.28, 0.43], [0.28, 0.47], [0.29, 0.53]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122063633#349.jpg", + "gt": "[[191, 201], [195, 207], [195, 163], [205, 144], [206, 171], [202, 157], [187, 148], [180, 153], [177, 171], [183, 191]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "58247_exterior_image_1_left#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the microwave door, place the bowl inside the microwave', and now you are performing subtask 'pick up the bowl from the countertop.'\nYour initial coordinate is [0.47, 0.22]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.22], [0.51, 0.27], [0.51, 0.3], [0.48, 0.44], [0.5, 0.61], [0.46, 0.65], [0.49, 0.36], [0.53, 0.32], [0.54, 0.38], [0.54, 0.33]]}" + } + ], + "images": "58247_exterior_image_1_left#422", + "gt": "[[150, 39], [164, 48], [164, 54], [153, 80], [161, 110], [146, 117], [158, 64], [171, 58], [174, 68], [174, 60]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "22851_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'cover the lid of the pink object on the table', and now you are performing subtask 'pick up the lid on the table.'\nStart at [0.74, 0.21]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.74, 0.21], [0.58, 0.15], [0.54, 0.18], [0.52, 0.17], [0.56, 0.17], [0.5, 0.27], [0.47, 0.43], [0.48, 0.47], [0.48, 0.42], [0.5, 0.46]]}" + } + ], + "images": "22851_exterior_image_1_left#0", + "gt": "[[237, 38], [185, 27], [174, 33], [168, 31], [180, 30], [161, 48], [149, 77], [153, 85], [155, 76], [161, 83]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "64234_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the towel diagonally', and now you are performing subtask 'fold the towel diagonally.'\nFrom start point [0.9, 0.23], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.9, 0.23], [0.77, 0.41], [0.69, 0.62], [0.68, 0.63], [0.68, 0.57], [0.63, 0.56], [0.54, 0.63], [0.55, 0.68], [0.53, 0.69], [0.56, 0.66]]}" + } + ], + "images": "64234_exterior_image_1_left#0", + "gt": "[[287, 42], [248, 74], [222, 111], [216, 113], [218, 102], [202, 101], [173, 113], [177, 123], [170, 124], [179, 119]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "41917_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the grey spoon in the bowl', and now you are performing subtask 'transfer the grey spoon from the table to the bowl.'\nFrom start point [0.73, 0.38], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.73, 0.38], [0.6, 0.49], [0.58, 0.59], [0.52, 0.69], [0.56, 0.72], [0.52, 0.69], [0.49, 0.76], [0.55, 0.77], [0.49, 0.77], [0.52, 0.79]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/41917_exterior_image_1_left#0.jpg", + "gt": "[[233, 69], [191, 88], [185, 107], [165, 124], [178, 129], [168, 125], [158, 136], [175, 138], [157, 138], [166, 142]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0038_user_0020_scene_0006_cfg_0007_104122063633#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the cup', and now you are performing subtask 'pick up the cup.'\nBegin at [0.37, 0.34] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.34], [0.33, 0.38], [0.3, 0.48], [0.27, 0.49], [0.26, 0.52], [0.26, 0.5], [0.27, 0.5], [0.26, 0.41], [0.28, 0.37], [0.27, 0.35]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0038_user_0020_scene_0006_cfg_0007_104122063633#0.jpg", + "gt": "[[234, 121], [213, 138], [194, 174], [171, 175], [166, 188], [167, 180], [172, 181], [169, 148], [177, 134], [175, 125]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "48713_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack up the cups', and now you are performing subtask 'transfer the cup from the table to the top of another cup.'\nStarting at [0.36, 0.13], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.13], [0.39, 0.23], [0.39, 0.35], [0.34, 0.66], [0.38, 0.59], [0.47, 0.37], [0.49, 0.37], [0.49, 0.44], [0.5, 0.4], [0.5, 0.52]]}" + } + ], + "images": "48713_exterior_image_2_left#0", + "gt": "[[116, 24], [125, 42], [126, 63], [109, 118], [121, 107], [149, 67], [156, 67], [158, 79], [159, 72], [159, 94]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0063_user_0007_scene_0007_cfg_0005_104422070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use a shovel to scoop up corn', and now you are performing subtask 'use a shovel to scoop up corn.'\nThe trajectory starts at [0.58, 0.37]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.37], [0.62, 0.38], [0.68, 0.49], [0.69, 0.41], [0.67, 0.4], [0.67, 0.38], [0.69, 0.35], [0.66, 0.39], [0.63, 0.46], [0.67, 0.39]]}" + } + ], + "images": "RH20T_cfg5_task_0063_user_0007_scene_0007_cfg_0005_104422070044#0", + "gt": "[[374, 132], [398, 138], [435, 177], [440, 149], [429, 145], [428, 138], [444, 126], [421, 139], [403, 165], [430, 142]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "37195_exterior_image_2_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'uncover the bottle and put the whisk and the spoon inside the bottle', and now you are performing subtask 'insert the spoon into the bottle.'\nBegin at [0.39, 0.45] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.45], [0.37, 0.66], [0.42, 0.75], [0.38, 0.76], [0.4, 0.83], [0.49, 0.22], [0.47, 0.21], [0.44, 0.46], [0.44, 0.36], [0.37, 0.3]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/37195_exterior_image_2_left#549.jpg", + "gt": "[[126, 81], [117, 119], [134, 135], [123, 137], [129, 149], [157, 40], [150, 37], [142, 83], [140, 64], [119, 54]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "68303_exterior_image_1_left#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the carrot plush toy backwards, move the blue towel backwards', and now you are performing subtask 'push the blue towel backwards.'\nFrom start point [0.42, 0.57], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.57], [0.42, 0.56], [0.42, 0.56], [0.42, 0.56], [0.42, 0.58], [0.43, 0.57], [0.43, 0.56], [0.38, 0.53], [0.36, 0.51], [0.37, 0.52]]}" + } + ], + "images": "68303_exterior_image_1_left#203", + "gt": "[[135.5, 102.5], [135.5, 100.5], [135.5, 101.5], [134.5, 101.5], [134.5, 104.5], [136.5, 103.5], [137.5, 100.5], [120.5, 95.5], [115.5, 92.5], [118.5, 94.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "69906_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer, take out the black object, and place it on the countertop, pick up the pen on the countertop, put it back in the drawer, close the drawer', and now you are performing subtask 'place the black object on the countertop.'\nYour starting position is [0.28, 0.83]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.28, 0.83], [0.27, 0.79], [0.26, 0.73], [0.3, 0.63], [0.4, 0.58], [0.49, 0.57], [0.53, 0.58], [0.58, 0.61], [0.59, 0.63], [0.56, 0.58]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/69906_exterior_image_1_left#311.jpg", + "gt": "[[91, 149], [85, 143], [83, 131], [96, 114], [128, 105], [156, 103], [170, 104], [187, 109], [190, 113], [178, 105]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0017_user_0014_scene_0002_cfg_0006_104122064161#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the pen into the pen holder', and now you are performing subtask 'pick up the pen from the table, transfer the pen from the table to the position of the pen holder, place the pen in the pen holder.'\nBegin at [0.46, 0.27] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.27], [0.48, 0.34], [0.47, 0.46], [0.47, 0.34], [0.46, 0.24], [0.45, 0.21], [0.44, 0.23], [0.43, 0.23], [0.46, 0.26], [0.46, 0.26]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0017_user_0014_scene_0002_cfg_0006_104122064161#102.jpg", + "gt": "[[295, 96], [309, 123], [298, 166], [300, 123], [294, 88], [286, 76], [284, 82], [274, 82], [292, 95], [292, 95]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0004_user_0006_scene_0003_cfg_0002_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the small block on the left and move it to the right', and now you are performing subtask 'pick up the small block on the left and move it to the right.'\nWith start location [0.45, 0.54], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.54], [0.56, 0.67], [0.58, 0.8], [0.5, 0.89], [0.51, 0.98], [0.52, 0.94], [0.47, 0.79], [0.46, 0.84], [0.45, 0.71], [0.45, 0.71]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg2_task_0004_user_0006_scene_0003_cfg_0002_104122063678#0.jpg", + "gt": "[[289, 196], [358, 241], [371, 289], [322, 322], [325, 353], [331, 337], [303, 286], [297, 302], [289, 257], [289, 257]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0061_user_0014_scene_0008_cfg_0001_035622060973#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'pick up the small cube from the large cube, transfer the small cube from the large cube to the lid of the gift box, place the small cube on the lid of the gift box.'\nThe trajectory starts at [0.39, 0.36]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.36], [0.4, 0.41], [0.42, 0.43], [0.44, 0.49], [0.49, 0.52], [0.49, 0.54], [0.48, 0.49], [0.46, 0.44], [0.43, 0.41], [0.4, 0.41]]}" + } + ], + "images": "RH20T_cfg1_task_0061_user_0014_scene_0008_cfg_0001_035622060973#0", + "gt": "[[247, 130], [255, 146], [268, 156], [282, 178], [311, 188], [311, 196], [306, 178], [292, 158], [277, 148], [257, 149]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "54044_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up and put down the green block, then put the salt cellar into the pot and the green block into the sink', and now you are performing subtask 'place the green block in the sink.'\nYour starting position is [0.68, 0.44]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.44], [0.62, 0.52], [0.65, 0.56], [0.62, 0.58], [0.65, 0.56], [0.63, 0.52], [0.66, 0.54], [0.64, 0.58], [0.55, 0.63], [0.72, 0.49]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/54044_exterior_image_1_left#531.jpg", + "gt": "[[217, 80], [200, 94], [208, 101], [199, 105], [208, 100], [201, 94], [210, 98], [204, 105], [175, 113], [231, 89]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "42060_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the green bowl to the designated position', and now you are performing subtask 'move the green bowl to the designated position.'\nStarting at [0.73, 0.33], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.73, 0.33], [0.77, 0.49], [0.72, 0.6], [0.74, 0.67], [0.77, 0.54], [0.73, 0.49], [0.83, 0.5], [0.89, 0.62], [0.85, 0.6], [0.85, 0.6]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/42060_exterior_image_1_left#0.jpg", + "gt": "[[234, 59], [245, 89], [229, 108], [236, 121], [247, 97], [233, 88], [266, 90], [284, 111], [273, 108], [273, 108]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_038522062547#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'pick up the third square from the right side of the table, transfer the third square from the right side of the table to the top of the second square, place the third square beside the second square, pick up the fourth square on the table.'\nThe trajectory starts at [0.67, 0.5]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.67, 0.5], [0.73, 0.54], [0.71, 0.51], [0.63, 0.48], [0.62, 0.45], [0.61, 0.46], [0.61, 0.49], [0.6, 0.48], [0.65, 0.46], [0.71, 0.47]]}" + } + ], + "images": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_038522062547#361", + "gt": "[[426, 180], [470, 195], [455, 183], [406, 172], [397, 163], [390, 164], [393, 175], [387, 174], [415, 164], [454, 170]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0032_user_0020_scene_0009_cfg_0006_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'slide the outer casing onto the gift box', and now you are performing subtask 'slide the outer casing onto the gift box.'\nBegin at [0.53, 0.31] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.31], [0.53, 0.38], [0.56, 0.53], [0.58, 0.78], [0.59, 0.64], [0.59, 0.57], [0.61, 0.55], [0.6, 0.52], [0.61, 0.56], [0.61, 0.34]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0032_user_0020_scene_0009_cfg_0006_104122064161#0.jpg", + "gt": "[[338, 111], [337, 136], [358, 190], [370, 281], [380, 229], [377, 205], [391, 199], [387, 189], [391, 200], [389, 121]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122061330#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the block on the other block, pick up the block on the surface, transfer the block from the surface to the other block, place the block on the other block.'\nStart at [0.46, 0.38]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.38], [0.5, 0.28], [0.52, 0.29], [0.52, 0.26], [0.53, 0.29], [0.52, 0.27], [0.47, 0.28], [0.45, 0.34], [0.47, 0.31], [0.52, 0.31]]}" + } + ], + "images": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122061330#404", + "gt": "[[294, 135], [317, 99], [334, 106], [334, 93], [340, 106], [330, 96], [303, 100], [290, 122], [301, 110], [335, 113]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_104122061850#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the block on the other blocks, pick up the block on the plane, transfer the block from the plane to the other blocks, place the block on the other blocks.'\nStarting at [0.51, 0.18], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.18], [0.55, 0.2], [0.58, 0.38], [0.63, 0.52], [0.62, 0.52], [0.6, 0.34], [0.52, 0.16], [0.5, 0.15], [0.5, 0.11], [0.5, 0.12]]}" + } + ], + "images": "RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_104122061850#580", + "gt": "[[327, 64], [349, 72], [370, 136], [405, 188], [400, 186], [384, 122], [332, 56], [318, 53], [320, 40], [323, 44]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "55751_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the black ladle out of the glass jar', and now you are performing subtask 'take the black ladle out of the glass jar.'\nYour initial coordinate is [0.25, 0.0]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.25, 0.0], [0.42, 0.35], [0.4, 0.49], [0.44, 0.18], [0.34, 0.42], [0.34, 0.54], [0.38, 0.58], [0.34, 0.71], [0.36, 0.59], [0.33, 0.45]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/55751_exterior_image_2_left#0.jpg", + "gt": "[[81, 0], [135, 63], [129, 88], [140, 32], [110, 76], [109, 97], [122, 105], [109, 127], [116, 107], [107, 81]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "815_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the lid on the cup', and now you are performing subtask 'open the lid on the cup.'\nBegin at [1.0, 0.23] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[1.0, 0.22], [0.91, 0.39], [0.88, 0.56], [0.86, 0.57], [0.85, 0.6], [0.87, 0.43], [0.88, 0.39], [0.89, 0.38], [0.9, 0.36], [0.9, 0.37]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/815_exterior_image_1_left#0.jpg", + "gt": "[[318.5, 40.5], [292.5, 69.5], [281.5, 101.5], [275.5, 102.5], [271.5, 107.5], [277.5, 76.5], [280.5, 69.5], [285.5, 68.5], [289.5, 64.5], [288.5, 66.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "72609_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker in the cup', and now you are performing subtask 'put the marker in the cup.'\nYou start from [0.74, 0.17]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.74, 0.17], [0.59, 0.27], [0.54, 0.39], [0.52, 0.4], [0.49, 0.53], [0.47, 0.37], [0.53, 0.29], [0.55, 0.3], [0.55, 0.38], [0.55, 0.31]]}" + } + ], + "images": "72609_exterior_image_2_left#0", + "gt": "[[236, 30], [190, 49], [174, 71], [165, 72], [157, 95], [151, 66], [170, 53], [175, 54], [177, 68], [175, 56]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "64124_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the green spoon slightly to the left', and now you are performing subtask 'move the green spoon slightly to the left.'\nYou start from [0.4, 0.12]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.12], [0.51, 0.12], [0.66, 0.21], [0.71, 0.28], [0.71, 0.52], [0.73, 0.53], [0.73, 0.57], [0.7, 0.52], [0.6, 0.59], [0.6, 0.44]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/64124_exterior_image_1_left#0.jpg", + "gt": "[[128.5, 21.0], [162.5, 21.0], [211.5, 38.0], [228.5, 50.0], [226.5, 93.0], [232.5, 95.0], [232.5, 103.0], [223.5, 94.0], [192.5, 106.0], [192.5, 79.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "40844_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the toy to the pot in the sink', and now you are performing subtask 'move the toy to the pot in the sink.'\nYour initial coordinate is [0.66, 0.36]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.66, 0.36], [0.65, 0.51], [0.66, 0.63], [0.62, 0.8], [0.66, 0.71], [0.71, 0.7], [0.76, 0.75], [0.79, 0.8], [0.79, 0.85], [0.72, 0.68]]}" + } + ], + "images": "40844_exterior_image_1_left#0", + "gt": "[[211.0, 64.5], [207.0, 92.5], [212.0, 113.5], [197.0, 143.5], [211.0, 128.5], [227.0, 126.5], [244.0, 135.5], [252.0, 143.5], [252.0, 152.5], [229.0, 122.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0029_user_0010_scene_0006_cfg_0005_036422060215#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp the handle and open the drawer', and now you are performing subtask 'grab the handle of the drawer, pull the handle to open the drawer.'\nThe trajectory starts at [0.4, 0.34]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.34], [0.43, 0.35], [0.43, 0.36], [0.43, 0.36], [0.42, 0.35], [0.42, 0.36], [0.41, 0.36], [0.41, 0.39], [0.42, 0.38], [0.43, 0.37]]}" + } + ], + "images": "RH20T_cfg5_task_0029_user_0010_scene_0006_cfg_0005_036422060215#151", + "gt": "[[256, 124], [274, 127], [275, 130], [276, 128], [268, 125], [267, 128], [264, 128], [263, 139], [271, 138], [277, 134]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "36877_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'twist the upper knob on the oven', and now you are performing subtask 'twist the upper knob on the oven.'\nBegin at [0.66, 0.3] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.66, 0.3], [0.63, 0.29], [0.57, 0.34], [0.62, 0.35], [0.64, 0.4], [0.59, 0.38], [0.57, 0.44], [0.63, 0.39], [0.62, 0.34], [0.62, 0.39]]}" + } + ], + "images": "36877_exterior_image_1_left#0", + "gt": "[[211, 54], [202, 52], [184, 61], [199, 63], [204, 72], [188, 69], [182, 79], [203, 71], [198, 62], [197, 70]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0037_user_0007_scene_0006_cfg_0002_104122063678#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the first block on the second block, pick up the second block on the table, transfer the second block from the table to the top of the third block, place the second block on the third block.'\nYour initial coordinate is [0.54, 0.61]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.61], [0.54, 0.6], [0.58, 0.59], [0.58, 0.64], [0.58, 0.66], [0.58, 0.63], [0.58, 0.59], [0.56, 0.59], [0.54, 0.56], [0.55, 0.58]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg2_task_0037_user_0007_scene_0006_cfg_0002_104122063678#310.jpg", + "gt": "[[343, 221], [344, 216], [370, 211], [373, 232], [369, 238], [373, 226], [371, 213], [356, 211], [348, 203], [349, 209]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0085_user_0015_scene_0002_cfg_0001_039422060546#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the bubble ring from the assembled bubble ring and ball', and now you are performing subtask 'remove the bubble ring from the assembled bubble ring and ball.'\nThe trajectory starts at [0.41, 0.34]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.34], [0.49, 0.35], [0.54, 0.42], [0.57, 0.41], [0.5, 0.52], [0.48, 0.6], [0.5, 0.66], [0.47, 0.56], [0.49, 0.65], [0.47, 0.73]]}" + } + ], + "images": "RH20T_cfg1_task_0085_user_0015_scene_0002_cfg_0001_039422060546#0", + "gt": "[[265, 123], [311, 126], [345, 152], [364, 149], [321, 189], [310, 216], [318, 238], [304, 200], [315, 235], [299, 264]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "40480_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the marker from the green cup', and now you are performing subtask 'remove the marker from the green cup.'\nStarting at [0.84, 0.57], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.84, 0.57], [0.92, 0.46], [0.9, 0.47], [0.94, 0.52], [0.91, 0.61], [0.88, 0.61], [0.88, 0.64], [0.96, 0.56], [0.83, 0.61], [0.75, 0.72]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/40480_exterior_image_2_left#0.jpg", + "gt": "[[268, 103], [294, 83], [289, 85], [300, 93], [290, 110], [282, 110], [280, 116], [306, 100], [267, 110], [240, 130]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0039_user_0014_scene_0003_cfg_0006_104122060811#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pour the water from one cup into another empty cup', and now you are performing subtask 'pour the water from one cup into another empty cup.'\nThe trajectory starts at [0.6, 0.35]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.35], [0.54, 0.41], [0.63, 0.33], [0.64, 0.32], [0.63, 0.32], [0.6, 0.33], [0.55, 0.42], [0.57, 0.4], [0.59, 0.4], [0.58, 0.41]]}" + } + ], + "images": "RH20T_cfg6_task_0039_user_0014_scene_0003_cfg_0006_104122060811#0", + "gt": "[[385, 127], [347, 149], [402, 120], [407, 115], [401, 114], [387, 119], [352, 150], [362, 143], [378, 143], [374, 146]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "18085_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'folded shirt', and now you are performing subtask 'fold the shirt on the table.'\nWith start location [0.33, 0.83], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.33, 0.83], [0.34, 0.61], [0.37, 0.59], [0.38, 0.61], [0.39, 0.69], [0.38, 0.64], [0.36, 0.61], [0.34, 0.69], [0.35, 0.71], [0.35, 0.74]]}" + } + ], + "images": "18085_exterior_image_1_left#233", + "gt": "[[107, 149], [110, 110], [118, 106], [122, 110], [126, 124], [123, 116], [114, 109], [110, 125], [113, 128], [113, 134]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_037522061512#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'pick up the lid on the box, transfer the lid from the box to the tabletop, place the lid on the tabletop, grab the object in the box.'\nYour initial coordinate is [0.41, 0.24]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.24], [0.38, 0.3], [0.35, 0.32], [0.35, 0.39], [0.33, 0.35], [0.28, 0.47], [0.28, 0.52], [0.29, 0.59], [0.29, 0.48], [0.33, 0.4]]}" + } + ], + "images": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_037522061512#0", + "gt": "[[261, 86], [243, 108], [222, 117], [226, 142], [212, 125], [182, 169], [177, 189], [184, 211], [184, 172], [213, 144]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "57664_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the spoon out of the cup, place it on the table, and pour the contents of the cup into another cup', and now you are performing subtask 'place the cup on the table.'\nYour starting position is [0.47, 0.27]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.27], [0.51, 0.3], [0.48, 0.21], [0.48, 0.27], [0.44, 0.31], [0.46, 0.28], [0.47, 0.34], [0.49, 0.29], [0.49, 0.36], [0.48, 0.24]]}" + } + ], + "images": "57664_exterior_image_2_left#231", + "gt": "[[152, 48], [162, 54], [154, 38], [153, 49], [141, 56], [146, 51], [150, 62], [158, 53], [158, 65], [154, 44]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "56707_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the landfill bin to the left then right', and now you are performing subtask 'place the trash can to the right side of the table.'\nWith start location [0.53, 0.64], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.64], [0.53, 0.67], [0.53, 0.63], [0.55, 0.64], [0.55, 0.69], [0.58, 0.68], [0.59, 0.69], [0.6, 0.68], [0.57, 0.61], [0.52, 0.58]]}" + } + ], + "images": "56707_exterior_image_1_left#197", + "gt": "[[171.0, 115.0], [171.0, 120.0], [171.0, 114.0], [176.0, 116.0], [177.0, 124.0], [187.0, 122.0], [190.0, 125.0], [191.0, 122.0], [183.0, 110.0], [167.0, 104.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0033_user_0020_scene_0001_cfg_0007_104122061602#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pushing another small block to the designated position while grasping one small block', and now you are performing subtask 'pick up the front square from the cardboard, transfer the front square from the position to the position aligned with the second square.'\nYou start from [0.7, 0.32]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.32], [0.68, 0.33], [0.68, 0.36], [0.66, 0.29], [0.64, 0.26], [0.63, 0.27], [0.63, 0.25], [0.65, 0.27], [0.64, 0.32], [0.65, 0.31]]}" + } + ], + "images": "RH20T_cfg7_task_0033_user_0020_scene_0001_cfg_0007_104122061602#0", + "gt": "[[448, 116], [432, 118], [432, 129], [422, 105], [407, 95], [404, 98], [403, 90], [416, 98], [411, 117], [417, 113]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0091_user_0007_scene_0004_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move an object from one box to another', and now you are performing subtask 'pick up the block from the pink box, transfer the block from the pink box to the table, throw the block onto the table, pick up the block from the table.'\nWith start location [0.5, 0.37], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.54, 0.41], [0.57, 0.42], [0.6, 0.49], [0.63, 0.63], [0.61, 0.69], [0.63, 0.54], [0.58, 0.41], [0.54, 0.4], [0.52, 0.44]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0091_user_0007_scene_0004_cfg_0005_104122063678#0.jpg", + "gt": "[[323, 132], [343, 148], [363, 151], [383, 178], [401, 228], [392, 249], [404, 196], [373, 146], [346, 145], [331, 158]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "10406_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the green marker inside the red cup', and now you are performing subtask 'place the green marker inside the red cup.'\nYou begin at [0.86, 0.14]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.86, 0.14], [0.68, 0.19], [0.61, 0.42], [0.5, 0.5], [0.48, 0.59], [0.5, 0.46], [0.48, 0.55], [0.45, 0.47], [0.47, 0.59], [0.5, 0.33]]}" + } + ], + "images": "10406_exterior_image_2_left#0", + "gt": "[[275, 26], [216, 35], [196, 76], [160, 90], [155, 106], [160, 83], [154, 99], [143, 85], [149, 106], [159, 60]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0023_user_0010_scene_0007_cfg_0005_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the power strip by pressing the button', and now you are performing subtask 'turn on the power strip by pressing the button.'\nStarting at [0.48, 0.14], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.14], [0.44, 0.21], [0.47, 0.17], [0.49, 0.17], [0.52, 0.23], [0.48, 0.26], [0.49, 0.18], [0.51, 0.18], [0.51, 0.2], [0.47, 0.21]]}" + } + ], + "images": "RH20T_cfg5_task_0023_user_0010_scene_0007_cfg_0005_104122061850#0", + "gt": "[[305, 50], [283, 75], [299, 61], [315, 60], [330, 82], [306, 94], [314, 65], [329, 66], [326, 72], [298, 74]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0037_user_0014_scene_0001_cfg_0007_104122063633#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line', and now you are performing subtask 'place the first block at the stacking position, pick up the second block from the table, transfer the second block from the table to the stacking position, place the second block on top of the first block.'\nStart at [0.34, 0.4]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.34, 0.4], [0.32, 0.41], [0.31, 0.43], [0.31, 0.45], [0.31, 0.42], [0.32, 0.38], [0.33, 0.37], [0.34, 0.36], [0.35, 0.38], [0.35, 0.38]]}" + } + ], + "images": "RH20T_cfg7_task_0037_user_0014_scene_0001_cfg_0007_104122063633#187", + "gt": "[[220, 145], [206, 149], [201, 154], [198, 163], [200, 152], [206, 138], [211, 132], [217, 129], [224, 136], [224, 136]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "29717_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the clothes on the chair back', and now you are performing subtask 'put the clothes on the chair back.'\nYour starting position is [0.65, 0.67]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.67], [0.61, 0.75], [0.6, 0.92], [0.58, 0.57], [0.62, 0.32], [0.64, 0.55], [0.72, 0.54], [0.79, 0.73], [0.69, 0.48], [0.69, 0.48]]}" + } + ], + "images": "29717_exterior_image_1_left#0", + "gt": "[[208, 121], [196, 135], [193, 166], [185, 103], [198, 58], [206, 99], [229, 98], [253, 132], [222, 86], [222, 86]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "28295_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the second switch from the right on the adapter', and now you are performing subtask 'press the second switch on the right side of the adapter.'\nFrom start point [0.4, 0.63], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.63], [0.4, 0.63], [0.4, 0.65], [0.4, 0.64], [0.39, 0.64], [0.39, 0.63], [0.39, 0.63], [0.38, 0.64], [0.38, 0.63], [0.38, 0.63]]}" + } + ], + "images": "28295_exterior_image_2_left#112", + "gt": "[[127, 113], [128, 114], [127, 117], [128, 115], [125, 116], [125, 114], [124, 114], [121, 116], [120, 113], [120, 113]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_039422060546#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'fold the second lid on the box on the table.'\nYou begin at [0.28, 0.57]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.28, 0.57], [0.36, 0.49], [0.36, 0.48], [0.38, 0.47], [0.42, 0.44], [0.43, 0.45], [0.44, 0.5], [0.41, 0.46], [0.31, 0.5], [0.32, 0.53]]}" + } + ], + "images": "RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_039422060546#91", + "gt": "[[182, 204], [230, 177], [233, 172], [246, 169], [266, 157], [272, 162], [279, 181], [264, 164], [201, 179], [202, 191]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0057_user_0014_scene_0008_cfg_0006_104122061018#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn on the power strips', and now you are performing subtask 'press the fourth button on the power strip.'\nFrom start point [0.41, 0.31], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.31], [0.41, 0.32], [0.42, 0.31], [0.43, 0.27], [0.46, 0.21], [0.5, 0.19], [0.51, 0.19], [0.51, 0.21], [0.5, 0.22], [0.51, 0.22]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0057_user_0014_scene_0008_cfg_0006_104122061018#224.jpg", + "gt": "[[265, 110], [263, 115], [268, 111], [273, 96], [296, 74], [317, 68], [328, 68], [327, 76], [323, 80], [326, 79]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "19614_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the orange doll to the right, then put it back in the original position, and then move the green towel to the left', and now you are performing subtask 'slide the green cloth to the left.'\nYour starting position is [0.47, 0.32]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.32], [0.51, 0.34], [0.5, 0.42], [0.45, 0.32], [0.38, 0.27], [0.35, 0.3], [0.35, 0.37], [0.35, 0.34], [0.31, 0.27], [0.31, 0.22]]}" + } + ], + "images": "19614_exterior_image_1_left#133", + "gt": "[[152, 58], [162, 62], [161, 75], [145, 58], [121, 48], [112, 54], [113, 66], [113, 62], [99, 49], [100, 39]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "67887_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'arrange the cushions on the sofa', and now you are performing subtask 'pick up the second sofa cushion on the sofa.'\nStarting at [0.64, 0.89], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.64, 0.89], [0.61, 0.94], [0.63, 0.92], [0.69, 0.71], [0.67, 0.65], [0.6, 0.73], [0.61, 0.84], [0.62, 0.74], [0.58, 0.64], [0.5, 0.7]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/67887_exterior_image_2_left#327.jpg", + "gt": "[[206, 161], [194, 169], [202, 166], [222, 128], [214, 117], [193, 132], [195, 151], [200, 134], [187, 116], [160, 126]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_037522061512#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'place the object on the tabletop, pick up the cube in the box, transfer the cube from the box to the tabletop, place the cube on the tabletop.'\nWith start location [0.43, 0.29], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.29], [0.4, 0.26], [0.38, 0.3], [0.37, 0.38], [0.38, 0.43], [0.38, 0.3], [0.41, 0.26], [0.44, 0.32], [0.42, 0.27], [0.37, 0.23]]}" + } + ], + "images": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_037522061512#686", + "gt": "[[276, 106], [258, 94], [242, 109], [235, 136], [245, 153], [243, 108], [264, 94], [281, 117], [267, 97], [239, 82]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "27562_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the towel then remove the object from the drawer', and now you are performing subtask 'pick up the object from the drawer.'\nYou start from [0.32, 0.0]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.0], [0.56, 0.4], [0.62, 0.6], [0.62, 0.42], [0.61, 0.37], [0.56, 0.37], [0.53, 0.45], [0.52, 0.32], [0.5, 0.27], [0.46, 0.24]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/27562_exterior_image_1_left#0.jpg", + "gt": "[[102, 0], [179, 72], [197, 108], [200, 76], [196, 66], [178, 66], [169, 81], [165, 57], [159, 49], [147, 44]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "43225_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the cloth', and now you are performing subtask 'fold the cloth.'\nYour starting position is [0.7, 0.53]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.53], [0.48, 0.47], [0.33, 0.52], [0.47, 0.26], [0.57, 0.41], [0.49, 0.45], [0.46, 0.4], [0.44, 0.44], [0.47, 0.49], [0.44, 0.45]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/43225_exterior_image_2_left#0.jpg", + "gt": "[[223, 96], [155, 84], [107, 94], [149, 46], [183, 73], [157, 81], [146, 72], [141, 80], [151, 89], [142, 81]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0038_user_0015_scene_0010_cfg_0001_035622060973#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the cup', and now you are performing subtask 'pick up the cup.'\nStarting at [0.38, 0.36], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.36], [0.38, 0.38], [0.36, 0.42], [0.38, 0.4], [0.39, 0.47], [0.39, 0.47], [0.39, 0.4], [0.41, 0.26], [0.39, 0.21], [0.39, 0.22]]}" + } + ], + "images": "RH20T_cfg1_task_0038_user_0015_scene_0010_cfg_0001_035622060973#0", + "gt": "[[246, 130], [246, 135], [228, 151], [244, 144], [251, 168], [248, 169], [250, 143], [260, 94], [252, 75], [250, 81]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "3657_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push the box into the cabinet', and now you are performing subtask 'push the box into the cabinet.'\nThe trajectory starts at [0.71, 0.59]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.71, 0.59], [0.66, 0.66], [0.65, 0.67], [0.62, 0.66], [0.59, 0.73], [0.54, 0.74], [0.48, 0.72], [0.48, 0.71], [0.57, 0.73], [0.62, 0.7]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/3657_exterior_image_1_left#0.jpg", + "gt": "[[226, 106], [212, 118], [207, 120], [199, 118], [189, 132], [172, 133], [155, 130], [154, 128], [183, 131], [199, 126]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "16617_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the cover off the can and put it on the table', and now you are performing subtask 'take the cover off the can and put it on the table.'\nFrom start point [0.62, 0.0], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.0], [0.62, 0.04], [0.58, 0.13], [0.58, 0.17], [0.59, 0.04], [0.58, 0.03], [0.5, 0.09], [0.44, 0.21], [0.44, 0.29], [0.46, 0.18]]}" + } + ], + "images": "16617_exterior_image_2_left#0", + "gt": "[[198, 0], [197, 8], [187, 23], [187, 30], [188, 7], [185, 6], [161, 16], [140, 37], [142, 52], [146, 33]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "62326_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the right door of the cabinet', and now you are performing subtask 'close the right door of the cabinet.'\nWith start location [0.3, 0.03], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.3, 0.03], [0.54, 0.76], [0.55, 0.77], [0.57, 0.75], [0.57, 0.71], [0.57, 0.62], [0.5, 0.71], [0.52, 0.71], [0.51, 0.71], [0.51, 0.71]]}" + } + ], + "images": "62326_exterior_image_1_left#0", + "gt": "[[95, 5], [174, 137], [177, 138], [184, 135], [182, 127], [184, 111], [160, 128], [168, 127], [163, 128], [163, 128]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0030_user_0015_scene_0001_cfg_0001_039422060546#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the lid of the gift box on the gift box', and now you are performing subtask 'place the lid of the gift box on the gift box.'\nYour starting position is [0.41, 0.34]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.34], [0.46, 0.32], [0.53, 0.12], [0.56, 0.15], [0.58, 0.12], [0.58, 0.16], [0.6, 0.1], [0.62, 0.2], [0.59, 0.0], [0.59, 0.0]]}" + } + ], + "images": "RH20T_cfg1_task_0030_user_0015_scene_0001_cfg_0001_039422060546#0", + "gt": "[[265, 124], [292, 114], [339, 45], [358, 54], [371, 42], [372, 56], [385, 37], [398, 72], [377, 0], [377, 0]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0015_user_0014_scene_0008_cfg_0007_104122060811#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'play the drum', and now you are performing subtask 'play the drum.'\nYou start from [0.6, 0.36]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.36], [0.62, 0.32], [0.61, 0.36], [0.58, 0.3], [0.55, 0.31], [0.54, 0.34], [0.55, 0.32], [0.56, 0.32], [0.57, 0.38], [0.6, 0.34]]}" + } + ], + "images": "RH20T_cfg7_task_0015_user_0014_scene_0008_cfg_0007_104122060811#0", + "gt": "[[387, 130], [394, 114], [392, 128], [372, 109], [353, 113], [344, 123], [353, 114], [361, 115], [366, 138], [381, 122]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0051_user_0014_scene_0002_cfg_0006_104122061330#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the dish off the dish rack', and now you are performing subtask 'take the dish off the dish rack.'\nYou begin at [0.42, 0.27]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.27], [0.45, 0.44], [0.43, 0.65], [0.42, 0.46], [0.45, 0.34], [0.52, 0.28], [0.59, 0.34], [0.57, 0.26], [0.52, 0.33], [0.54, 0.25]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0051_user_0014_scene_0002_cfg_0006_104122061330#0.jpg", + "gt": "[[266, 97], [285, 159], [272, 233], [266, 166], [286, 122], [335, 100], [375, 124], [367, 95], [332, 118], [345, 91]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0057_user_0010_scene_0006_cfg_0003_104122062823#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn on the power strips', and now you are performing subtask 'press the button on the left power strip.'\nStart at [0.19, 0.14]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.19, 0.14], [0.2, 0.15], [0.19, 0.16], [0.17, 0.12], [0.14, 0.03], [0.14, 0.02], [0.16, 0.0], [0.28, 0.0], [0.3, 0.08], [0.29, 0.08]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg3_task_0057_user_0010_scene_0006_cfg_0003_104122062823#144.jpg", + "gt": "[[120, 52], [126, 53], [124, 59], [108, 43], [90, 12], [91, 6], [102, 0], [181, 0], [195, 29], [186, 29]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0064_user_0001_scene_0008_cfg_0001_038522063145#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the roll of paper down from the shelf', and now you are performing subtask 'take the roll of paper down from the shelf.'\nYour starting position is [0.69, 0.28]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.28], [0.77, 0.23], [0.8, 0.29], [0.81, 0.36], [0.79, 0.36], [0.77, 0.46], [0.76, 0.42], [0.7, 0.3], [0.7, 0.32], [0.73, 0.23]]}" + } + ], + "images": "RH20T_cfg1_task_0064_user_0001_scene_0008_cfg_0001_038522063145#0", + "gt": "[[444, 99], [492, 83], [511, 104], [518, 128], [505, 131], [492, 166], [488, 151], [451, 108], [450, 117], [470, 83]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0104_user_0007_scene_0007_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'drag the plate back after holding it down', and now you are performing subtask 'drag the plate back after holding it down.'\nWith start location [0.5, 0.37], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.57, 0.41], [0.58, 0.66], [0.59, 0.69], [0.58, 0.7], [0.58, 0.73], [0.57, 0.69], [0.57, 0.72], [0.57, 0.62], [0.6, 0.58]]}" + } + ], + "images": "RH20T_cfg5_task_0104_user_0007_scene_0007_cfg_0005_104122063678#0", + "gt": "[[323, 132], [364, 146], [369, 236], [380, 247], [371, 252], [372, 262], [365, 248], [366, 258], [366, 224], [385, 210]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "91549_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the towel then move the pot to the right', and now you are performing subtask 'place the pot on the specified position on the surface.'\nYou start from [0.27, 0.52]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.27, 0.52], [0.24, 0.6], [0.26, 0.58], [0.27, 0.61], [0.22, 0.54], [0.26, 0.49], [0.32, 0.5], [0.36, 0.55], [0.35, 0.7], [0.36, 0.55]]}" + } + ], + "images": "91549_exterior_image_1_left#163", + "gt": "[[86.5, 93.5], [76.5, 108.5], [82.5, 104.5], [87.5, 109.5], [70.5, 97.5], [82.5, 87.5], [103.5, 89.5], [116.5, 98.5], [113.5, 125.5], [116.5, 98.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_104122063633#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the third block on the table, transfer the third block from the table to above the first block, place the third block on top of the first block, pick up the fourth block on the table.'\nYou begin at [0.37, 0.4]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.4], [0.4, 0.47], [0.35, 0.56], [0.35, 0.6], [0.38, 0.49], [0.42, 0.44], [0.41, 0.33], [0.39, 0.36], [0.38, 0.31], [0.35, 0.36]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_104122063633#622.jpg", + "gt": "[[237, 144], [254, 171], [221, 200], [224, 216], [242, 178], [269, 159], [263, 119], [251, 128], [241, 112], [222, 131]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "88538_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'folding the towel', and now you are performing subtask 'fold the towel on the desk.'\nStarting at [0.62, 0.54], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.54], [0.59, 0.49], [0.62, 0.56], [0.63, 0.67], [0.62, 0.66], [0.59, 0.46], [0.54, 0.45], [0.53, 0.41], [0.55, 0.37], [0.55, 0.41]]}" + } + ], + "images": "88538_exterior_image_2_left#163", + "gt": "[[197, 98], [189, 88], [199, 101], [202, 120], [197, 119], [190, 82], [172, 81], [169, 73], [175, 67], [176, 73]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0007_user_0020_scene_0004_cfg_0007_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'holding a small block with the gripper and sweeping it from left to right on the table', and now you are performing subtask 'holding a small block with the gripper and sweeping it from left to right on the table.'\nYour starting position is [0.53, 0.32]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.32], [0.55, 0.6], [0.57, 0.68], [0.59, 0.7], [0.59, 0.74], [0.6, 0.43], [0.59, 0.36], [0.61, 0.31], [0.59, 0.32], [0.59, 0.32]]}" + } + ], + "images": "RH20T_cfg7_task_0007_user_0020_scene_0004_cfg_0007_104122064161#0", + "gt": "[[337, 115], [352, 216], [368, 245], [380, 251], [378, 268], [387, 155], [380, 128], [388, 112], [377, 117], [377, 117]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "55040_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'throw the yellow square into the silver pot', and now you are performing subtask 'throw the yellow square into the silver pot.'\nYou begin at [0.79, 0.01]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.79, 0.01], [0.74, 0.26], [0.66, 0.47], [0.63, 0.51], [0.57, 0.51], [0.6, 0.57], [0.59, 0.61], [0.62, 0.55], [0.67, 0.59], [0.63, 0.57]]}" + } + ], + "images": "55040_exterior_image_2_left#0", + "gt": "[[252, 2], [236, 47], [210, 85], [201, 92], [183, 91], [191, 102], [188, 109], [200, 99], [213, 106], [202, 102]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0046_user_0014_scene_0001_cfg_0007_104122061018#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the cup on the cup rack', and now you are performing subtask 'put the cup on the cup rack.'\nStarting at [0.57, 0.08], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.08], [0.51, 0.21], [0.51, 0.35], [0.57, 0.29], [0.72, 0.25], [0.8, 0.29], [0.85, 0.34], [0.85, 0.38], [0.74, 0.3], [0.66, 0.14]]}" + } + ], + "images": "RH20T_cfg7_task_0046_user_0014_scene_0001_cfg_0007_104122061018#0", + "gt": "[[363, 30], [329, 76], [328, 125], [363, 104], [460, 89], [513, 104], [545, 123], [544, 135], [472, 108], [425, 52]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "26555_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the cup from the right side of the table to the left side of the table', and now you are performing subtask 'move the cup from the right side of the table to the left side of the table.'\nStart at [0.77, 0.34]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.77, 0.34], [0.74, 0.35], [0.73, 0.38], [0.74, 0.47], [0.7, 0.62], [0.67, 0.56], [0.67, 0.49], [0.65, 0.53], [0.64, 0.53], [0.63, 0.49]]}" + } + ], + "images": "26555_exterior_image_1_left#0", + "gt": "[[247.5, 62.0], [237.5, 63.0], [232.5, 68.0], [237.5, 84.0], [223.5, 111.0], [213.5, 101.0], [213.5, 89.0], [207.5, 96.0], [203.5, 95.0], [202.5, 89.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0035_user_0017_scene_0005_cfg_0004_038522062288#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up a small block', and now you are performing subtask 'pick up a small block.'\nThe trajectory starts at [0.36, 0.33]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.33], [0.35, 0.32], [0.35, 0.3], [0.39, 0.28], [0.36, 0.3], [0.38, 0.32], [0.38, 0.47], [0.41, 0.54], [0.44, 0.56], [0.36, 0.32]]}" + } + ], + "images": "RH20T_cfg4_task_0035_user_0017_scene_0005_cfg_0004_038522062288#0", + "gt": "[[233, 119], [223, 114], [224, 109], [251, 102], [232, 108], [240, 117], [246, 171], [262, 193], [281, 202], [232, 116]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0052_user_0007_scene_0010_cfg_0005_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grab a basketball, release it and shoot it into the basket', and now you are performing subtask 'grab a basketball, release it and shoot it into the basket.'\nYour initial coordinate is [0.48, 0.14]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.14], [0.53, 0.13], [0.6, 0.35], [0.53, 0.24], [0.4, 0.18], [0.27, 0.3], [0.27, 0.39], [0.3, 0.45], [0.43, 0.19], [0.46, 0.08]]}" + } + ], + "images": "RH20T_cfg5_task_0052_user_0007_scene_0010_cfg_0005_104122061850#0", + "gt": "[[305, 50], [341, 46], [384, 125], [341, 85], [256, 64], [172, 108], [174, 140], [192, 161], [275, 70], [292, 28]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "21793_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn the caps on the bed', and now you are performing subtask 'pick up the black hat on the bed.'\nYou start from [0.55, 0.29]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.29], [0.46, 0.41], [0.29, 0.91], [0.24, 0.92], [0.28, 0.88], [0.26, 0.79], [0.25, 0.88], [0.29, 0.86], [0.28, 0.67], [0.17, 0.84]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/21793_exterior_image_1_left#0.jpg", + "gt": "[[176, 52], [146, 73], [94, 163], [78, 165], [90, 159], [82, 142], [81, 159], [94, 155], [88, 121], [54, 152]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "16095_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pour the contents in the red bowl into the silver bowl', and now you are performing subtask 'pour the contents in the red bowl into the silver bowl.'\nWith start location [0.42, 0.0], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.0], [0.45, 0.57], [0.45, 0.49], [0.54, 0.41], [0.67, 0.4], [0.61, 0.32], [0.43, 0.52], [0.49, 0.54], [0.47, 0.34], [0.51, 0.23]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/16095_exterior_image_1_left#0.jpg", + "gt": "[[135, 0], [143, 102], [145, 89], [174, 73], [213, 72], [196, 57], [138, 93], [156, 98], [150, 62], [163, 41]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0015_user_0001_scene_0007_cfg_0001_035622060973#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'play the drum', and now you are performing subtask 'play the drum.'\nThe trajectory starts at [0.44, 0.39]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.39], [0.32, 0.49], [0.32, 0.54], [0.29, 0.61], [0.35, 0.47], [0.44, 0.39], [0.47, 0.41], [0.33, 0.45], [0.32, 0.49], [0.34, 0.47]]}" + } + ], + "images": "RH20T_cfg1_task_0015_user_0001_scene_0007_cfg_0001_035622060973#0", + "gt": "[[279, 141], [205, 176], [207, 196], [184, 221], [223, 169], [280, 141], [304, 149], [212, 161], [204, 175], [220, 171]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "6951_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unhang the cloth and put it on the table', and now you are performing subtask 'unhang the cloth and put it on the table.'\nYour starting position is [0.63, 0.02]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.02], [0.55, 0.25], [0.51, 0.31], [0.46, 0.32], [0.45, 0.34], [0.49, 0.29], [0.52, 0.28], [0.6, 0.36], [0.61, 0.42], [0.62, 0.42]]}" + } + ], + "images": "6951_exterior_image_1_left#0", + "gt": "[[201, 4], [177, 45], [163, 55], [147, 57], [145, 62], [156, 52], [165, 51], [191, 64], [196, 75], [199, 75]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "35690_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the black and yellow bottle to the left, move the water bottle to the right, then move the black and yellow bottle closer to the water bottle', and now you are performing subtask 'push the black and yellow bottle next to the water bottle.'\nThe trajectory starts at [0.44, 0.64]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.64], [0.47, 0.51], [0.51, 0.47], [0.59, 0.51], [0.61, 0.57], [0.56, 0.59], [0.64, 0.58], [0.61, 0.58], [0.63, 0.42], [0.55, 0.46]]}" + } + ], + "images": "35690_exterior_image_1_left#262", + "gt": "[[142, 115], [149, 91], [163, 85], [188, 92], [195, 102], [178, 106], [204, 104], [196, 105], [201, 75], [176, 82]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "265_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the bowl to the right and put the rubik's cube inside it', and now you are performing subtask 'place the bowl in the right side area.'\nStarting at [0.2, 0.17], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.2, 0.17], [0.35, 0.31], [0.4, 0.28], [0.43, 0.4], [0.46, 0.41], [0.46, 0.47], [0.49, 0.38], [0.46, 0.45], [0.48, 0.43], [0.51, 0.49]]}" + } + ], + "images": "265_exterior_image_2_left#0", + "gt": "[[63, 30], [111, 56], [127, 51], [139, 72], [148, 73], [146, 85], [156, 69], [147, 81], [153, 77], [162, 88]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0016_user_0014_scene_0001_cfg_0001_035622060973#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'hit the pool ball', and now you are performing subtask 'pick up the pool ball from the surface.'\nStart at [0.38, 0.36]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.36], [0.38, 0.42], [0.4, 0.46], [0.43, 0.63], [0.45, 0.69], [0.49, 0.82], [0.48, 0.82], [0.46, 0.81], [0.47, 0.82], [0.46, 0.82]]}" + } + ], + "images": "RH20T_cfg1_task_0016_user_0014_scene_0001_cfg_0001_035622060973#0", + "gt": "[[246, 130], [242, 150], [257, 167], [275, 228], [291, 249], [314, 294], [307, 297], [296, 293], [298, 295], [292, 297]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "92163_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the towel', and now you are performing subtask 'unfold the towel on the table.'\nThe trajectory starts at [0.37, 0.06]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.06], [0.53, 0.21], [0.53, 0.37], [0.46, 0.4], [0.55, 0.43], [0.5, 0.54], [0.54, 0.61], [0.59, 0.54], [0.37, 0.47], [0.45, 0.66]]}" + } + ], + "images": "92163_exterior_image_1_left#0", + "gt": "[[119, 10], [169, 37], [171, 67], [146, 72], [176, 78], [161, 98], [173, 110], [189, 97], [119, 85], [145, 118]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "46904_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the doll from the plate', and now you are performing subtask 'remove the doll from the plate.'\nThe trajectory starts at [0.99, 0.07]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.99, 0.07], [0.85, 0.19], [0.6, 0.45], [0.54, 0.6], [0.59, 0.52], [0.63, 0.52], [0.65, 0.57], [0.66, 0.6], [0.65, 0.61], [0.63, 0.52]]}" + } + ], + "images": "46904_exterior_image_2_left#0", + "gt": "[[317, 13], [273, 34], [192, 81], [173, 108], [190, 94], [201, 93], [209, 102], [210, 108], [207, 110], [203, 94]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0023_user_0014_scene_0002_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the power strip by pressing the button', and now you are performing subtask 'turn on the power strip by pressing the button.'\nYour initial coordinate is [0.63, 0.37]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.64, 0.36], [0.63, 0.36], [0.63, 0.37], [0.62, 0.35], [0.6, 0.34], [0.59, 0.38], [0.58, 0.38], [0.58, 0.39], [0.59, 0.36]]}" + } + ], + "images": "RH20T_cfg1_task_0023_user_0014_scene_0002_cfg_0001_038522062547#0", + "gt": "[[406, 134], [408, 130], [405, 130], [406, 133], [398, 125], [381, 123], [377, 137], [369, 136], [371, 141], [376, 129]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "41945_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'throw the bottle cap on the table', and now you are performing subtask 'throw the bottle cap on the table.'\nThe trajectory starts at [0.38, 0.16]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.16], [0.38, 0.15], [0.38, 0.16], [0.36, 0.17], [0.35, 0.16], [0.35, 0.18], [0.35, 0.17], [0.38, 0.18], [0.41, 0.27], [0.41, 0.27]]}" + } + ], + "images": "41945_exterior_image_2_left#0", + "gt": "[[120, 29], [121, 27], [120, 29], [115, 30], [113, 29], [111, 32], [111, 31], [122, 32], [131, 48], [131, 48]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "79222_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the carrot plushie in the pan', and now you are performing subtask 'put the carrot plushie in the pan.'\nYou begin at [0.6, 0.01]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.01], [0.55, 0.22], [0.49, 0.31], [0.48, 0.37], [0.47, 0.33], [0.43, 0.31], [0.41, 0.34], [0.41, 0.38], [0.42, 0.31], [0.41, 0.32]]}" + } + ], + "images": "79222_exterior_image_2_left#0", + "gt": "[[193, 1], [176, 39], [157, 56], [153, 66], [149, 60], [136, 56], [131, 61], [131, 68], [135, 55], [132, 57]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "1158_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull out a tissue from the box and put it on the table', and now you are performing subtask 'pull out a tissue from the box and put it on the table.'\nStarting at [0.49, 0.36], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.36], [0.59, 0.56], [0.61, 0.56], [0.58, 0.56], [0.54, 0.59], [0.55, 0.61], [0.59, 0.51], [0.6, 0.6], [0.57, 0.7], [0.59, 0.66]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/1158_exterior_image_2_left#0.jpg", + "gt": "[[157, 65], [188, 100], [194, 101], [186, 100], [174, 107], [175, 110], [189, 92], [191, 108], [182, 126], [188, 119]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "15888_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the cup to the left', and now you are performing subtask 'move the cup to the left.'\nYour initial coordinate is [0.68, 0.0]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.0], [0.61, 0.07], [0.6, 0.12], [0.52, 0.16], [0.55, 0.16], [0.52, 0.18], [0.56, 0.28], [0.53, 0.37], [0.42, 0.43], [0.42, 0.33]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/15888_exterior_image_1_left#0.jpg", + "gt": "[[217, 0], [195, 13], [193, 22], [168, 28], [175, 28], [168, 32], [180, 50], [171, 67], [133, 78], [135, 60]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "89189_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the faucet spout to the left and move the doll forward', and now you are performing subtask 'pick up doll from countertop.'\nYour initial coordinate is [0.55, 0.24]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.24], [0.54, 0.3], [0.55, 0.34], [0.56, 0.33], [0.59, 0.38], [0.63, 0.39], [0.61, 0.32], [0.57, 0.26], [0.53, 0.24], [0.48, 0.27]]}" + } + ], + "images": "89189_exterior_image_1_left#108", + "gt": "[[176, 44], [173, 54], [176, 62], [180, 60], [188, 69], [203, 70], [196, 57], [182, 46], [170, 44], [154, 49]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "81550_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the mug from the coffee maker and put it on the counter, and pour the content from the transparent cup into the mug', and now you are performing subtask 'pick up the transparent cup from the countertop.'\nThe trajectory starts at [0.34, 0.56]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.34, 0.56], [0.36, 0.52], [0.4, 0.56], [0.38, 0.59], [0.38, 0.54], [0.3, 0.51], [0.31, 0.48], [0.34, 0.53], [0.32, 0.49], [0.34, 0.47]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/81550_exterior_image_2_left#0.jpg", + "gt": "[[108, 100], [115, 94], [127, 101], [123, 107], [120, 97], [95, 91], [99, 87], [108, 96], [102, 89], [110, 85]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "62555_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick the marker and put it in the cup', and now you are performing subtask 'pick the marker and put it in the cup.'\nWith start location [0.44, 0.24], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.24], [0.53, 0.24], [0.53, 0.48], [0.55, 0.42], [0.57, 0.43], [0.55, 0.49], [0.58, 0.53], [0.57, 0.57], [0.58, 0.46], [0.62, 0.39]]}" + } + ], + "images": "62555_exterior_image_1_left#0", + "gt": "[[141, 44], [169, 44], [171, 87], [177, 76], [184, 77], [175, 89], [185, 95], [182, 103], [185, 82], [198, 71]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0002_user_0001_scene_0006_cfg_0001_819112070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull out a napkin', and now you are performing subtask 'pull out a napkin.'\nStarting at [0.61, 0.3], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.61, 0.3], [0.58, 0.31], [0.57, 0.36], [0.53, 0.36], [0.52, 0.39], [0.55, 0.32], [0.56, 0.29], [0.57, 0.28], [0.58, 0.25], [0.64, 0.27]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0002_user_0001_scene_0006_cfg_0001_819112070044#0.jpg", + "gt": "[[389, 108], [371, 110], [363, 128], [338, 128], [332, 141], [354, 117], [358, 105], [367, 102], [369, 89], [410, 96]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "53731_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the glue stick out of the cup and place it on the black box', and now you are performing subtask 'take the glue stick out of the cup and place it on the black box.'\nYou begin at [1.0, 0.29]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[1.0, 0.29], [0.64, 0.46], [0.68, 0.5], [0.6, 0.47], [0.62, 0.5], [0.57, 0.54], [0.64, 0.41], [0.72, 0.45], [0.72, 0.52], [0.72, 0.44]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/53731_exterior_image_1_left#0.jpg", + "gt": "[[319.0, 51.5], [205.0, 83.5], [218.0, 89.5], [191.0, 85.5], [200.0, 90.5], [184.0, 97.5], [204.0, 73.5], [231.0, 80.5], [231.0, 94.5], [230.0, 78.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "12057_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the plush toy inside the open drawer', and now you are performing subtask 'put the plush toy inside the open drawer.'\nStarting at [0.68, 0.37], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.37], [0.62, 0.5], [0.62, 0.65], [0.51, 0.8], [0.45, 0.79], [0.61, 0.62], [0.65, 0.74], [0.61, 0.73], [0.55, 0.8], [0.52, 0.77]]}" + } + ], + "images": "12057_exterior_image_1_left#0", + "gt": "[[218, 67], [199, 90], [198, 117], [162, 144], [143, 142], [195, 112], [209, 133], [196, 131], [175, 144], [166, 138]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "86543_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the glass lid from the black pot', and now you are performing subtask 'remove the glass lid from the black pot.'\nWith start location [0.63, 0.09], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.09], [0.51, 0.38], [0.5, 0.63], [0.5, 0.58], [0.45, 0.51], [0.35, 0.49], [0.25, 0.58], [0.27, 0.74], [0.29, 0.73], [0.29, 0.73]]}" + } + ], + "images": "86543_exterior_image_2_left#0", + "gt": "[[201, 17], [162, 68], [159, 113], [160, 104], [144, 91], [111, 88], [80, 104], [85, 133], [93, 132], [93, 132]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0010_user_0010_scene_0007_cfg_0005_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the handset of the telephone on the corresponding phone cradle', and now you are performing subtask 'place the handset of the telephone on the corresponding phone cradle.'\nYou begin at [0.48, 0.14]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.14], [0.48, 0.19], [0.47, 0.26], [0.45, 0.26], [0.48, 0.31], [0.48, 0.38], [0.4, 0.45], [0.41, 0.52], [0.4, 0.36], [0.45, 0.31]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0010_user_0010_scene_0007_cfg_0005_104122061850#0.jpg", + "gt": "[[305, 50], [310, 68], [302, 92], [290, 92], [308, 110], [306, 138], [258, 163], [261, 187], [253, 130], [291, 110]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0025_user_0020_scene_0008_cfg_0007_104122061602#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold a piece of paper that has been folded once (one end is pressed down by a weight)', and now you are performing subtask 'unfold a piece of paper that has been folded once (one end is pressed down by a weight).'\nYou start from [0.7, 0.32]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.32], [0.69, 0.44], [0.71, 0.48], [0.71, 0.52], [0.69, 0.41], [0.64, 0.33], [0.59, 0.31], [0.6, 0.29], [0.64, 0.38], [0.64, 0.28]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0025_user_0020_scene_0008_cfg_0007_104122061602#0.jpg", + "gt": "[[448, 116], [442, 160], [454, 173], [452, 188], [440, 147], [410, 120], [376, 113], [382, 103], [411, 138], [407, 101]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0034_user_0003_scene_0006_cfg_0002_037522062165#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'transfer the fourth square from the countertop to the first and third squares, place the fourth square on the first and third squares, slide the fifth square towards the middle of the countertop, pick up the fifth square from the countertop, transfer the fifth square from the countertop to the first and second squares.'\nThe trajectory starts at [0.25, 0.5]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.25, 0.5], [0.28, 0.46], [0.24, 0.47], [0.19, 0.55], [0.19, 0.6], [0.17, 0.65], [0.19, 0.65], [0.24, 0.61], [0.25, 0.63], [0.3, 0.5]]}" + } + ], + "images": "RH20T_cfg2_task_0034_user_0003_scene_0006_cfg_0002_037522062165#756", + "gt": "[[157, 180], [176, 164], [154, 170], [121, 198], [124, 216], [111, 234], [119, 235], [155, 219], [163, 228], [191, 181]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "13502_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the faucet, turn it off, and then turn it on again', and now you are performing subtask 'press the faucet handle to turn off.'\nThe trajectory starts at [0.51, 0.4]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.4], [0.5, 0.41], [0.45, 0.34], [0.38, 0.32], [0.37, 0.34], [0.38, 0.34], [0.32, 0.36], [0.31, 0.43], [0.28, 0.4], [0.32, 0.42]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/13502_exterior_image_1_left#0.jpg", + "gt": "[[164, 72], [161, 73], [144, 62], [123, 58], [118, 61], [122, 61], [101, 65], [99, 77], [91, 72], [102, 76]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "90950_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the blue marker out of the white bowl', and now you are performing subtask 'take the blue marker out of the white bowl.'\nYour starting position is [0.28, 0.26]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.28, 0.26], [0.41, 0.23], [0.47, 0.27], [0.5, 0.33], [0.51, 0.46], [0.53, 0.47], [0.52, 0.49], [0.5, 0.42], [0.49, 0.61], [0.45, 0.46]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/90950_exterior_image_1_left#0.jpg", + "gt": "[[91, 46], [130, 42], [149, 49], [161, 59], [162, 83], [170, 84], [165, 88], [161, 76], [157, 109], [144, 83]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "51023_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the rectangular block on the table', and now you are performing subtask 'put the rectangular block on the table.'\nStarting at [0.61, 0.01], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.61, 0.01], [0.58, 0.2], [0.6, 0.32], [0.58, 0.35], [0.57, 0.26], [0.52, 0.23], [0.46, 0.31], [0.45, 0.37], [0.46, 0.22], [0.46, 0.22]]}" + } + ], + "images": "51023_exterior_image_2_left#0", + "gt": "[[196, 2], [185, 36], [193, 58], [186, 63], [183, 47], [165, 42], [147, 55], [143, 67], [146, 39], [146, 39]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0061_user_0001_scene_0006_cfg_0001_038522062547#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'place the large block on the table, pick up the small block from the gift box, transfer the small block from the gift box to the table, place the small block on the table.'\nThe trajectory starts at [0.83, 0.54]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.83, 0.54], [0.68, 0.43], [0.64, 0.45], [0.6, 0.41], [0.66, 0.42], [0.64, 0.46], [0.69, 0.34], [0.7, 0.64], [0.73, 0.57], [0.72, 0.46]]}" + } + ], + "images": "RH20T_cfg1_task_0061_user_0001_scene_0006_cfg_0001_038522062547#198", + "gt": "[[531, 193], [433, 154], [407, 163], [384, 149], [421, 151], [411, 166], [441, 124], [451, 232], [467, 204], [460, 166]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122060811#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the block on the other block, pick up the block on the surface, transfer the block from the surface to the other block, place the block on the other block.'\nStarting at [0.52, 0.44], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.44], [0.51, 0.44], [0.51, 0.46], [0.52, 0.44], [0.56, 0.44], [0.57, 0.45], [0.56, 0.4], [0.55, 0.38], [0.54, 0.36], [0.54, 0.36]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122060811#498.jpg", + "gt": "[[332, 157], [326, 158], [325, 166], [335, 159], [361, 157], [362, 162], [357, 143], [353, 136], [343, 131], [345, 129]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0038_user_0007_scene_0009_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the red cup from the cube to the table', and now you are performing subtask 'move the red cup from the cube to the table.'\nYour starting position is [0.5, 0.37]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.51, 0.42], [0.42, 0.38], [0.35, 0.39], [0.34, 0.4], [0.35, 0.47], [0.35, 0.41], [0.41, 0.37], [0.43, 0.39], [0.43, 0.39]]}" + } + ], + "images": "RH20T_cfg5_task_0038_user_0007_scene_0009_cfg_0005_104122063678#0", + "gt": "[[323, 132], [326, 152], [267, 135], [226, 139], [216, 145], [221, 169], [227, 147], [263, 134], [274, 142], [274, 142]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "56854_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the tissue on the table', and now you are performing subtask 'put the tissue on the table.'\nYour starting position is [0.16, 0.17]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.16, 0.17], [0.33, 0.25], [0.37, 0.42], [0.43, 0.4], [0.4, 0.46], [0.43, 0.29], [0.55, 0.3], [0.45, 0.81], [0.47, 0.41], [0.44, 0.27]]}" + } + ], + "images": "56854_exterior_image_2_left#0", + "gt": "[[52, 30], [106, 45], [117, 76], [136, 72], [128, 82], [138, 52], [175, 54], [143, 146], [152, 73], [141, 48]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0083_user_0014_scene_0003_cfg_0001_039422060546#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'swat with a flyswatter', and now you are performing subtask 'swat with a flyswatter.'\nThe trajectory starts at [0.42, 0.34]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.34], [0.47, 0.31], [0.51, 0.32], [0.58, 0.31], [0.6, 0.28], [0.61, 0.37], [0.58, 0.34], [0.59, 0.34], [0.57, 0.3], [0.54, 0.31]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0083_user_0014_scene_0003_cfg_0001_039422060546#0.jpg", + "gt": "[[266, 122], [301, 113], [327, 116], [373, 110], [383, 101], [391, 132], [373, 122], [380, 121], [365, 108], [348, 113]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0079_user_0015_scene_0003_cfg_0001_819112070044#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the microwave door', and now you are performing subtask 'pull the handle of the microwave door to open the door.'\nStart at [0.54, 0.51]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.51], [0.53, 0.46], [0.55, 0.42], [0.58, 0.49], [0.54, 0.41], [0.47, 0.44], [0.48, 0.46], [0.52, 0.39], [0.53, 0.33], [0.63, 0.28]]}" + } + ], + "images": "RH20T_cfg1_task_0079_user_0015_scene_0003_cfg_0001_819112070044#153", + "gt": "[[346, 182], [342, 166], [351, 152], [373, 175], [347, 146], [302, 159], [308, 165], [335, 142], [340, 120], [405, 99]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "60558_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the green marker out of the grey bowl', and now you are performing subtask 'take the green marker out of the grey bowl.'\nYour starting position is [0.02, 0.26]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.02, 0.26], [0.35, 0.44], [0.41, 0.54], [0.45, 0.56], [0.44, 0.56], [0.47, 0.6], [0.49, 0.68], [0.46, 0.52], [0.41, 0.46], [0.4, 0.51]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/60558_exterior_image_1_left#0.jpg", + "gt": "[[5.5, 46.5], [111.5, 78.5], [131.5, 96.5], [144.5, 100.5], [139.5, 101.5], [151.5, 107.5], [156.5, 122.5], [148.5, 94.5], [130.5, 83.5], [126.5, 92.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "27445_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the door of the cabinet', and now you are performing subtask 'close the door of the cabinet.'\nStarting at [0.5, 0.38], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.38], [0.49, 0.39], [0.49, 0.38], [0.46, 0.37], [0.43, 0.44], [0.38, 0.51], [0.36, 0.48], [0.21, 0.47], [0.28, 0.46], [0.34, 0.43]]}" + } + ], + "images": "27445_exterior_image_1_left#0", + "gt": "[[161, 68], [157, 71], [156, 68], [147, 67], [137, 80], [122, 91], [115, 87], [66, 84], [91, 83], [109, 78]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "90367_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take up the marker', and now you are performing subtask 'take up the marker.'\nStarting at [0.7, 0.0], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.0], [0.63, 0.11], [0.61, 0.13], [0.63, 0.13], [0.57, 0.2], [0.55, 0.24], [0.54, 0.28], [0.55, 0.33], [0.55, 0.36], [0.52, 0.39]]}" + } + ], + "images": "90367_exterior_image_1_left#0", + "gt": "[[225, 0], [203, 19], [195, 24], [203, 24], [182, 36], [175, 43], [174, 50], [177, 59], [177, 64], [167, 70]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0001_user_0010_scene_0001_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the button from top to bottom', and now you are performing subtask 'press the button from top to bottom.'\nFrom start point [0.5, 0.37], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.52, 0.41], [0.52, 0.43], [0.51, 0.54], [0.5, 0.59], [0.49, 0.61], [0.49, 0.62], [0.49, 0.62], [0.52, 0.54], [0.52, 0.52]]}" + } + ], + "images": "RH20T_cfg5_task_0001_user_0010_scene_0001_cfg_0005_104122063678#0", + "gt": "[[323, 132], [332, 147], [332, 156], [327, 195], [320, 213], [313, 220], [316, 223], [313, 223], [331, 195], [331, 187]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0083_user_0007_scene_0005_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'swat with a flyswatter', and now you are performing subtask 'swat with a flyswatter.'\nYour starting position is [0.5, 0.37]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.5, 0.35], [0.52, 0.34], [0.53, 0.38], [0.57, 0.37], [0.57, 0.41], [0.52, 0.33], [0.49, 0.35], [0.49, 0.32], [0.53, 0.42]]}" + } + ], + "images": "RH20T_cfg5_task_0083_user_0007_scene_0005_cfg_0005_104122063678#0", + "gt": "[[323, 132], [321, 126], [335, 123], [339, 136], [362, 134], [364, 147], [335, 119], [315, 125], [311, 114], [337, 151]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0069_user_0015_scene_0001_cfg_0004_104122062823#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'assemble one piece of a puzzle', and now you are performing subtask 'pick up the puzzle piece from the table, transfer the puzzle piece from the table to the right side of the puzzle board, place the puzzle piece on the puzzle board, push the puzzle piece to the left.'\nThe trajectory starts at [0.61, 0.2]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.61, 0.2], [0.63, 0.23], [0.63, 0.16], [0.61, 0.16], [0.6, 0.24], [0.55, 0.25], [0.55, 0.31], [0.55, 0.24], [0.58, 0.33], [0.58, 0.33]]}" + } + ], + "images": "RH20T_cfg4_task_0069_user_0015_scene_0001_cfg_0004_104122062823#511", + "gt": "[[388, 72], [406, 83], [405, 58], [392, 56], [381, 85], [352, 91], [351, 111], [353, 85], [369, 119], [369, 119]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0201_user_0015_scene_0007_cfg_0001_750612070853#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'the medium pipette tip is inserted into the holder for medium pipette tips', and now you are performing subtask 'place the tip of the medium pipette into the holder for medium pipette tips, place the tip of the medium pipette into the holder for medium pipette tips.'\nFrom start point [0.63, 0.36], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.36], [0.64, 0.35], [0.63, 0.34], [0.63, 0.33], [0.63, 0.29], [0.64, 0.29], [0.65, 0.32], [0.66, 0.31], [0.65, 0.34], [0.65, 0.33]]}" + } + ], + "images": "RH20T_cfg1_task_0201_user_0015_scene_0007_cfg_0001_750612070853#209", + "gt": "[[406, 131], [407, 127], [402, 123], [401, 119], [404, 103], [407, 103], [413, 116], [421, 113], [413, 122], [413, 119]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0018_user_0014_scene_0007_cfg_0006_104122061602#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'sliding the Lego blocks on the desktop', and now you are performing subtask 'sliding the Lego blocks on the desktop.'\nStart at [0.7, 0.29]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.29], [0.69, 0.47], [0.65, 0.54], [0.66, 0.57], [0.68, 0.57], [0.69, 0.53], [0.69, 0.5], [0.71, 0.47], [0.69, 0.44], [0.67, 0.44]]}" + } + ], + "images": "RH20T_cfg6_task_0018_user_0014_scene_0007_cfg_0006_104122061602#0", + "gt": "[[446, 106], [439, 171], [418, 196], [423, 204], [432, 204], [441, 191], [442, 181], [452, 169], [444, 160], [427, 160]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0032_user_0016_scene_0002_cfg_0004_104122062295#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'slide the outer casing onto the gift box', and now you are performing subtask 'slide the outer casing onto the gift box.'\nYour initial coordinate is [0.33, 0.56]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.33, 0.56], [0.31, 0.55], [0.26, 0.55], [0.26, 0.62], [0.27, 0.63], [0.28, 0.57], [0.3, 0.55], [0.34, 0.55], [0.42, 0.56], [0.45, 0.59]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0032_user_0016_scene_0002_cfg_0004_104122062295#173.jpg", + "gt": "[[214, 203], [196, 197], [169, 198], [169, 223], [172, 226], [179, 205], [194, 199], [215, 198], [271, 203], [291, 211]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0081_user_0020_scene_0009_cfg_0006_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the leg of the glasses (with one hand)', and now you are performing subtask 'fold the leg of the glasses outward.'\nWith start location [0.53, 0.31], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.31], [0.53, 0.41], [0.5, 0.5], [0.51, 0.54], [0.5, 0.54], [0.5, 0.55], [0.53, 0.45], [0.54, 0.45], [0.53, 0.42], [0.54, 0.47]]}" + } + ], + "images": "RH20T_cfg6_task_0081_user_0020_scene_0009_cfg_0006_104122064161#0", + "gt": "[[338, 111], [339, 146], [321, 180], [324, 196], [318, 195], [321, 198], [338, 163], [344, 163], [339, 152], [346, 170]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "46745_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the plate in the dishwasher on its side', and now you are performing subtask 'put the plate in the dishwasher on its side.'\nStart at [0.49, 0.37]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.37], [0.56, 0.32], [0.6, 0.35], [0.61, 0.62], [0.59, 0.53], [0.55, 0.49], [0.52, 0.48], [0.53, 0.56], [0.52, 0.47], [0.52, 0.47]]}" + } + ], + "images": "46745_exterior_image_2_left#0", + "gt": "[[157, 67], [178, 58], [193, 63], [194, 111], [188, 95], [177, 88], [168, 87], [169, 100], [165, 84], [165, 84]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0073_user_0001_scene_0003_cfg_0001_039422060546#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the things in the drawer', and now you are performing subtask 'place the block in the drawer, push the drawer handle to close the drawer.'\nWith start location [0.33, 0.38], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.33, 0.38], [0.32, 0.28], [0.35, 0.22], [0.36, 0.17], [0.33, 0.18], [0.33, 0.25], [0.26, 0.42], [0.31, 0.32], [0.38, 0.24], [0.39, 0.25]]}" + } + ], + "images": "RH20T_cfg1_task_0073_user_0001_scene_0003_cfg_0001_039422060546#214", + "gt": "[[214, 137], [205, 102], [225, 81], [230, 62], [210, 66], [210, 91], [169, 151], [196, 115], [246, 86], [247, 91]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "71065_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push the lid', and now you are performing subtask 'push the lid.'\nYou start from [0.52, 0.0]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.0], [0.58, 0.19], [0.57, 0.19], [0.56, 0.31], [0.54, 0.37], [0.52, 0.37], [0.57, 0.27], [0.59, 0.15], [0.59, 0.09], [0.58, 0.11]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/71065_exterior_image_2_left#0.jpg", + "gt": "[[167, 0], [187, 35], [181, 34], [179, 56], [174, 66], [167, 67], [183, 48], [189, 27], [188, 16], [186, 19]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0003_user_0015_scene_0008_cfg_0001_819112071037#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the three buttons from left to right in sequence', and now you are performing subtask 'press the orange button on the left, press the yellow button in the middle.'\nYou start from [0.6, 0.04]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.04], [0.68, 0.06], [0.62, 0.12], [0.63, 0.22], [0.65, 0.0], [0.76, 0.06], [0.85, 0.04], [0.91, 0.11], [0.9, 0.18], [0.9, 0.18]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0003_user_0015_scene_0008_cfg_0001_819112071037#161.jpg", + "gt": "[[385, 15], [437, 22], [400, 43], [402, 79], [418, 0], [486, 21], [547, 14], [583, 38], [573, 64], [573, 64]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "91405_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the can and place it on the lower shelf', and now you are performing subtask 'pick up the can and place it on the lower shelf.'\nBegin at [0.71, 0.59] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.71, 0.59], [0.62, 0.66], [0.61, 0.83], [0.55, 0.78], [0.6, 0.6], [0.58, 0.13], [0.57, 0.08], [0.52, 0.22], [0.62, 0.26], [0.66, 0.39]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/91405_exterior_image_2_left#0.jpg", + "gt": "[[226, 107], [199, 118], [195, 150], [177, 141], [193, 108], [185, 23], [181, 15], [165, 40], [197, 46], [212, 70]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0016_user_0014_scene_0003_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'hit the pool ball', and now you are performing subtask 'pick up the pool cue from the table, strike the pool ball on the table.'\nWith start location [0.64, 0.38], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.64, 0.38], [0.66, 0.3], [0.62, 0.24], [0.62, 0.17], [0.56, 0.16], [0.56, 0.18], [0.56, 0.17], [0.57, 0.14], [0.59, 0.19], [0.59, 0.19]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0016_user_0014_scene_0003_cfg_0001_038522062547#0.jpg", + "gt": "[[407, 136], [422, 108], [400, 87], [398, 63], [360, 57], [357, 66], [356, 60], [367, 52], [377, 68], [377, 68]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_038522063145#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'place the first Megablock on the green base, pick up the second Megablock from the left side of the table, place the second Megablock on the green base.'\nYou begin at [0.74, 0.45]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.74, 0.45], [0.77, 0.4], [0.8, 0.4], [0.81, 0.42], [0.8, 0.44], [0.83, 0.38], [0.87, 0.35], [0.74, 0.32], [0.7, 0.36], [0.68, 0.42]]}" + } + ], + "images": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_038522063145#184", + "gt": "[[473, 161], [492, 144], [510, 145], [517, 151], [510, 157], [533, 135], [555, 127], [471, 116], [450, 129], [437, 152]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "36754_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the carrot plushie in the pot, then turn the tap nozzle to the left, put the carrot plushie back on the table', and now you are performing subtask 'place the carrot plushie on the table.'\nYour starting position is [0.47, 0.53]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.53], [0.47, 0.54], [0.45, 0.53], [0.42, 0.45], [0.41, 0.46], [0.39, 0.54], [0.4, 0.53], [0.39, 0.48], [0.4, 0.5], [0.39, 0.48]]}" + } + ], + "images": "36754_exterior_image_1_left#287", + "gt": "[[149, 95], [149, 97], [145, 96], [133, 81], [130, 83], [125, 97], [127, 96], [124, 87], [127, 90], [125, 87]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "88032_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'switch on the adapter', and now you are performing subtask 'switch on the adapter.'\nFrom start point [0.42, 0.9], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.9], [0.41, 0.64], [0.47, 0.45], [0.47, 0.48], [0.62, 0.38], [0.62, 0.36], [0.68, 0.39], [0.7, 0.38], [0.51, 0.4], [0.49, 0.37]]}" + } + ], + "images": "88032_exterior_image_2_left#0", + "gt": "[[133, 162], [131, 116], [151, 81], [152, 87], [198, 68], [200, 64], [217, 70], [225, 69], [163, 72], [157, 66]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0014_user_0014_scene_0001_cfg_0001_819112071037#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the block from the scale', and now you are performing subtask 'remove the block from the scale.'\nStarting at [0.48, 0.0], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.0], [0.73, 0.0], [0.77, 0.04], [0.75, 0.11], [0.74, 0.35], [0.74, 0.24], [0.62, 0.15], [0.51, 0.13], [0.47, 0.15], [0.45, 0.17]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0014_user_0014_scene_0001_cfg_0001_819112071037#0.jpg", + "gt": "[[309, 0], [464, 0], [493, 14], [481, 39], [476, 125], [474, 86], [394, 54], [329, 47], [299, 54], [290, 63]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0037_user_0014_scene_0010_cfg_0001_750612070851#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up a block from the table, transfer a block from the table to the top of a block, place the block on another block.'\nBegin at [0.49, 0.45] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.45], [0.52, 0.49], [0.57, 0.62], [0.58, 0.62], [0.6, 0.65], [0.6, 0.72], [0.61, 0.68], [0.64, 0.72], [0.65, 0.79], [0.64, 0.88]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0037_user_0014_scene_0010_cfg_0001_750612070851#0.jpg", + "gt": "[[315, 161], [334, 176], [366, 224], [370, 222], [382, 235], [381, 259], [393, 245], [409, 260], [414, 283], [410, 317]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0014_user_0014_scene_0008_cfg_0006_104122061018#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the object from the scale', and now you are performing subtask 'remove the object from the scale.'\nYour starting position is [0.57, 0.08]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.08], [0.53, 0.26], [0.53, 0.38], [0.59, 0.16], [0.58, 0.14], [0.57, 0.16], [0.58, 0.26], [0.59, 0.29], [0.61, 0.29], [0.61, 0.29]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0014_user_0014_scene_0008_cfg_0006_104122061018#0.jpg", + "gt": "[[367, 28], [342, 92], [339, 138], [377, 58], [373, 51], [365, 56], [373, 93], [378, 106], [391, 106], [391, 106]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "14825_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker inside the silver pot', and now you are performing subtask 'put the marker inside the silver pot.'\nYour initial coordinate is [0.78, 0.28]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.77, 0.28], [0.7, 0.44], [0.6, 0.58], [0.6, 0.74], [0.57, 0.62], [0.53, 0.64], [0.54, 0.72], [0.55, 0.66], [0.61, 0.56], [0.61, 0.56]]}" + } + ], + "images": "14825_exterior_image_2_left#0", + "gt": "[[248, 50], [224, 80], [193, 105], [191, 133], [181, 112], [171, 115], [174, 130], [176, 119], [196, 101], [196, 101]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "26830_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the packet from the sink', and now you are performing subtask 'remove the packet from the sink.'\nWith start location [0.31, 0.49], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.31, 0.49], [0.52, 0.71], [0.55, 0.86], [0.52, 0.88], [0.53, 0.93], [0.53, 0.75], [0.51, 0.72], [0.46, 0.72], [0.32, 0.78], [0.35, 0.72]]}" + } + ], + "images": "26830_exterior_image_1_left#0", + "gt": "[[98, 89], [166, 127], [175, 154], [168, 158], [170, 167], [171, 135], [162, 130], [148, 130], [104, 141], [111, 129]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "70471_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the white and black mugs from the drawer, push the drawer closed, close the open bottom cabinet', and now you are performing subtask 'place the white cup on the countertop.'\nYou start from [0.7, 0.69]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.69], [0.61, 0.67], [0.59, 0.71], [0.61, 0.71], [0.6, 0.76], [0.63, 0.69], [0.62, 0.62], [0.53, 0.54], [0.52, 0.59], [0.52, 0.59]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/70471_exterior_image_1_left#0.jpg", + "gt": "[[225, 125], [195, 121], [189, 127], [196, 128], [192, 136], [203, 124], [199, 112], [170, 98], [168, 107], [168, 107]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0044_user_0010_scene_0008_cfg_0005_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up a bag of things', and now you are performing subtask 'pick up a bag of things.'\nYou start from [0.49, 0.14]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.14], [0.53, 0.24], [0.54, 0.27], [0.54, 0.32], [0.55, 0.39], [0.54, 0.39], [0.55, 0.42], [0.55, 0.34], [0.49, 0.12], [0.5, 0.11]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0044_user_0010_scene_0008_cfg_0005_104122061850#0.jpg", + "gt": "[[311, 50], [337, 88], [347, 96], [344, 117], [352, 141], [348, 139], [349, 150], [350, 121], [316, 43], [317, 38]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "45684_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the pen from inside the mug and place it on the desk', and now you are performing subtask 'remove the pen from inside the mug and place it on the desk.'\nWith start location [0.35, 0.09], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.09], [0.42, 0.36], [0.43, 0.47], [0.49, 0.73], [0.53, 0.49], [0.48, 0.51], [0.48, 0.65], [0.5, 0.62], [0.47, 0.49], [0.45, 0.47]]}" + } + ], + "images": "45684_exterior_image_1_left#0", + "gt": "[[110.5, 15.5], [135.5, 65.5], [136.5, 84.5], [156.5, 131.5], [168.5, 87.5], [154.5, 91.5], [154.5, 117.5], [160.5, 111.5], [151.5, 88.5], [144.5, 84.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0038_user_0016_scene_0009_cfg_0001_819112070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the cup', and now you are performing subtask 'pick up the cup.'\nWith start location [0.62, 0.47], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.47], [0.62, 0.47], [0.63, 0.47], [0.62, 0.46], [0.61, 0.43], [0.62, 0.46], [0.61, 0.47], [0.61, 0.49], [0.62, 0.45], [0.63, 0.39]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0038_user_0016_scene_0009_cfg_0001_819112070044#0.jpg", + "gt": "[[400, 168], [399, 170], [401, 170], [400, 166], [393, 153], [394, 165], [388, 170], [389, 176], [397, 161], [401, 141]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_104122063633#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'pick up the lid on the box, transfer the lid from the box to the tabletop, place the lid on the tabletop, grab the object in the box.'\nStart at [0.37, 0.34]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.34], [0.32, 0.41], [0.29, 0.44], [0.28, 0.49], [0.29, 0.52], [0.28, 0.47], [0.27, 0.7], [0.28, 0.73], [0.28, 0.55], [0.3, 0.52]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_104122063633#0.jpg", + "gt": "[[234, 121], [207, 147], [187, 158], [181, 178], [184, 187], [181, 169], [175, 251], [179, 262], [179, 197], [190, 186]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0058_user_0015_scene_0003_cfg_0001_035622060973#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn off the power strips', and now you are performing subtask 'press the third button on the power socket, press the fourth button on the power socket.'\nThe trajectory starts at [0.35, 0.27]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.27], [0.35, 0.26], [0.34, 0.28], [0.35, 0.31], [0.35, 0.31], [0.35, 0.28], [0.35, 0.28], [0.35, 0.28], [0.35, 0.37], [0.35, 0.39]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0058_user_0015_scene_0003_cfg_0001_035622060973#236.jpg", + "gt": "[[227, 98], [226, 94], [220, 101], [221, 112], [224, 111], [224, 101], [223, 99], [222, 101], [227, 134], [224, 142]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0201_user_0015_scene_0007_cfg_0001_750612070853#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'the medium pipette tip is inserted into the holder for medium pipette tips', and now you are performing subtask 'pick up the medium pipette from the table, transfer the medium pipette from the table to the holder for medium pipette tips, place the tip of the medium pipette into the holder for medium pipette tips.'\nYour initial coordinate is [0.55, 0.23]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.23], [0.47, 0.28], [0.48, 0.22], [0.43, 0.34], [0.48, 0.55], [0.49, 0.45], [0.52, 0.38], [0.63, 0.27], [0.62, 0.31], [0.63, 0.36]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0201_user_0015_scene_0007_cfg_0001_750612070853#0.jpg", + "gt": "[[355, 82], [300, 100], [308, 79], [275, 123], [309, 197], [312, 162], [336, 136], [405, 98], [399, 110], [406, 131]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "89619_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push the black door handle down and open the door', and now you are performing subtask 'push the black door handle down and open the door.'\nYour starting position is [0.82, 0.43]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.82, 0.43], [0.71, 0.48], [0.64, 0.45], [0.49, 0.56], [0.48, 0.52], [0.5, 0.63], [0.64, 0.63], [0.69, 0.57], [0.67, 0.37], [0.79, 0.36]]}" + } + ], + "images": "89619_exterior_image_1_left#0", + "gt": "[[263, 78], [226, 87], [204, 81], [157, 100], [154, 94], [160, 114], [205, 114], [220, 102], [213, 66], [253, 65]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "32668_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the doors of the second shelf of the cabinet', and now you are performing subtask 'pull the handle of the left door of the second shelf to open the door.'\nYou begin at [0.4, 0.0]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.0], [0.33, 0.18], [0.34, 0.27], [0.39, 0.3], [0.37, 0.33], [0.32, 0.31], [0.3, 0.21], [0.31, 0.1], [0.36, 0.07], [0.34, 0.03]]}" + } + ], + "images": "32668_exterior_image_1_left#340", + "gt": "[[128, 0], [107, 33], [109, 48], [124, 54], [118, 59], [103, 55], [95, 38], [99, 18], [116, 13], [109, 5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0005_user_0004_scene_0007_cfg_0001_039422060546#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push the wooden block to the left side of the table', and now you are performing subtask 'push the wooden block to the left side of the table.'\nYour starting position is [0.4, 0.35]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.35], [0.43, 0.39], [0.41, 0.39], [0.51, 0.43], [0.5, 0.44], [0.53, 0.47], [0.59, 0.5], [0.59, 0.52], [0.55, 0.5], [0.59, 0.51]]}" + } + ], + "images": "RH20T_cfg1_task_0005_user_0004_scene_0007_cfg_0001_039422060546#0", + "gt": "[[255, 127], [273, 139], [261, 139], [329, 153], [323, 157], [337, 171], [375, 179], [375, 187], [352, 181], [380, 185]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0058_user_0007_scene_0002_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press a row of two power strip buttons to turn off the power strips', and now you are performing subtask 'press the first button on the power socket.'\nStarting at [0.5, 0.37], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.52, 0.37], [0.54, 0.41], [0.55, 0.46], [0.54, 0.55], [0.54, 0.6], [0.55, 0.6], [0.54, 0.64], [0.54, 0.62], [0.52, 0.62]]}" + } + ], + "images": "RH20T_cfg5_task_0058_user_0007_scene_0002_cfg_0005_104122063678#0", + "gt": "[[323, 132], [335, 133], [347, 146], [352, 167], [345, 199], [344, 215], [349, 216], [346, 231], [346, 223], [333, 224]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0061_user_0015_scene_0001_cfg_0001_750612070851#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the things out of the gift box and put them on the table', and now you are performing subtask 'place the orange ball on the table.'\nYour initial coordinate is [0.46, 0.74]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.74], [0.47, 0.69], [0.55, 0.57], [0.63, 0.54], [0.64, 0.5], [0.69, 0.46], [0.65, 0.47], [0.66, 0.51], [0.65, 0.49], [0.66, 0.47]]}" + } + ], + "images": "RH20T_cfg1_task_0061_user_0015_scene_0001_cfg_0001_750612070851#813", + "gt": "[[296, 265], [300, 247], [349, 207], [406, 194], [412, 179], [439, 165], [416, 171], [422, 183], [413, 178], [422, 171]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "52870_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the lemons and an apple out of the glass bowl and place them on the countertop', and now you are performing subtask 'place the first lemon on the tabletop.'\nFrom start point [0.94, 0.38], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.94, 0.38], [0.81, 0.47], [0.81, 0.48], [0.74, 0.61], [0.74, 0.58], [0.72, 0.53], [0.68, 0.52], [0.61, 0.59], [0.57, 0.67], [0.62, 0.57]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/52870_exterior_image_1_left#0.jpg", + "gt": "[[300, 68], [258, 84], [260, 86], [236, 109], [237, 105], [230, 96], [219, 93], [194, 106], [183, 121], [200, 102]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "54044_exterior_image_2_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up and put down the green block, then put the salt cellar into the pot and the green block into the sink', and now you are performing subtask 'place the green block in the sink.'\nYou begin at [0.56, 0.35]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.56, 0.35], [0.45, 0.22], [0.43, 0.27], [0.47, 0.32], [0.43, 0.33], [0.46, 0.39], [0.47, 0.31], [0.42, 0.31], [0.45, 0.39], [0.36, 0.2]]}" + } + ], + "images": "54044_exterior_image_2_left#513", + "gt": "[[179, 63], [145, 40], [139, 48], [151, 57], [136, 59], [146, 71], [149, 55], [133, 55], [144, 70], [116, 36]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "10535_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the lid from the pot and put it on the counter', and now you are performing subtask 'pick up the lid from the pot and put it on the counter.'\nStarting at [0.76, 0.0], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.76, 0.0], [0.84, 0.14], [0.81, 0.2], [0.83, 0.35], [0.82, 0.28], [0.69, 0.33], [0.54, 0.51], [0.58, 0.52], [0.56, 0.58], [0.65, 0.34]]}" + } + ], + "images": "10535_exterior_image_2_left#0", + "gt": "[[243, 0], [268, 25], [258, 36], [265, 63], [264, 51], [220, 60], [174, 91], [185, 94], [178, 104], [207, 61]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "78150_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the granola bar in the cabinet', and now you are performing subtask 'place the granola bar in the cabinet.'\nStart at [0.42, 0.52]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.52], [0.4, 0.8], [0.4, 0.71], [0.35, 0.73], [0.33, 0.56], [0.35, 0.02], [0.38, 0.02], [0.42, 0.14], [0.37, 0.09], [0.32, 0.34]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/78150_exterior_image_1_left#0.jpg", + "gt": "[[135, 94], [129, 144], [129, 128], [111, 131], [105, 101], [111, 4], [121, 4], [134, 26], [117, 16], [102, 61]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0073_user_0020_scene_0005_cfg_0006_037522061512#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the things in the drawer', and now you are performing subtask 'pick up the wooden block on the table, transfer the wooden block from the table to the drawer, place the wooden block in the drawer, push the drawer to close the drawer.'\nFrom start point [0.37, 0.22], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.22], [0.41, 0.2], [0.39, 0.17], [0.42, 0.26], [0.36, 0.13], [0.3, 0.16], [0.28, 0.28], [0.31, 0.38], [0.29, 0.23], [0.36, 0.24]]}" + } + ], + "images": "RH20T_cfg6_task_0073_user_0020_scene_0005_cfg_0006_037522061512#351", + "gt": "[[235, 80], [263, 72], [252, 62], [270, 94], [230, 48], [191, 56], [178, 99], [197, 136], [185, 84], [230, 88]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0058_user_0007_scene_0002_cfg_0005_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press a row of two power strip buttons to turn off the power strips', and now you are performing subtask 'press the first button on the power socket, press the second button on the power socket.'\nStart at [0.48, 0.14]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.14], [0.49, 0.15], [0.52, 0.19], [0.54, 0.31], [0.55, 0.4], [0.55, 0.41], [0.55, 0.45], [0.54, 0.43], [0.52, 0.41], [0.52, 0.41]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0058_user_0007_scene_0002_cfg_0005_104122061850#0.jpg", + "gt": "[[305, 50], [316, 53], [330, 70], [348, 110], [349, 145], [353, 148], [351, 163], [347, 153], [332, 149], [332, 149]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0094_user_0007_scene_0009_cfg_0005_036422060215#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'press the cardboard box on the side, press the cardboard box on the side.'\nFrom start point [0.45, 0.33], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.33], [0.45, 0.34], [0.43, 0.35], [0.4, 0.34], [0.39, 0.38], [0.4, 0.35], [0.43, 0.33], [0.43, 0.34], [0.43, 0.38], [0.43, 0.34]]}" + } + ], + "images": "RH20T_cfg5_task_0094_user_0007_scene_0009_cfg_0005_036422060215#188", + "gt": "[[290, 119], [285, 123], [272, 125], [256, 123], [248, 136], [253, 125], [272, 120], [275, 122], [276, 137], [274, 122]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "89488_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the towel from the middle shelf and put it on the table', and now you are performing subtask 'remove the towel from the middle shelf and put it on the table.'\nStart at [0.05, 0.0]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.05, 0.0], [0.29, 0.37], [0.43, 0.37], [0.32, 0.3], [0.22, 0.14], [0.24, 0.32], [0.22, 0.49], [0.07, 0.69], [0.1, 0.67], [0.09, 0.76]]}" + } + ], + "images": "89488_exterior_image_1_left#0", + "gt": "[[16, 0], [94, 67], [138, 67], [104, 54], [71, 26], [78, 58], [69, 88], [23, 124], [31, 120], [28, 136]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "69906_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer, take out the black object, and place it on the countertop, pick up the pen on the countertop, put it back in the drawer, close the drawer', and now you are performing subtask 'pick up the black object in the drawer.'\nBegin at [0.23, 0.76] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.23, 0.76], [0.26, 0.76], [0.28, 0.77], [0.28, 0.79], [0.28, 0.78], [0.27, 0.8], [0.28, 0.83], [0.27, 0.82], [0.28, 0.81], [0.28, 0.83]]}" + } + ], + "images": "69906_exterior_image_1_left#233", + "gt": "[[75, 137], [83, 136], [88, 138], [89, 142], [90, 141], [87, 144], [88, 150], [86, 147], [89, 145], [91, 149]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0016_user_0010_scene_0010_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'hitting the pool ball', and now you are performing subtask 'hitting the pool ball.'\nYou start from [0.5, 0.37]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.51, 0.49], [0.48, 0.57], [0.51, 0.51], [0.5, 0.56], [0.44, 0.76], [0.44, 0.84], [0.51, 0.83], [0.49, 0.67], [0.52, 0.41]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0016_user_0010_scene_0010_cfg_0005_104122063678#0.jpg", + "gt": "[[323, 132], [328, 175], [307, 206], [324, 184], [323, 201], [282, 274], [279, 301], [327, 300], [313, 242], [336, 146]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0037_user_0015_scene_0004_cfg_0001_035622060973#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the first block on the table, transfer the first block from the table to above the second block, place the first block on top of the second block, pick up the third block on the table.'\nFrom start point [0.38, 0.36], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.36], [0.36, 0.36], [0.34, 0.38], [0.34, 0.47], [0.35, 0.43], [0.34, 0.4], [0.41, 0.37], [0.43, 0.39], [0.41, 0.37], [0.37, 0.45]]}" + } + ], + "images": "RH20T_cfg1_task_0037_user_0015_scene_0004_cfg_0001_035622060973#0", + "gt": "[[246, 130], [233, 129], [217, 138], [216, 170], [223, 156], [219, 145], [261, 134], [273, 140], [262, 132], [234, 163]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0037_user_0020_scene_0009_cfg_0006_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the first block from the table, transfer the first block from the table to the second block, place the first block on the second block.'\nThe trajectory starts at [0.53, 0.31]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.31], [0.51, 0.3], [0.48, 0.33], [0.48, 0.41], [0.46, 0.43], [0.51, 0.37], [0.53, 0.39], [0.54, 0.47], [0.56, 0.4], [0.55, 0.35]]}" + } + ], + "images": "RH20T_cfg6_task_0037_user_0020_scene_0009_cfg_0006_104122064161#0", + "gt": "[[338, 110], [326, 108], [308, 119], [306, 149], [296, 155], [326, 134], [338, 139], [348, 169], [358, 144], [354, 125]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "54621_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'rotate the faucet counterclockwise, press the button on the faucet', and now you are performing subtask 'press the button on the faucet.'\nYou start from [1.0, 0.25]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[1.0, 0.25], [0.6, 0.64], [0.32, 0.68], [0.32, 0.77], [0.45, 0.76], [0.51, 0.73], [0.55, 0.39], [0.65, 0.21], [0.58, 0.35], [0.42, 0.45]]}" + } + ], + "images": "54621_exterior_image_1_left#0", + "gt": "[[319.0, 44.5], [191.0, 115.5], [103.0, 122.5], [102.0, 139.5], [145.0, 137.5], [164.0, 130.5], [177.0, 70.5], [209.0, 38.5], [186.0, 62.5], [134.0, 80.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_819112070044#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the fourth block on top of the third block, pick up the fifth block from the table, transfer the fifth block from the table to the top of the stack, place the fifth block on the top of the stack.'\nYour starting position is [0.53, 0.48]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.48], [0.54, 0.41], [0.57, 0.39], [0.64, 0.59], [0.67, 0.6], [0.67, 0.44], [0.58, 0.45], [0.53, 0.38], [0.53, 0.43], [0.53, 0.32]]}" + } + ], + "images": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_819112070044#582", + "gt": "[[336, 174], [345, 146], [364, 141], [410, 213], [427, 215], [430, 160], [374, 163], [341, 137], [337, 155], [342, 115]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_038522063145#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the third block on top of the second block, pick up the fourth block from the table, transfer the fourth block from the table to the position above the third block, place the fourth block on top of the third block.'\nYou start from [0.58, 0.18]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.18], [0.58, 0.15], [0.63, 0.16], [0.67, 0.26], [0.64, 0.32], [0.6, 0.38], [0.65, 0.24], [0.64, 0.07], [0.6, 0.06], [0.59, 0.1]]}" + } + ], + "images": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_038522063145#381", + "gt": "[[371, 66], [374, 54], [401, 57], [426, 95], [412, 116], [381, 136], [417, 87], [407, 25], [387, 20], [375, 35]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "52069_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the yellow plate to the front right side of the table', and now you are performing subtask 'place the yellow plate in the front right side of the table.'\nFrom start point [0.7, 0.56], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.56], [0.7, 0.44], [0.68, 0.41], [0.67, 0.44], [0.69, 0.57], [0.72, 0.67], [0.73, 0.67], [0.76, 0.53], [0.77, 0.55], [0.77, 0.54]]}" + } + ], + "images": "52069_exterior_image_1_left#133", + "gt": "[[223, 101], [223, 79], [217, 74], [215, 79], [222, 103], [231, 120], [234, 120], [244, 95], [247, 99], [245, 97]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0029_user_0020_scene_0002_cfg_0007_104122061330#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp the handle and open the drawer', and now you are performing subtask 'grasp the handle and open the drawer.'\nYou begin at [0.42, 0.29]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.29], [0.43, 0.29], [0.44, 0.33], [0.47, 0.32], [0.55, 0.18], [0.56, 0.28], [0.54, 0.42], [0.49, 0.22], [0.48, 0.21], [0.47, 0.22]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0029_user_0020_scene_0002_cfg_0007_104122061330#0.jpg", + "gt": "[[266, 103], [278, 106], [283, 118], [301, 115], [354, 66], [359, 102], [347, 151], [313, 78], [307, 74], [303, 81]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "14883_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the pen on the table and put it in the cup', and now you are performing subtask 'pick up the pen on the table and put it in the cup.'\nYou begin at [0.32, 0.63]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.63], [0.36, 0.64], [0.35, 0.7], [0.37, 0.76], [0.43, 0.63], [0.46, 0.63], [0.47, 0.66], [0.48, 0.61], [0.5, 0.68], [0.43, 0.6]]}" + } + ], + "images": "14883_exterior_image_2_left#0", + "gt": "[[101, 114], [114, 115], [113, 126], [119, 136], [136, 113], [146, 113], [149, 119], [154, 109], [159, 122], [137, 108]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0037_user_0014_scene_0001_cfg_0007_104122063633#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line', and now you are performing subtask 'place the second block on top of the first block, pick up the third block from the table, transfer the third block from the table to the stacking position.'\nYour starting position is [0.35, 0.38]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.38], [0.34, 0.37], [0.32, 0.39], [0.3, 0.42], [0.3, 0.45], [0.3, 0.43], [0.32, 0.39], [0.33, 0.36], [0.34, 0.34], [0.36, 0.35]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0037_user_0014_scene_0001_cfg_0007_104122063633#252.jpg", + "gt": "[[224, 136], [217, 132], [205, 139], [193, 152], [189, 162], [191, 154], [206, 140], [209, 128], [220, 121], [231, 127]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "60714_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the green sachet in the bag', and now you are performing subtask 'put the green sachet in the bag.'\nYour initial coordinate is [0.52, 0.68]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.68], [0.44, 0.89], [0.51, 0.84], [0.53, 0.88], [0.52, 1.0], [0.45, 0.64], [0.33, 0.54], [0.3, 0.47], [0.32, 0.52], [0.31, 0.55]]}" + } + ], + "images": "60714_exterior_image_2_left#0", + "gt": "[[165, 122], [142, 160], [163, 151], [170, 158], [166, 180], [143, 115], [105, 97], [96, 84], [104, 94], [100, 99]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_037522061512#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the third block on the table, transfer the third block from the table to above the first block, place the third block on top of the first block, pick up the fourth block on the table.'\nBegin at [0.44, 0.38] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.38], [0.38, 0.44], [0.38, 0.49], [0.39, 0.5], [0.39, 0.46], [0.41, 0.4], [0.45, 0.38], [0.47, 0.26], [0.43, 0.22], [0.39, 0.26]]}" + } + ], + "images": "RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_037522061512#616", + "gt": "[[280, 136], [246, 157], [246, 175], [252, 179], [248, 165], [263, 144], [289, 135], [301, 93], [273, 78], [247, 94]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0085_user_0007_scene_0007_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the bubble ring from the assembled bubble ring and ball', and now you are performing subtask 'remove the bubble ring from the assembled bubble ring and ball.'\nThe trajectory starts at [0.5, 0.37]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.56, 0.37], [0.57, 0.45], [0.57, 0.54], [0.55, 0.57], [0.54, 0.66], [0.55, 0.62], [0.52, 0.57], [0.52, 0.5], [0.52, 0.49]]}" + } + ], + "images": "RH20T_cfg5_task_0085_user_0007_scene_0007_cfg_0005_104122063678#0", + "gt": "[[323, 132], [356, 132], [368, 162], [363, 195], [353, 205], [346, 236], [349, 225], [335, 204], [331, 180], [336, 178]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0130_user_0020_scene_0002_cfg_0006_104122060811#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb', and now you are performing subtask 'insert the bulb into the socket, press the switch on the socket to turn on the bulb.'\nYour starting position is [0.56, 0.33]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.56, 0.33], [0.55, 0.31], [0.55, 0.32], [0.55, 0.32], [0.54, 0.32], [0.55, 0.33], [0.54, 0.34], [0.54, 0.35], [0.54, 0.35], [0.54, 0.33]]}" + } + ], + "images": "RH20T_cfg6_task_0130_user_0020_scene_0002_cfg_0006_104122060811#510", + "gt": "[[356, 119], [355, 113], [349, 114], [349, 116], [348, 117], [349, 120], [347, 121], [347, 125], [348, 125], [347, 118]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "80330_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pour the contents from the cup into the bowl', and now you are performing subtask 'pour the contents from the cup into the bowl.'\nYour starting position is [0.47, 0.61]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.61], [0.43, 0.61], [0.42, 0.58], [0.43, 0.61], [0.43, 0.64], [0.46, 0.62], [0.39, 0.58], [0.4, 0.54], [0.42, 0.54], [0.4, 0.53]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/80330_exterior_image_2_left#0.jpg", + "gt": "[[151, 110], [139, 109], [133, 105], [137, 109], [136, 115], [146, 111], [125, 105], [129, 98], [133, 98], [129, 95]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0038_user_0016_scene_0009_cfg_0001_750612070851#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the cup', and now you are performing subtask 'pick up the cup.'\nYour starting position is [0.62, 0.47]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.47], [0.62, 0.46], [0.62, 0.47], [0.62, 0.46], [0.62, 0.51], [0.6, 0.55], [0.61, 0.63], [0.6, 0.65], [0.59, 0.56], [0.6, 0.43]]}" + } + ], + "images": "RH20T_cfg1_task_0038_user_0016_scene_0009_cfg_0001_750612070851#0", + "gt": "[[398, 168], [398, 166], [400, 168], [398, 165], [394, 185], [386, 197], [389, 226], [386, 233], [379, 203], [384, 155]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "10614_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the shower door, push the shower door closed', and now you are performing subtask 'open the shower door, push the shower door closed.'\nFrom start point [0.81, 0.16], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.81, 0.16], [0.81, 0.22], [0.77, 0.26], [0.8, 0.28], [0.75, 0.28], [0.82, 0.3], [0.85, 0.1], [0.8, 0.18], [0.75, 0.16], [0.75, 0.21]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/10614_exterior_image_1_left#0.jpg", + "gt": "[[260, 29], [260, 40], [248, 46], [257, 50], [240, 51], [261, 54], [273, 18], [255, 32], [241, 28], [241, 37]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "92163_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the towel', and now you are performing subtask 'unfold the towel on the table again.'\nStart at [0.45, 0.63]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.63], [0.49, 0.67], [0.51, 0.63], [0.58, 0.43], [0.58, 0.31], [0.53, 0.39], [0.38, 0.38], [0.47, 0.46], [0.51, 0.55], [0.5, 0.58]]}" + } + ], + "images": "92163_exterior_image_1_left#383", + "gt": "[[143, 114], [157, 121], [164, 114], [186, 77], [186, 55], [171, 70], [120, 69], [150, 82], [162, 99], [159, 105]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0009_user_0014_scene_0004_cfg_0001_819112070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take out a Hanoi block and throw it aside', and now you are performing subtask 'take out a Hanoi block and throw it aside.'\nWith start location [0.62, 0.46], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.46], [0.55, 0.42], [0.52, 0.5], [0.53, 0.54], [0.52, 0.48], [0.57, 0.44], [0.59, 0.44], [0.59, 0.5], [0.58, 0.53], [0.59, 0.44]]}" + } + ], + "images": "RH20T_cfg1_task_0009_user_0014_scene_0004_cfg_0001_819112070044#0", + "gt": "[[399, 167], [355, 150], [336, 180], [339, 196], [336, 174], [362, 160], [375, 159], [380, 180], [374, 191], [377, 157]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0075_user_0007_scene_0004_cfg_0005_105422061350#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'finish setting up the starting position of the chessboard that is almost arranged', and now you are performing subtask 'pick up the black chess piece on the table, transfer the black chess piece from the table to the chessboard, place the black chess piece next to the black chess piece on the chessboard.'\nStarting at [0.43, 0.13], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.13], [0.42, 0.21], [0.46, 0.36], [0.46, 0.42], [0.47, 0.46], [0.45, 0.39], [0.44, 0.38], [0.41, 0.4], [0.39, 0.34], [0.39, 0.34]]}" + } + ], + "images": "RH20T_cfg5_task_0075_user_0007_scene_0004_cfg_0005_105422061350#0", + "gt": "[[276, 47], [268, 75], [294, 130], [296, 152], [304, 166], [288, 140], [280, 136], [261, 145], [251, 124], [251, 124]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0062_user_0014_scene_0003_cfg_0006_104122061018#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the toilet paper on its holder', and now you are performing subtask 'put the toilet paper on its holder.'\nYour starting position is [0.56, 0.08]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.56, 0.08], [0.56, 0.11], [0.42, 0.12], [0.38, 0.17], [0.39, 0.2], [0.39, 0.16], [0.48, 0.24], [0.6, 0.31], [0.6, 0.22], [0.54, 0.11]]}" + } + ], + "images": "RH20T_cfg6_task_0062_user_0014_scene_0003_cfg_0006_104122061018#0", + "gt": "[[359, 30], [359, 39], [267, 42], [245, 62], [247, 73], [250, 59], [310, 88], [385, 111], [384, 81], [348, 41]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "37562_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the banana and pineapple plush on the table', and now you are performing subtask 'put the banana and pineapple plush on the table.'\nStart at [0.62, 0.09]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.09], [0.64, 0.06], [0.59, 0.11], [0.47, 0.33], [0.47, 0.37], [0.5, 0.41], [0.49, 0.43], [0.48, 0.28], [0.5, 0.19], [0.52, 0.18]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/37562_exterior_image_1_left#0.jpg", + "gt": "[[199, 17], [204, 10], [190, 19], [150, 59], [149, 66], [159, 73], [156, 78], [155, 51], [159, 35], [166, 32]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "52635_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the towel', and now you are performing subtask 'pull the folded towel.'\nYou begin at [0.72, 0.01]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.72, 0.01], [0.61, 0.13], [0.62, 0.14], [0.6, 0.24], [0.54, 0.3], [0.57, 0.36], [0.59, 0.25], [0.66, 0.22], [0.73, 0.21], [0.86, 0.23]]}" + } + ], + "images": "52635_exterior_image_1_left#0", + "gt": "[[230, 1], [196, 23], [199, 25], [191, 43], [174, 54], [181, 65], [190, 45], [210, 39], [234, 37], [274, 42]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0057_user_0016_scene_0001_cfg_0002_037522062165#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn on the power strips', and now you are performing subtask 'press the button of the second row socket, press the button of the third row socket, press the button of the fourth row socket.'\nYou start from [0.24, 0.56]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.24, 0.56], [0.23, 0.63], [0.21, 0.64], [0.22, 0.62], [0.24, 0.65], [0.22, 0.62], [0.21, 0.69], [0.2, 0.67], [0.21, 0.6], [0.25, 0.51]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg2_task_0057_user_0016_scene_0001_cfg_0002_037522062165#395.jpg", + "gt": "[[152, 200], [145, 227], [132, 230], [143, 225], [156, 234], [143, 225], [135, 250], [130, 240], [133, 216], [159, 183]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "58247_exterior_image_1_left#TrajectoryPrediction_Task#12", + "task": "TrajectoryPrediction_Task", + "p_index": 12, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the microwave door, place the bowl inside the microwave', and now you are performing subtask 'place the bowl inside the microwave.'\nThe trajectory starts at [0.46, 0.74]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.74], [0.51, 0.87], [0.47, 0.76], [0.47, 0.65], [0.45, 0.69], [0.55, 0.76], [0.6, 0.87], [0.48, 0.96], [0.41, 0.64], [0.47, 0.68]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/58247_exterior_image_1_left#695.jpg", + "gt": "[[147, 133], [163, 156], [149, 136], [151, 117], [145, 124], [176, 136], [193, 156], [153, 173], [131, 116], [149, 123]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "31298_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'straighten out the towel', and now you are performing subtask 'straighten out the towel.'\nYou begin at [0.56, 0.53]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.56, 0.53], [0.51, 0.52], [0.43, 0.68], [0.44, 0.64], [0.48, 0.68], [0.5, 0.32], [0.44, 0.22], [0.3, 0.21], [0.42, 0.32], [0.47, 0.53]]}" + } + ], + "images": "31298_exterior_image_2_left#0", + "gt": "[[178, 95], [162, 94], [138, 123], [141, 116], [153, 122], [161, 57], [142, 39], [96, 37], [133, 58], [149, 95]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "36160_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the cloth on the table', and now you are performing subtask 'unfold the cloth.'\nWith start location [0.63, 0.29], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.29], [0.51, 0.34], [0.5, 0.37], [0.5, 0.46], [0.52, 0.5], [0.55, 0.39], [0.57, 0.36], [0.64, 0.34], [0.63, 0.29], [0.65, 0.18]]}" + } + ], + "images": "36160_exterior_image_2_left#150", + "gt": "[[202, 52], [163, 62], [161, 67], [159, 83], [166, 90], [175, 70], [182, 64], [205, 61], [203, 52], [208, 33]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "64889_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the lid on the bottle', and now you are performing subtask 'put the lid on the bottle.'\nThe trajectory starts at [0.44, 0.19]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.19], [0.47, 0.09], [0.52, 0.07], [0.49, 0.12], [0.53, 0.32], [0.52, 0.23], [0.45, 0.17], [0.39, 0.26], [0.44, 0.27], [0.44, 0.27]]}" + } + ], + "images": "64889_exterior_image_2_left#0", + "gt": "[[140, 35], [150, 17], [167, 12], [158, 22], [169, 57], [165, 42], [144, 31], [126, 47], [141, 48], [141, 48]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "4134_exterior_image_1_left#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'insert the small object into the green building block', and now you are performing subtask 'press the blue block onto the green block.'\nWith start location [0.67, 0.34], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.67, 0.34], [0.67, 0.28], [0.69, 0.24], [0.68, 0.28], [0.68, 0.4], [0.65, 0.29], [0.65, 0.27], [0.64, 0.26], [0.62, 0.28], [0.62, 0.28]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/4134_exterior_image_1_left#621.jpg", + "gt": "[[214.5, 61.0], [215.5, 51.0], [219.5, 44.0], [217.5, 51.0], [217.5, 72.0], [208.5, 53.0], [209.5, 48.0], [204.5, 47.0], [198.5, 50.0], [198.5, 50.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "15216_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the pen from the cup and put it on the table', and now you are performing subtask 'place the pen on the table.'\nThe trajectory starts at [0.59, 0.33]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.33], [0.58, 0.37], [0.57, 0.3], [0.57, 0.24], [0.58, 0.22], [0.61, 0.23], [0.68, 0.29], [0.72, 0.37], [0.73, 0.41], [0.7, 0.28]]}" + } + ], + "images": "15216_exterior_image_1_left#128", + "gt": "[[188, 60], [187, 67], [184, 54], [184, 44], [187, 40], [196, 41], [217, 52], [231, 66], [233, 74], [224, 51]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0077_user_0014_scene_0006_cfg_0002_105422061350#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stacking blocks (small Lego) randomly one at a time', and now you are performing subtask 'place the purple block on the yellow block, pick up the green block on the table, transfer the green block from the table to the purple block, place the green block on the purple block.'\nYou start from [0.3, 0.26]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.3, 0.26], [0.3, 0.22], [0.31, 0.25], [0.23, 0.26], [0.22, 0.32], [0.25, 0.26], [0.3, 0.26], [0.32, 0.28], [0.32, 0.3], [0.32, 0.3]]}" + } + ], + "images": "RH20T_cfg2_task_0077_user_0014_scene_0006_cfg_0002_105422061350#914", + "gt": "[[194, 94], [191, 81], [198, 90], [149, 92], [144, 117], [159, 95], [193, 93], [208, 99], [206, 107], [206, 107]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "38910_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the glue stick out of the cup and put it on the silver object', and now you are performing subtask 'take the glue stick out of the cup and put it on the silver object.'\nBegin at [0.51, 0.0] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.0], [0.52, 0.31], [0.45, 0.57], [0.46, 0.48], [0.43, 0.42], [0.31, 0.36], [0.28, 0.37], [0.3, 0.38], [0.28, 0.41], [0.28, 0.37]]}" + } + ], + "images": "38910_exterior_image_2_left#0", + "gt": "[[164, 0], [167, 56], [143, 103], [147, 86], [139, 75], [99, 65], [89, 66], [95, 69], [91, 74], [91, 67]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0055_user_0015_scene_0009_cfg_0001_038522062547#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'transfer liquid using a dropper', and now you are performing subtask 'place the dropper in the container with liquid, press the dropper to draw liquid.'\nThe trajectory starts at [0.72, 0.11]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.72, 0.11], [0.72, 0.12], [0.72, 0.13], [0.72, 0.15], [0.73, 0.15], [0.73, 0.14], [0.73, 0.15], [0.73, 0.14], [0.73, 0.16], [0.73, 0.16]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0055_user_0015_scene_0009_cfg_0001_038522062547#206.jpg", + "gt": "[[458, 40], [458, 43], [461, 48], [461, 54], [464, 54], [464, 51], [465, 55], [465, 52], [464, 59], [464, 59]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "11415_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the red cup from the sink and put it in the dish rack', and now you are performing subtask 'pick up the red cup from the sink and put it in the dish rack.'\nStart at [0.85, 0.37]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.85, 0.37], [0.61, 0.56], [0.68, 0.43], [0.52, 0.29], [0.45, 0.37], [0.48, 0.56], [0.48, 0.34], [0.57, 0.27], [0.64, 0.28], [0.68, 0.41]]}" + } + ], + "images": "11415_exterior_image_1_left#0", + "gt": "[[273, 67], [195, 101], [216, 77], [165, 52], [143, 67], [153, 101], [154, 61], [183, 48], [206, 51], [217, 73]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0026_user_0014_scene_0007_cfg_0006_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the gripper to push and close the drawer', and now you are performing subtask 'use the gripper to push and close the drawer.'\nStarting at [0.53, 0.31], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.31], [0.55, 0.33], [0.56, 0.32], [0.55, 0.31], [0.54, 0.27], [0.51, 0.19], [0.51, 0.18], [0.52, 0.19], [0.52, 0.18], [0.52, 0.18]]}" + } + ], + "images": "RH20T_cfg6_task_0026_user_0014_scene_0007_cfg_0006_104122064161#0", + "gt": "[[338, 110], [351, 119], [359, 114], [354, 113], [347, 97], [324, 68], [328, 64], [332, 69], [331, 66], [331, 66]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0034_user_0020_scene_0001_cfg_0006_104122063633#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'place the first square in the center, pick up the second square on the right, transfer the second square from the right to below the first square, place the second square below the first square.'\nStart at [0.35, 0.38]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.38], [0.27, 0.29], [0.28, 0.32], [0.31, 0.33], [0.3, 0.37], [0.33, 0.35], [0.36, 0.38], [0.35, 0.44], [0.34, 0.42], [0.36, 0.35]]}" + } + ], + "images": "RH20T_cfg6_task_0034_user_0020_scene_0001_cfg_0006_104122063633#562", + "gt": "[[221, 135], [173, 103], [177, 117], [200, 119], [195, 132], [209, 126], [228, 138], [226, 157], [218, 151], [228, 125]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0037_user_0016_scene_0002_cfg_0004_036422060909#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the first block on the second block, pick up the third block on the surface.'\nStart at [0.46, 0.33]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.33], [0.44, 0.33], [0.44, 0.32], [0.44, 0.3], [0.46, 0.26], [0.46, 0.27], [0.47, 0.29], [0.47, 0.31], [0.46, 0.31], [0.54, 0.36]]}" + } + ], + "images": "RH20T_cfg4_task_0037_user_0016_scene_0002_cfg_0004_036422060909#104", + "gt": "[[293, 120], [283, 118], [282, 115], [282, 109], [292, 94], [297, 98], [299, 105], [298, 111], [293, 112], [345, 128]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0096_user_0015_scene_0005_cfg_0001_038522063145#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the block out of the drawer and put it on the table', and now you are performing subtask 'pull the drawer handle to open the drawer.'\nFrom start point [0.7, 0.13], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.13], [0.69, 0.14], [0.69, 0.11], [0.69, 0.09], [0.68, 0.12], [0.68, 0.13], [0.67, 0.14], [0.68, 0.13], [0.69, 0.12], [0.7, 0.1]]}" + } + ], + "images": "RH20T_cfg1_task_0096_user_0015_scene_0005_cfg_0001_038522063145#187", + "gt": "[[445, 48], [443, 49], [443, 39], [440, 34], [432, 43], [435, 48], [429, 51], [433, 47], [441, 44], [446, 37]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "65705_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the towel', and now you are performing subtask 'fold the towel on the counter.'\nYour initial coordinate is [0.45, 0.54]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.54], [0.46, 0.53], [0.48, 0.54], [0.5, 0.6], [0.52, 0.56], [0.51, 0.56], [0.49, 0.59], [0.47, 0.61], [0.46, 0.63], [0.51, 0.57]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/65705_exterior_image_2_left#182.jpg", + "gt": "[[145, 98], [148, 96], [154, 98], [160, 108], [166, 101], [163, 100], [158, 107], [152, 109], [148, 114], [163, 103]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "53685_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the bowl to the right of the paper', and now you are performing subtask 'move the bowl to the right of the paper.'\nThe trajectory starts at [0.39, 0.31]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.31], [0.41, 0.55], [0.43, 0.61], [0.47, 0.58], [0.51, 0.62], [0.6, 0.79], [0.62, 0.92], [0.57, 0.99], [0.52, 0.94], [0.54, 0.81]]}" + } + ], + "images": "53685_exterior_image_1_left#0", + "gt": "[[124, 55], [130, 99], [139, 109], [152, 105], [164, 111], [191, 142], [198, 165], [181, 178], [168, 169], [173, 146]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "37690_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the candy from the table and put it in the bowl', and now you are performing subtask 'pick up the candy from the table and put it in the bowl.'\nStart at [0.27, 0.72]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.27, 0.72], [0.31, 0.64], [0.38, 0.6], [0.43, 0.64], [0.52, 0.56], [0.69, 0.58], [0.65, 0.52], [0.66, 0.38], [0.68, 0.33], [0.64, 0.34]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/37690_exterior_image_1_left#0.jpg", + "gt": "[[85.0, 129.5], [100.0, 115.5], [120.0, 108.5], [137.0, 114.5], [165.0, 101.5], [221.0, 104.5], [207.0, 93.5], [211.0, 68.5], [219.0, 59.5], [206.0, 60.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "31630_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the marker on the table', and now you are performing subtask 'place the marker on the table.'\nWith start location [0.22, 0.03], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.22, 0.03], [0.22, 0.12], [0.24, 0.26], [0.22, 0.34], [0.24, 0.42], [0.21, 0.31], [0.22, 0.28], [0.25, 0.31], [0.29, 0.46], [0.29, 0.46]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/31630_exterior_image_1_left#0.jpg", + "gt": "[[71.5, 5.0], [70.5, 22.0], [77.5, 47.0], [71.5, 61.0], [75.5, 76.0], [67.5, 56.0], [69.5, 51.0], [79.5, 55.0], [93.5, 83.0], [93.5, 83.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "14341_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the brush inside the silver flask up and down', and now you are performing subtask 'move the brush inside the silver flask up and down.'\nFrom start point [0.99, 0.2], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.99, 0.2], [0.9, 0.29], [0.87, 0.46], [0.83, 0.33], [0.84, 0.44], [0.8, 0.16], [0.81, 0.41], [0.8, 0.12], [0.81, 0.47], [0.81, 0.22]]}" + } + ], + "images": "14341_exterior_image_1_left#0", + "gt": "[[316.5, 36.0], [286.5, 52.0], [279.5, 82.0], [265.5, 59.0], [268.5, 80.0], [257.5, 29.0], [260.5, 74.0], [257.5, 21.0], [260.5, 84.0], [258.5, 40.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "1654_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the whisk to the right, remove the spoon from the cup and put it beside the whisk', and now you are performing subtask 'pick up the spoon from the table.'\nFrom start point [0.6, 0.32], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.32], [0.58, 0.3], [0.54, 0.32], [0.52, 0.41], [0.52, 0.41], [0.52, 0.24], [0.54, 0.17], [0.57, 0.18], [0.62, 0.44], [0.63, 0.61]]}" + } + ], + "images": "1654_exterior_image_1_left#235", + "gt": "[[191, 57], [187, 54], [172, 57], [165, 74], [168, 73], [165, 43], [174, 31], [184, 32], [199, 79], [202, 110]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "31630_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the marker on the table', and now you are performing subtask 'place the marker on the table.'\nYou start from [0.65, 0.0]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.0], [0.57, 0.16], [0.61, 0.16], [0.48, 0.46], [0.5, 0.48], [0.48, 0.57], [0.52, 0.43], [0.6, 0.4], [0.59, 0.51], [0.54, 0.53]]}" + } + ], + "images": "31630_exterior_image_2_left#0", + "gt": "[[208, 0], [183, 28], [194, 29], [155, 82], [160, 87], [155, 103], [165, 78], [191, 72], [189, 92], [173, 95]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_104122064161#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the fourth block on the table, transfer the fourth block from the table to above the third block, place the fourth block on top of the third block, pick up the fifth block on the table.'\nThe trajectory starts at [0.61, 0.42]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.61, 0.42], [0.61, 0.49], [0.62, 0.46], [0.58, 0.41], [0.57, 0.36], [0.51, 0.38], [0.51, 0.4], [0.55, 0.3], [0.56, 0.26], [0.55, 0.23]]}" + } + ], + "images": "RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_104122064161#647", + "gt": "[[391, 150], [391, 178], [396, 165], [372, 147], [365, 128], [326, 136], [326, 143], [354, 109], [359, 95], [354, 82]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "85017_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'slide the white cup on the desk', and now you are performing subtask 'slide the white cup on the desk.'\nYour initial coordinate is [0.8, 0.01]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.8, 0.01], [0.75, 0.27], [0.76, 0.33], [0.69, 0.41], [0.73, 0.42], [0.69, 0.47], [0.64, 0.46], [0.6, 0.49], [0.61, 0.51], [0.6, 0.44]]}" + } + ], + "images": "85017_exterior_image_2_left#0", + "gt": "[[255, 2], [240, 48], [243, 59], [221, 73], [234, 76], [220, 85], [204, 83], [193, 88], [196, 92], [192, 79]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "53689_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the napkin completely', and now you are performing subtask 'push the fabric on the table to the left.'\nYour starting position is [0.63, 0.63]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.63], [0.66, 0.73], [0.64, 0.78], [0.56, 0.77], [0.49, 0.68], [0.45, 0.72], [0.57, 0.64], [0.61, 0.52], [0.61, 0.41], [0.62, 0.45]]}" + } + ], + "images": "53689_exterior_image_1_left#333", + "gt": "[[203, 114], [212, 132], [204, 140], [178, 139], [156, 123], [143, 130], [183, 115], [195, 93], [195, 74], [198, 81]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0079_user_0015_scene_0003_cfg_0001_819112070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the microwave door', and now you are performing subtask 'pull the handle of the microwave door to open the door.'\nYour initial coordinate is [0.62, 0.47]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.47], [0.63, 0.39], [0.64, 0.38], [0.64, 0.42], [0.63, 0.43], [0.63, 0.41], [0.62, 0.45], [0.55, 0.47], [0.56, 0.54], [0.54, 0.51]]}" + } + ], + "images": "RH20T_cfg1_task_0079_user_0015_scene_0003_cfg_0001_819112070044#0", + "gt": "[[400, 168], [404, 142], [412, 136], [410, 150], [401, 156], [404, 146], [400, 161], [352, 171], [357, 193], [346, 182]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "48959_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the pineapple plush toy in the black pot, and the pear in the box', and now you are performing subtask 'transfer the pineapple plush toy from the plate to the black pot.'\nYou begin at [0.93, 0.11]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.93, 0.11], [0.82, 0.01], [0.73, 0.0], [0.68, 0.06], [0.64, 0.13], [0.63, 0.19], [0.63, 0.32], [0.62, 0.26], [0.63, 0.25], [0.62, 0.09]]}" + } + ], + "images": "48959_exterior_image_1_left#0", + "gt": "[[297.5, 19.0], [263.5, 1.0], [233.5, 0.0], [217.5, 10.0], [204.5, 23.0], [200.5, 34.0], [202.5, 57.0], [197.5, 46.0], [200.5, 45.0], [197.5, 17.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0018_user_0010_scene_0009_cfg_0005_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'throw a block pulled out from the middle of the Jenga tower onto the table', and now you are performing subtask 'throw a block pulled out from the middle of the Jenga tower onto the table.'\nStart at [0.48, 0.14]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.14], [0.5, 0.2], [0.6, 0.34], [0.61, 0.36], [0.58, 0.39], [0.6, 0.36], [0.59, 0.29], [0.54, 0.23], [0.54, 0.2], [0.5, 0.18]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0018_user_0010_scene_0009_cfg_0005_104122061850#0.jpg", + "gt": "[[305, 50], [317, 73], [385, 121], [389, 128], [374, 139], [387, 129], [380, 104], [347, 82], [344, 71], [319, 66]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0014_user_0014_scene_0008_cfg_0006_104122061602#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the object from the scale', and now you are performing subtask 'remove the object from the scale.'\nWith start location [0.7, 0.29], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.29], [0.67, 0.45], [0.66, 0.43], [0.66, 0.47], [0.64, 0.53], [0.64, 0.69], [0.66, 0.57], [0.65, 0.48], [0.65, 0.44], [0.61, 0.42]]}" + } + ], + "images": "RH20T_cfg6_task_0014_user_0014_scene_0008_cfg_0006_104122061602#0", + "gt": "[[446, 106], [426, 161], [420, 155], [423, 170], [410, 192], [407, 249], [420, 206], [419, 174], [415, 160], [393, 151]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0016_user_0020_scene_0004_cfg_0007_104122061602#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'hit the pool ball', and now you are performing subtask 'hit the pool ball.'\nYou start from [0.7, 0.32]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.32], [0.69, 0.41], [0.7, 0.37], [0.7, 0.29], [0.74, 0.31], [0.75, 0.37], [0.74, 0.33], [0.72, 0.37], [0.73, 0.31], [0.72, 0.25]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0016_user_0020_scene_0004_cfg_0007_104122061602#0.jpg", + "gt": "[[448, 116], [441, 149], [450, 134], [446, 104], [472, 110], [479, 133], [473, 118], [460, 132], [469, 111], [461, 91]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "68635_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer', and now you are performing subtask 'open the drawer.'\nThe trajectory starts at [0.42, 0.5]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.5], [0.49, 0.56], [0.43, 0.57], [0.41, 0.61], [0.39, 0.71], [0.43, 0.65], [0.42, 0.72], [0.43, 0.66], [0.42, 0.77], [0.42, 0.77]]}" + } + ], + "images": "68635_exterior_image_2_left#0", + "gt": "[[134, 90], [156, 100], [138, 102], [130, 109], [125, 128], [137, 117], [133, 129], [139, 119], [134, 138], [134, 138]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "79914_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the small pillow against the right armrest, move the pillow in the middle of the couch to the right side of the couch', and now you are performing subtask 'transfer the pillow from the middle of the sofa to the right side of the sofa.'\nYour starting position is [0.42, 0.68]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.68], [0.41, 0.71], [0.43, 0.74], [0.41, 0.85], [0.47, 0.88], [0.43, 1.0], [0.42, 0.93], [0.47, 0.75], [0.57, 0.71], [0.52, 0.72]]}" + } + ], + "images": "79914_exterior_image_1_left#282", + "gt": "[[133.5, 122.5], [130.5, 127.5], [138.5, 132.5], [130.5, 152.5], [150.5, 157.5], [138.5, 179.5], [133.5, 167.5], [151.5, 134.5], [182.5, 128.5], [167.5, 129.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg3_task_0003_user_0016_scene_0001_cfg_0003_104122063550#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the green button on the right, press the green button in the middle.'\nStart at [0.47, 0.96]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.96], [0.46, 0.95], [0.46, 0.88], [0.44, 0.76], [0.45, 0.77], [0.46, 0.76], [0.46, 0.74], [0.46, 0.63], [0.47, 0.62], [0.48, 0.63]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg3_task_0003_user_0016_scene_0001_cfg_0003_104122063550#160.jpg", + "gt": "[[303, 347], [296, 342], [296, 316], [284, 272], [287, 276], [292, 274], [296, 265], [295, 226], [298, 224], [305, 228]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "81280_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the white cloth down from the black stand', and now you are performing subtask 'take the white cloth down from the black stand.'\nYou start from [0.45, 0.0]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.0], [0.44, 0.06], [0.45, 0.09], [0.43, 0.09], [0.44, 0.04], [0.37, 0.01], [0.38, 0.09], [0.41, 0.12], [0.43, 0.11], [0.4, 0.12]]}" + } + ], + "images": "81280_exterior_image_2_left#0", + "gt": "[[144, 0], [141, 11], [144, 17], [139, 16], [140, 7], [118, 1], [121, 16], [131, 22], [137, 20], [128, 22]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "91403_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the banana and the pear plush toys from the box to the plate then move the box to the right', and now you are performing subtask 'push the box to the right.'\nThe trajectory starts at [0.37, 0.51]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.51], [0.42, 0.52], [0.39, 0.53], [0.41, 0.57], [0.32, 0.48], [0.25, 0.54], [0.21, 0.63], [0.26, 0.56], [0.35, 0.51], [0.47, 0.54]]}" + } + ], + "images": "91403_exterior_image_1_left#191", + "gt": "[[117, 91], [135, 93], [126, 96], [132, 102], [103, 87], [80, 98], [68, 114], [82, 100], [112, 92], [149, 97]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0094_user_0007_scene_0003_cfg_0004_104122062295#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the box', and now you are performing subtask 'push the lid of the box, push the lid of the box.'\nYour starting position is [0.41, 0.29]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.29], [0.35, 0.32], [0.31, 0.29], [0.32, 0.15], [0.31, 0.06], [0.29, 0.1], [0.28, 0.18], [0.29, 0.19], [0.31, 0.17], [0.35, 0.3]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0094_user_0007_scene_0003_cfg_0004_104122062295#271.jpg", + "gt": "[[264, 104], [225, 117], [197, 103], [204, 55], [199, 23], [186, 37], [181, 64], [183, 67], [198, 63], [227, 109]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "10383_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'partially unfold the towel on the table', and now you are performing subtask 'fold the towel partially unfolded.'\nYou begin at [0.87, 0.11]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.87, 0.11], [0.54, 0.49], [0.5, 0.7], [0.56, 0.79], [0.55, 0.85], [0.52, 0.74], [0.55, 0.49], [0.74, 0.23], [0.95, 0.14], [0.77, 0.29]]}" + } + ], + "images": "10383_exterior_image_1_left#0", + "gt": "[[278, 19], [173, 89], [159, 126], [179, 142], [177, 153], [165, 133], [176, 89], [237, 41], [305, 25], [248, 52]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "7069_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the bowl to the right on the table', and now you are performing subtask 'move the bowl to the right on the table.'\nStarting at [0.38, 0.14], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.14], [0.49, 0.02], [0.64, 0.01], [0.64, 0.03], [0.6, 0.13], [0.56, 0.31], [0.56, 0.28], [0.68, 0.37], [0.71, 0.46], [0.66, 0.3]]}" + } + ], + "images": "7069_exterior_image_1_left#0", + "gt": "[[121, 26], [156, 3], [204, 1], [206, 6], [191, 23], [178, 56], [180, 51], [217, 67], [227, 82], [210, 54]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0016_user_0003_scene_0009_cfg_0002_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'an action of hitting the pool ball', and now you are performing subtask 'pick up the pool cue from the table, transfer the pool cue from the table to the pool ball, strike the pool ball on the green pool table.'\nWith start location [0.44, 0.48], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.48], [0.5, 0.53], [0.51, 0.63], [0.46, 0.71], [0.41, 0.71], [0.38, 0.67], [0.38, 0.69], [0.51, 0.66], [0.46, 0.66], [0.46, 0.64]]}" + } + ], + "images": "RH20T_cfg2_task_0016_user_0003_scene_0009_cfg_0002_037522062165#0", + "gt": "[[284, 174], [317, 192], [324, 227], [297, 254], [262, 255], [246, 242], [245, 248], [325, 237], [297, 239], [295, 230]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0004_user_0020_scene_0005_cfg_0007_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the big block from the desktop and move it to the left', and now you are performing subtask 'pick up the big block from the desktop and move it to the left.'\nFrom start point [0.53, 0.32], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.32], [0.54, 0.54], [0.57, 0.61], [0.58, 0.71], [0.58, 0.56], [0.56, 0.48], [0.6, 0.4], [0.58, 0.31], [0.55, 0.36], [0.58, 0.31]]}" + } + ], + "images": "RH20T_cfg7_task_0004_user_0020_scene_0005_cfg_0007_104122064161#0", + "gt": "[[337, 115], [346, 195], [368, 218], [371, 254], [373, 203], [361, 173], [386, 143], [373, 113], [352, 128], [373, 113]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0014_user_0010_scene_0005_cfg_0004_104422070011#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the object from the scale', and now you are performing subtask 'remove the object from the scale.'\nBegin at [0.53, 0.29] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.29], [0.55, 0.32], [0.52, 0.32], [0.56, 0.47], [0.54, 0.47], [0.53, 0.44], [0.52, 0.48], [0.51, 0.47], [0.65, 0.5], [0.6, 0.57]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0014_user_0010_scene_0005_cfg_0004_104422070011#0.jpg", + "gt": "[[340, 104], [352, 116], [336, 116], [359, 168], [346, 170], [338, 158], [330, 172], [324, 168], [417, 179], [383, 206]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "90700_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the towel and the shirt off the back of the chair and put them on the seat', and now you are performing subtask 'pull the shirt onto the seat of the chair.'\nWith start location [0.77, 0.46], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.77, 0.46], [0.93, 0.66], [0.98, 0.79], [1.0, 0.53], [0.98, 0.47], [0.91, 0.36], [0.83, 0.32], [0.78, 0.35], [0.81, 0.44], [0.76, 0.44]]}" + } + ], + "images": "90700_exterior_image_2_left#158", + "gt": "[[245, 83], [298, 119], [315, 142], [320, 95], [314, 85], [290, 64], [267, 58], [249, 63], [260, 80], [243, 80]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "40658_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the green cloth', and now you are performing subtask 'unfold the green cloth.'\nBegin at [0.49, 0.01] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.01], [0.55, 0.37], [0.53, 0.35], [0.55, 0.41], [0.53, 0.51], [0.46, 0.38], [0.48, 0.29], [0.52, 0.34], [0.5, 0.4], [0.48, 0.37]]}" + } + ], + "images": "40658_exterior_image_1_left#0", + "gt": "[[157, 1], [177, 67], [171, 63], [175, 73], [170, 92], [148, 68], [153, 52], [168, 62], [161, 72], [153, 66]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "4936_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer, place the black clip in the open drawer', and now you are performing subtask 'pull the drawer handle to open the drawer.'\nYou start from [0.68, 0.14]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.14], [0.64, 0.17], [0.59, 0.26], [0.52, 0.58], [0.51, 0.61], [0.5, 0.56], [0.49, 0.58], [0.48, 0.56], [0.53, 0.63], [0.55, 0.53]]}" + } + ], + "images": "4936_exterior_image_2_left#0", + "gt": "[[219, 26], [206, 30], [188, 46], [168, 105], [162, 109], [161, 100], [156, 104], [155, 100], [171, 114], [176, 95]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0070_user_0020_scene_0008_cfg_0006_104122061602#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a sliding window', and now you are performing subtask 'slide the sliding window to the left to open.'\nYour initial coordinate is [0.65, 0.46]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.46], [0.68, 0.49], [0.67, 0.45], [0.64, 0.44], [0.64, 0.45], [0.66, 0.46], [0.66, 0.4], [0.65, 0.31], [0.66, 0.28], [0.68, 0.28]]}" + } + ], + "images": "RH20T_cfg6_task_0070_user_0020_scene_0008_cfg_0006_104122061602#278", + "gt": "[[416, 165], [436, 175], [428, 162], [412, 159], [409, 163], [422, 167], [424, 144], [417, 113], [421, 101], [432, 102]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "57652_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'hang up the cloth on the glass barrier', and now you are performing subtask 'pick up the cloth on the table.'\nBegin at [0.44, 0.62] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.62], [0.41, 0.71], [0.42, 0.77], [0.43, 0.78], [0.44, 0.29], [0.34, 0.57], [0.34, 0.66], [0.35, 0.69], [0.37, 0.71], [0.37, 0.73]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/57652_exterior_image_2_left#117.jpg", + "gt": "[[141, 111], [132, 128], [134, 138], [137, 140], [141, 52], [109, 102], [108, 118], [112, 125], [118, 128], [118, 132]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0033_user_0010_scene_0010_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp a small block to sweep the other small block into the designated position', and now you are performing subtask 'pick up the block to the left of the mark, push another block towards the mark.'\nStarting at [0.5, 0.37], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.51, 0.37], [0.5, 0.42], [0.52, 0.59], [0.5, 0.64], [0.49, 0.61], [0.5, 0.64], [0.48, 0.69], [0.5, 0.62], [0.62, 0.51]]}" + } + ], + "images": "RH20T_cfg5_task_0033_user_0010_scene_0010_cfg_0005_104122063678#0", + "gt": "[[323, 132], [329, 134], [323, 150], [332, 212], [319, 229], [315, 220], [319, 230], [306, 248], [319, 223], [397, 184]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0017_user_0014_scene_0002_cfg_0006_104122061018#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the pen into the pen holder', and now you are performing subtask 'pick up the pen from the table, transfer the pen from the table to the position of the pen holder, place the pen in the pen holder.'\nYou begin at [0.65, 0.33]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.33], [0.62, 0.34], [0.61, 0.39], [0.7, 0.25], [0.75, 0.28], [0.75, 0.34], [0.77, 0.31], [0.78, 0.4], [0.75, 0.33], [0.72, 0.33]]}" + } + ], + "images": "RH20T_cfg6_task_0017_user_0014_scene_0002_cfg_0006_104122061018#251", + "gt": "[[417, 120], [396, 123], [391, 142], [448, 91], [481, 102], [483, 123], [491, 112], [498, 144], [481, 120], [459, 119]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "27620_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the carrot plush inside the pot that is in the sink', and now you are performing subtask 'put the carrot plush inside the pot that is in the sink.'\nFrom start point [0.61, 0.19], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.61, 0.19], [0.52, 0.4], [0.52, 0.47], [0.59, 0.52], [0.47, 0.69], [0.42, 0.71], [0.42, 0.67], [0.38, 0.84], [0.39, 0.74], [0.52, 0.59]]}" + } + ], + "images": "27620_exterior_image_2_left#0", + "gt": "[[195, 35], [168, 72], [168, 85], [190, 94], [150, 125], [135, 128], [135, 120], [123, 152], [125, 134], [167, 106]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "47226_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the green block to the right side of the desk', and now you are performing subtask 'move the green block to the right side of the desk.'\nThe trajectory starts at [0.49, 0.0]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.0], [0.5, 0.36], [0.52, 0.23], [0.5, 0.27], [0.56, 0.39], [0.58, 0.66], [0.6, 0.65], [0.6, 0.44], [0.59, 0.31], [0.59, 0.31]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/47226_exterior_image_1_left#0.jpg", + "gt": "[[156, 0], [161, 65], [166, 42], [159, 48], [180, 70], [187, 118], [192, 117], [193, 79], [188, 55], [188, 55]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "14884_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the microwave light, turn off the microwave light', and now you are performing subtask 'press the button on the microwave again.'\nThe trajectory starts at [0.6, 0.51]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.51], [0.63, 0.49], [0.63, 0.49], [0.63, 0.49], [0.63, 0.49], [0.59, 0.49], [0.56, 0.52], [0.64, 0.52], [0.67, 0.51], [0.67, 0.51]]}" + } + ], + "images": "14884_exterior_image_1_left#150", + "gt": "[[192, 91], [203, 88], [202, 89], [202, 88], [202, 89], [189, 88], [178, 93], [204, 93], [214, 91], [214, 91]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "13937_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the green towel', and now you are performing subtask 'fold the green towel.'\nYour starting position is [0.53, 0.02]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.02], [0.52, 0.06], [0.57, 0.01], [0.47, 0.22], [0.35, 0.64], [0.36, 0.62], [0.4, 0.67], [0.47, 0.67], [0.42, 0.62], [0.42, 0.62]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/13937_exterior_image_2_left#0.jpg", + "gt": "[[168, 3], [165, 11], [182, 1], [149, 40], [112, 116], [114, 112], [129, 121], [150, 121], [133, 112], [133, 112]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg3_task_0058_user_0010_scene_0006_cfg_0003_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn off the power strips', and now you are performing subtask 'press the first button of the power strip.'\nYour initial coordinate is [0.38, 0.0]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.26, 0.03], [0.25, 0.04], [0.26, 0.09], [0.25, 0.1], [0.26, 0.11], [0.26, 0.17], [0.25, 0.21], [0.23, 0.16], [0.23, 0.16]]}" + } + ], + "images": "RH20T_cfg3_task_0058_user_0010_scene_0006_cfg_0003_104122062823#0", + "gt": "[[246, 0], [165, 11], [159, 16], [167, 33], [161, 36], [169, 40], [166, 63], [161, 74], [149, 58], [149, 58]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "14181_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'wipe the pan with the towel', and now you are performing subtask 'wipe the pan with the towel.'\nBegin at [0.6, 0.0] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.0], [0.54, 0.06], [0.47, 0.19], [0.5, 0.49], [0.44, 0.47], [0.49, 0.41], [0.52, 0.44], [0.46, 0.46], [0.44, 0.22], [0.47, 0.28]]}" + } + ], + "images": "14181_exterior_image_2_left#0", + "gt": "[[193, 0], [172, 10], [150, 35], [161, 88], [142, 84], [158, 73], [165, 80], [148, 82], [140, 40], [151, 50]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "89114_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the plush toy in the bowl, wipe the counter with the blue towel, and move the faucet to the center', and now you are performing subtask 'slide the blue towel on the table.'\nBegin at [0.28, 0.58] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.28, 0.58], [0.2, 0.64], [0.23, 0.64], [0.27, 0.61], [0.24, 0.61], [0.25, 0.63], [0.29, 0.62], [0.26, 0.59], [0.26, 0.52], [0.32, 0.53]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/89114_exterior_image_1_left#133.jpg", + "gt": "[[88, 104], [64, 116], [74, 116], [87, 109], [77, 109], [79, 114], [94, 112], [84, 106], [84, 94], [102, 95]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "83180_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the pen from the black mug', and now you are performing subtask 'remove the pen from the black mug.'\nYou begin at [0.22, 0.03]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.22, 0.03], [0.36, 0.01], [0.4, 0.21], [0.44, 0.29], [0.43, 0.31], [0.41, 0.2], [0.46, 0.38], [0.45, 0.42], [0.44, 0.38], [0.44, 0.38]]}" + } + ], + "images": "83180_exterior_image_1_left#0", + "gt": "[[70.5, 5.5], [115.5, 1.5], [128.5, 38.5], [140.5, 51.5], [137.5, 55.5], [130.5, 35.5], [146.5, 68.5], [145.5, 75.5], [141.5, 68.5], [141.5, 68.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg3_task_0071_user_0010_scene_0008_cfg_0003_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close a sliding window', and now you are performing subtask 'push the sliding window to close.'\nFrom start point [0.38, 0.0], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.35, 0.03], [0.41, 0.01], [0.4, 0.13], [0.33, 0.28], [0.41, 0.16], [0.33, 0.29], [0.35, 0.21], [0.33, 0.31], [0.28, 0.4]]}" + } + ], + "images": "RH20T_cfg3_task_0071_user_0010_scene_0008_cfg_0003_104122062823#0", + "gt": "[[246, 0], [226, 12], [263, 3], [259, 48], [214, 102], [261, 57], [209, 104], [227, 75], [212, 111], [178, 144]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122061018#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the block on the surface, transfer the block from the surface to the other block, place the block on the other block.'\nYou begin at [0.57, 0.08]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.08], [0.55, 0.18], [0.55, 0.21], [0.53, 0.22], [0.53, 0.27], [0.51, 0.31], [0.52, 0.28], [0.54, 0.32], [0.55, 0.31], [0.58, 0.39]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122061018#0.jpg", + "gt": "[[367, 28], [349, 65], [354, 76], [342, 81], [339, 98], [328, 113], [334, 101], [344, 115], [351, 110], [370, 139]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "51385_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'straighten the cloth', and now you are performing subtask 'place the cloth on the tabletop.'\nStarting at [0.41, 0.48], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.48], [0.31, 0.49], [0.38, 0.53], [0.38, 0.26], [0.34, 0.13], [0.36, 0.49], [0.43, 0.54], [0.46, 0.53], [0.42, 0.41], [0.39, 0.39]]}" + } + ], + "images": "51385_exterior_image_2_left#182", + "gt": "[[130, 86], [99, 88], [123, 96], [122, 47], [110, 23], [115, 88], [136, 98], [146, 95], [134, 74], [124, 71]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "30342_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the garlic on the counter into the glass bowl one by one', and now you are performing subtask 'place the second garlic into the glass bowl.'\nYou start from [0.48, 0.53]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.53], [0.48, 0.51], [0.53, 0.53], [0.56, 0.58], [0.55, 0.61], [0.56, 0.63], [0.54, 0.55], [0.51, 0.51], [0.48, 0.54], [0.49, 0.51]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/30342_exterior_image_2_left#146.jpg", + "gt": "[[154, 95], [155, 91], [170, 95], [178, 104], [176, 109], [178, 113], [173, 99], [163, 92], [153, 97], [156, 91]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "6770_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take out all the cloves of garlic in the bowl and put them on the table', and now you are performing subtask 'pick up the garlic in the bowl.'\nStarting at [0.68, 0.67], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.67], [0.63, 0.67], [0.61, 0.82], [0.62, 0.68], [0.65, 0.63], [0.67, 0.71], [0.68, 0.67], [0.66, 0.65], [0.62, 0.68], [0.63, 0.79]]}" + } + ], + "images": "6770_exterior_image_1_left#214", + "gt": "[[217, 121], [203, 120], [194, 148], [198, 123], [208, 114], [214, 127], [216, 121], [210, 117], [200, 123], [202, 142]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "81019_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the fruits on the bowl and put them on the counter', and now you are performing subtask 'transfer the third fruit from the bowl to the countertop.'\nFrom start point [0.78, 0.66], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.78, 0.66], [0.77, 0.58], [0.69, 0.56], [0.67, 0.59], [0.65, 0.67], [0.73, 0.52], [0.78, 0.58], [0.79, 0.66], [0.78, 0.69], [0.78, 0.64]]}" + } + ], + "images": "81019_exterior_image_1_left#275", + "gt": "[[251, 118], [246, 105], [221, 101], [213, 106], [209, 120], [233, 94], [251, 104], [254, 119], [251, 124], [250, 116]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0026_user_0014_scene_0007_cfg_0006_104122063633#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the gripper to push and close the drawer', and now you are performing subtask 'use the gripper to push and close the drawer.'\nStart at [0.36, 0.33]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.33], [0.33, 0.4], [0.32, 0.47], [0.33, 0.51], [0.34, 0.58], [0.35, 0.6], [0.37, 0.66], [0.35, 0.55], [0.36, 0.43], [0.37, 0.42]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0026_user_0014_scene_0007_cfg_0006_104122063633#0.jpg", + "gt": "[[233, 118], [210, 143], [202, 170], [211, 185], [215, 210], [223, 215], [237, 236], [227, 198], [231, 155], [236, 150]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0003_user_0015_scene_0008_cfg_0001_038522062547#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the three buttons from left to right in sequence', and now you are performing subtask 'press the orange button on the left, press the yellow button in the middle.'\nBegin at [0.65, 0.33] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.33], [0.65, 0.28], [0.61, 0.31], [0.59, 0.31], [0.59, 0.32], [0.59, 0.3], [0.58, 0.26], [0.6, 0.25], [0.59, 0.22], [0.59, 0.14]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0003_user_0015_scene_0008_cfg_0001_038522062547#156.jpg", + "gt": "[[419, 119], [413, 101], [392, 112], [378, 111], [378, 117], [380, 107], [372, 94], [384, 90], [376, 78], [380, 50]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0023_user_0010_scene_0001_cfg_0003_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the power strip by pressing the button', and now you are performing subtask 'turn on the power strip by pressing the button.'\nYour initial coordinate is [0.38, 0.0]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.35, 0.01], [0.36, 0.05], [0.46, 0.13], [0.46, 0.19], [0.45, 0.21], [0.44, 0.2], [0.44, 0.17], [0.46, 0.28], [0.46, 0.18]]}" + } + ], + "images": "RH20T_cfg3_task_0023_user_0010_scene_0001_cfg_0003_104122062823#0", + "gt": "[[246, 0], [224, 3], [228, 19], [297, 48], [296, 67], [287, 77], [279, 71], [283, 62], [292, 101], [293, 64]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "31421_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the blue object on the dish rack', and now you are performing subtask 'put the blue object on the dish rack.'\nBegin at [0.72, 0.19] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.72, 0.19], [0.68, 0.22], [0.7, 0.56], [0.66, 0.53], [0.68, 0.6], [0.71, 0.48], [0.73, 0.59], [0.67, 0.32], [0.52, 0.38], [0.66, 0.26]]}" + } + ], + "images": "31421_exterior_image_1_left#0", + "gt": "[[231, 34], [217, 40], [223, 100], [210, 95], [218, 108], [227, 87], [234, 107], [213, 58], [165, 69], [211, 46]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "49096_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the pen on the left side of the bowl', and now you are performing subtask 'put the pen on the left side of the bowl.'\nFrom start point [0.5, 0.02], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.02], [0.54, 0.15], [0.53, 0.25], [0.57, 0.48], [0.56, 0.54], [0.59, 0.45], [0.59, 0.35], [0.56, 0.25], [0.54, 0.32], [0.55, 0.24]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/49096_exterior_image_1_left#0.jpg", + "gt": "[[159.5, 3.5], [171.5, 27.5], [170.5, 45.5], [183.5, 86.5], [180.5, 96.5], [189.5, 80.5], [189.5, 62.5], [179.5, 44.5], [171.5, 58.5], [177.5, 42.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "72433_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the lid', and now you are performing subtask 'close the lid.'\nYour starting position is [0.94, 0.01]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.94, 0.01], [0.96, 0.06], [1.0, 0.06], [1.0, 0.09], [0.9, 0.18], [0.88, 0.29], [0.81, 0.36], [0.79, 0.42], [0.77, 0.42], [0.82, 0.14]]}" + } + ], + "images": "72433_exterior_image_1_left#0", + "gt": "[[300, 2], [306, 10], [320, 11], [320, 17], [288, 33], [282, 53], [258, 64], [253, 75], [247, 76], [262, 25]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0003_user_0006_scene_0007_cfg_0002_104422070044#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the three buttons in sequence', and now you are performing subtask 'press the gray base red button, press the yellow base green button.'\nWith start location [0.49, 0.56], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.56], [0.47, 0.5], [0.46, 0.49], [0.46, 0.48], [0.46, 0.52], [0.45, 0.49], [0.44, 0.49], [0.44, 0.5], [0.45, 0.55], [0.46, 0.55]]}" + } + ], + "images": "RH20T_cfg2_task_0003_user_0006_scene_0007_cfg_0002_104422070044#175", + "gt": "[[315, 201], [304, 181], [292, 177], [293, 174], [293, 187], [287, 177], [284, 175], [282, 179], [285, 197], [292, 198]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0029_user_0014_scene_0005_cfg_0006_037522061512#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer', and now you are performing subtask 'open the drawer.'\nWith start location [0.4, 0.23], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.23], [0.39, 0.21], [0.4, 0.21], [0.43, 0.24], [0.47, 0.22], [0.47, 0.24], [0.44, 0.33], [0.36, 0.41], [0.36, 0.38], [0.41, 0.32]]}" + } + ], + "images": "RH20T_cfg6_task_0029_user_0014_scene_0005_cfg_0006_037522061512#0", + "gt": "[[257, 83], [250, 77], [256, 75], [272, 85], [298, 80], [303, 88], [281, 118], [231, 147], [233, 135], [262, 117]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "16826_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'sliding cover', and now you are performing subtask 'sliding cover.'\nWith start location [0.29, 0.42], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.29, 0.42], [0.5, 0.17], [0.59, 0.11], [0.57, 0.16], [0.62, 0.2], [0.65, 0.41], [0.63, 0.59], [0.62, 0.61], [0.63, 0.53], [0.62, 0.38]]}" + } + ], + "images": "16826_exterior_image_2_left#0", + "gt": "[[93, 76], [161, 31], [190, 19], [183, 28], [199, 36], [208, 74], [203, 106], [198, 109], [202, 95], [197, 69]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0101_user_0007_scene_0010_cfg_0005_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'clean the table with a cloth', and now you are performing subtask 'clean the table with a cloth.'\nYou start from [0.45, 0.3]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.3], [0.46, 0.27], [0.49, 0.33], [0.47, 0.43], [0.4, 0.45], [0.47, 0.41], [0.4, 0.44], [0.39, 0.42], [0.44, 0.34], [0.47, 0.33]]}" + } + ], + "images": "RH20T_cfg5_task_0101_user_0007_scene_0010_cfg_0005_036422060215#0", + "gt": "[[286, 108], [292, 97], [311, 119], [300, 156], [257, 162], [304, 147], [259, 158], [248, 152], [280, 123], [304, 119]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0003_user_0014_scene_0010_cfg_0006_104122063633#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the green button in the middle, press the yellow button at the top.'\nYou begin at [0.3, 0.38]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.3, 0.38], [0.29, 0.38], [0.28, 0.4], [0.28, 0.42], [0.27, 0.48], [0.27, 0.44], [0.28, 0.41], [0.28, 0.41], [0.28, 0.41], [0.28, 0.42]]}" + } + ], + "images": "RH20T_cfg6_task_0003_user_0014_scene_0010_cfg_0006_104122063633#178", + "gt": "[[193, 135], [185, 135], [180, 143], [180, 151], [175, 172], [173, 160], [176, 149], [180, 149], [180, 148], [178, 151]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0004_user_0010_scene_0001_cfg_0005_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the small block on the left and move it to the right', and now you are performing subtask 'pick up the small block on the left and move it to the right.'\nStart at [0.45, 0.3]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.3], [0.45, 0.32], [0.44, 0.29], [0.48, 0.25], [0.55, 0.26], [0.55, 0.42], [0.51, 0.35], [0.46, 0.38], [0.42, 0.47], [0.36, 0.51]]}" + } + ], + "images": "RH20T_cfg5_task_0004_user_0010_scene_0001_cfg_0005_036422060215#0", + "gt": "[[286, 108], [285, 116], [283, 105], [310, 91], [351, 93], [349, 152], [328, 127], [297, 137], [270, 169], [231, 185]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "38730_exterior_image_2_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the towel to wipe on the table', and now you are performing subtask 'slide the towel to wipe the surface on the table.'\nThe trajectory starts at [0.67, 0.46]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.67, 0.46], [0.68, 0.42], [0.69, 0.44], [0.67, 0.52], [0.65, 0.54], [0.54, 0.58], [0.54, 0.56], [0.55, 0.52], [0.57, 0.52], [0.55, 0.52]]}" + } + ], + "images": "38730_exterior_image_2_left#156", + "gt": "[[213, 82], [219, 76], [221, 80], [214, 94], [209, 97], [173, 104], [173, 100], [177, 93], [181, 93], [177, 94]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0037_user_0007_scene_0006_cfg_0002_037522062165#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the second block on the table, transfer the second block from the table to the top of the third block, place the second block on the third block, pick up the third block on the table.'\nBegin at [0.34, 0.76] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.34, 0.76], [0.34, 0.76], [0.33, 0.77], [0.33, 0.73], [0.38, 0.71], [0.39, 0.71], [0.38, 0.72], [0.38, 0.73], [0.42, 0.72], [0.43, 0.76]]}" + } + ], + "images": "RH20T_cfg2_task_0037_user_0007_scene_0006_cfg_0002_037522062165#375", + "gt": "[[219, 275], [216, 272], [213, 278], [211, 264], [243, 256], [248, 257], [243, 258], [244, 261], [269, 260], [275, 275]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "63429_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker in the cup', and now you are performing subtask 'the marker falls on the printer.'\nYou begin at [0.56, 0.65]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.56, 0.65], [0.53, 0.59], [0.5, 0.58], [0.47, 0.63], [0.47, 0.68], [0.45, 0.66], [0.42, 0.61], [0.37, 0.55], [0.32, 0.57], [0.35, 0.57]]}" + } + ], + "images": "63429_exterior_image_1_left#0", + "gt": "[[179, 117], [169, 106], [159, 104], [151, 114], [152, 123], [143, 119], [135, 110], [119, 99], [104, 102], [112, 102]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0089_user_0020_scene_0009_cfg_0006_104122064161#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push a block with obstacles', and now you are performing subtask 'push a block forward, push a block to the right.'\nYour starting position is [0.56, 0.35]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.56, 0.35], [0.6, 0.36], [0.6, 0.33], [0.61, 0.32], [0.62, 0.35], [0.65, 0.43], [0.66, 0.47], [0.67, 0.48], [0.67, 0.47], [0.67, 0.47]]}" + } + ], + "images": "RH20T_cfg6_task_0089_user_0020_scene_0009_cfg_0006_104122064161#184", + "gt": "[[360, 125], [387, 130], [387, 120], [391, 114], [395, 126], [419, 154], [424, 168], [428, 173], [427, 170], [430, 169]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "4483_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the blue marker in the yellow mug', and now you are performing subtask 'put the blue marker in the yellow mug.'\nYour initial coordinate is [0.42, 0.07]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.07], [0.46, 0.29], [0.43, 0.56], [0.45, 0.62], [0.47, 0.61], [0.46, 0.61], [0.46, 0.45], [0.43, 0.37], [0.4, 0.37], [0.39, 0.48]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/4483_exterior_image_2_left#0.jpg", + "gt": "[[136, 13], [148, 53], [137, 100], [143, 112], [149, 109], [146, 109], [146, 81], [137, 67], [127, 66], [124, 87]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "6755_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'the cloth on the table was folded twice in total', and now you are performing subtask 'the cloth on the table was folded twice in total.'\nYou start from [0.47, 0.61]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.61], [0.47, 0.75], [0.56, 0.64], [0.62, 0.68], [0.48, 0.63], [0.48, 0.75], [0.56, 0.86], [0.51, 0.85], [0.56, 0.62], [0.47, 0.59]]}" + } + ], + "images": "6755_exterior_image_2_left#0", + "gt": "[[151, 109], [150, 135], [178, 116], [198, 122], [154, 114], [154, 135], [178, 154], [163, 153], [178, 112], [150, 107]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "59693_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the brown cup from the edge of the countertop and put it in the sink', and now you are performing subtask 'pick up the brown cup from the edge of the countertop and put it in the sink.'\nWith start location [0.31, 0.38], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.31, 0.38], [0.39, 0.54], [0.46, 0.58], [0.46, 0.69], [0.49, 0.74], [0.52, 0.74], [0.49, 0.79], [0.52, 0.63], [0.58, 0.72], [0.48, 0.53]]}" + } + ], + "images": "59693_exterior_image_2_left#0", + "gt": "[[98, 69], [124, 97], [146, 105], [146, 124], [157, 134], [167, 134], [158, 143], [168, 113], [187, 130], [154, 95]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "20591_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the blue bowl on top of the air fryer then set it down again', and now you are performing subtask 'put the blue bowl on top of the air fryer then set it down again.'\nYour initial coordinate is [0.77, 0.49]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.77, 0.49], [0.67, 0.77], [0.66, 0.55], [0.59, 0.45], [0.71, 0.57], [0.68, 0.54], [0.73, 0.72], [0.72, 0.74], [0.69, 0.74], [0.68, 0.62]]}" + } + ], + "images": "20591_exterior_image_1_left#0", + "gt": "[[247, 88], [215, 138], [212, 99], [189, 81], [228, 102], [219, 98], [234, 129], [230, 134], [222, 133], [216, 111]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0003_user_0006_scene_0007_cfg_0002_037522062165#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the three buttons in sequence', and now you are performing subtask 'press the yellow base green button.'\nBegin at [0.43, 0.47] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.47], [0.45, 0.53], [0.49, 0.6], [0.49, 0.68], [0.5, 0.7], [0.5, 0.63], [0.51, 0.52], [0.5, 0.47], [0.48, 0.47], [0.48, 0.46]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg2_task_0003_user_0006_scene_0007_cfg_0002_037522062165#301.jpg", + "gt": "[[275, 169], [287, 192], [313, 217], [314, 243], [318, 251], [319, 226], [326, 189], [318, 171], [309, 170], [307, 166]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "74681_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the rectangular block from the stack and put it on the table', and now you are performing subtask 'take the rectangular block from the stack and put it on the table.'\nBegin at [0.54, 0.02] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.02], [0.57, 0.16], [0.66, 0.27], [0.63, 0.41], [0.5, 0.32], [0.46, 0.41], [0.47, 0.51], [0.48, 0.27], [0.54, 0.04], [0.54, 0.04]]}" + } + ], + "images": "74681_exterior_image_1_left#0", + "gt": "[[172, 3], [184, 28], [210, 49], [203, 74], [161, 57], [146, 74], [152, 92], [153, 49], [173, 8], [173, 8]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "4673_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'transfer the marker from the black cup to the table', and now you are performing subtask 'transfer the marker from the black cup to the table.'\nYou start from [0.48, 0.04]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.04], [0.41, 0.11], [0.42, 0.11], [0.43, 0.19], [0.43, 0.31], [0.47, 0.33], [0.45, 0.34], [0.44, 0.19], [0.51, 0.44], [0.51, 0.36]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/4673_exterior_image_1_left#0.jpg", + "gt": "[[153.0, 6.5], [132.0, 19.5], [135.0, 19.5], [139.0, 33.5], [139.0, 56.5], [149.0, 59.5], [144.0, 61.5], [141.0, 33.5], [163.0, 78.5], [163.0, 65.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0001_user_0014_scene_0006_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the button from top to bottom', and now you are performing subtask 'press the button from top to bottom.'\nYour initial coordinate is [0.63, 0.37]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.67, 0.28], [0.66, 0.19], [0.66, 0.16], [0.67, 0.16], [0.64, 0.17], [0.63, 0.2], [0.62, 0.17], [0.61, 0.19], [0.62, 0.11]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0001_user_0014_scene_0006_cfg_0001_038522062547#0.jpg", + "gt": "[[406, 133], [431, 100], [422, 69], [423, 58], [429, 58], [412, 60], [401, 73], [395, 63], [392, 68], [400, 38]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "86182_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the top drawer and put the plush toy inside', and now you are performing subtask 'pull the handle of the top drawer to open the drawer.'\nStarting at [0.65, 0.77], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.77], [0.64, 0.75], [0.62, 0.75], [0.58, 0.79], [0.58, 0.83], [0.61, 0.88], [0.64, 0.89], [0.69, 0.87], [0.77, 0.86], [0.64, 0.76]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/86182_exterior_image_1_left#240.jpg", + "gt": "[[207, 138], [204, 135], [199, 135], [187, 143], [186, 149], [195, 159], [205, 160], [221, 156], [246, 154], [205, 137]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0015_user_0014_scene_0008_cfg_0007_104122061018#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'play the drum', and now you are performing subtask 'play the drum.'\nThe trajectory starts at [0.57, 0.09]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.09], [0.49, 0.11], [0.41, 0.18], [0.4, 0.23], [0.39, 0.29], [0.43, 0.21], [0.5, 0.36], [0.46, 0.29], [0.47, 0.37], [0.52, 0.14]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0015_user_0014_scene_0008_cfg_0007_104122061018#0.jpg", + "gt": "[[363, 31], [313, 40], [264, 64], [253, 82], [251, 105], [276, 75], [323, 130], [293, 103], [300, 132], [331, 51]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "48589_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the pen from the table and put it in the white bowl', and now you are performing subtask 'pick up the pen from the table and put it in the white bowl.'\nFrom start point [0.31, 0.05], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.31, 0.05], [0.42, 0.18], [0.42, 0.33], [0.47, 0.47], [0.45, 0.51], [0.47, 0.41], [0.5, 0.41], [0.52, 0.44], [0.5, 0.37], [0.48, 0.38]]}" + } + ], + "images": "48589_exterior_image_1_left#0", + "gt": "[[100.0, 9.0], [134.0, 33.0], [135.0, 60.0], [152.0, 84.0], [145.0, 92.0], [150.0, 74.0], [159.0, 73.0], [165.0, 80.0], [161.0, 67.0], [155.0, 68.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "2321_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up and put down the toy', and now you are performing subtask 'pick up and put down the toy.'\nThe trajectory starts at [0.72, 0.26]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.72, 0.26], [0.64, 0.38], [0.53, 0.79], [0.47, 0.92], [0.47, 0.83], [0.5, 0.74], [0.5, 0.86], [0.51, 0.78], [0.64, 0.66], [0.66, 0.58]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/2321_exterior_image_2_left#0.jpg", + "gt": "[[229, 47], [206, 68], [171, 143], [149, 166], [151, 149], [160, 134], [160, 155], [162, 141], [204, 118], [211, 104]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0096_user_0014_scene_0001_cfg_0001_035622060973#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take something out of a drawer', and now you are performing subtask 'pick up the object from the drawer, transfer the object from the drawer to the top of the table, place the object on the table.'\nFrom start point [0.45, 0.5], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.5], [0.46, 0.5], [0.47, 0.54], [0.48, 0.52], [0.44, 0.42], [0.39, 0.39], [0.39, 0.44], [0.4, 0.46], [0.41, 0.39], [0.42, 0.36]]}" + } + ], + "images": "RH20T_cfg1_task_0096_user_0014_scene_0001_cfg_0001_035622060973#158", + "gt": "[[288, 180], [296, 181], [303, 195], [307, 189], [279, 152], [251, 141], [247, 159], [259, 165], [260, 142], [270, 131]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0044_user_0020_scene_0001_cfg_0007_104122063633#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up a bag of things', and now you are performing subtask 'pick up a bag of things.'\nYour starting position is [0.37, 0.34]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.34], [0.35, 0.39], [0.34, 0.4], [0.33, 0.43], [0.33, 0.45], [0.33, 0.46], [0.34, 0.46], [0.34, 0.41], [0.34, 0.37], [0.35, 0.36]]}" + } + ], + "images": "RH20T_cfg7_task_0044_user_0020_scene_0001_cfg_0007_104122063633#0", + "gt": "[[234, 121], [227, 139], [215, 145], [212, 154], [212, 163], [214, 166], [216, 164], [219, 147], [217, 133], [226, 128]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0029_user_0020_scene_0010_cfg_0007_104122061018#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp the handle and open the drawer', and now you are performing subtask 'grasp the handle and open the drawer.'\nThe trajectory starts at [0.57, 0.09]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.09], [0.53, 0.22], [0.55, 0.27], [0.56, 0.36], [0.49, 0.49], [0.48, 0.57], [0.43, 0.38], [0.43, 0.31], [0.46, 0.19], [0.53, 0.1]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0029_user_0020_scene_0010_cfg_0007_104122061018#0.jpg", + "gt": "[[363, 31], [339, 79], [354, 98], [356, 128], [311, 176], [305, 205], [275, 135], [275, 111], [293, 68], [337, 36]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0061_user_0007_scene_0010_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'pick up the object in the container, transfer the object from the container to the table, place the object on the table.'\nYour initial coordinate is [0.5, 0.37]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.52, 0.39], [0.52, 0.5], [0.5, 0.56], [0.51, 0.48], [0.49, 0.41], [0.47, 0.39], [0.45, 0.46], [0.47, 0.39], [0.47, 0.39]]}" + } + ], + "images": "RH20T_cfg5_task_0061_user_0007_scene_0010_cfg_0005_104122063678#0", + "gt": "[[323, 132], [333, 141], [334, 179], [319, 202], [324, 172], [312, 149], [301, 141], [291, 164], [300, 140], [300, 140]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0039_user_0014_scene_0002_cfg_0006_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pour the water from one cup into another empty cup', and now you are performing subtask 'pour the water from one cup into another empty cup.'\nThe trajectory starts at [0.53, 0.31]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.31], [0.53, 0.46], [0.53, 0.29], [0.55, 0.19], [0.54, 0.07], [0.54, 0.13], [0.56, 0.11], [0.57, 0.29], [0.54, 0.37], [0.53, 0.22]]}" + } + ], + "images": "RH20T_cfg6_task_0039_user_0014_scene_0002_cfg_0006_104122064161#0", + "gt": "[[338, 110], [340, 165], [339, 103], [351, 69], [343, 26], [348, 48], [361, 39], [366, 106], [344, 132], [342, 81]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0003_user_0006_scene_0007_cfg_0002_105422061350#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the three buttons in sequence', and now you are performing subtask 'press the gray base red button, press the yellow base green button.'\nWith start location [0.59, 0.44], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.44], [0.57, 0.43], [0.54, 0.45], [0.51, 0.5], [0.49, 0.51], [0.47, 0.57], [0.53, 0.51], [0.58, 0.53], [0.59, 0.52], [0.6, 0.49]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg2_task_0003_user_0006_scene_0007_cfg_0002_105422061350#275.jpg", + "gt": "[[378, 158], [362, 156], [345, 162], [326, 181], [311, 184], [303, 206], [337, 182], [369, 190], [378, 188], [382, 177]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0066_user_0014_scene_0007_cfg_0007_104122061018#TrajectoryPrediction_Task#12", + "task": "TrajectoryPrediction_Task", + "p_index": 12, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'pick up the building blocks on the table, transfer the building blocks from the desktop to the pile of building blocks on the building board, place the building blocks on the pile of building blocks on the building board.'\nWith start location [0.42, 0.42], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.43], [0.38, 0.54], [0.41, 0.65], [0.39, 0.54], [0.44, 0.08], [0.47, 0.06], [0.48, 0.15], [0.46, 0.11], [0.48, 0.04], [0.5, 0.06]]}" + } + ], + "images": "RH20T_cfg7_task_0066_user_0014_scene_0007_cfg_0007_104122061018#964", + "gt": "[[270, 153], [240, 193], [260, 233], [249, 195], [279, 29], [298, 22], [305, 55], [297, 40], [309, 15], [323, 23]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "62551_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the faucet, open the faucet, close the faucet', and now you are performing subtask 'press the faucet down.'\nStarting at [0.59, 0.2], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.2], [0.55, 0.22], [0.38, 0.44], [0.27, 0.46], [0.25, 0.42], [0.2, 0.41], [0.21, 0.47], [0.25, 0.49], [0.29, 0.53], [0.29, 0.44]]}" + } + ], + "images": "62551_exterior_image_1_left#0", + "gt": "[[188, 36], [177, 40], [121, 79], [87, 83], [79, 75], [64, 73], [67, 84], [80, 88], [93, 96], [92, 80]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "21793_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn the caps on the bed', and now you are performing subtask 'pick up the black and white hat on the bed.'\nYou begin at [0.33, 0.62]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.33, 0.62], [0.38, 0.7], [0.36, 0.73], [0.27, 0.59], [0.12, 0.62], [0.17, 0.63], [0.19, 0.77], [0.23, 0.74], [0.21, 0.72], [0.21, 0.72]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/21793_exterior_image_1_left#174.jpg", + "gt": "[[106, 112], [123, 126], [115, 131], [86, 106], [39, 111], [54, 113], [60, 138], [73, 134], [67, 129], [67, 129]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0037_user_0007_scene_0006_cfg_0002_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the first block on the table, transfer the first block from the table to the top of the second block, place the first block on the second block.'\nYour starting position is [0.45, 0.55]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.55], [0.56, 0.64], [0.56, 0.66], [0.6, 0.64], [0.6, 0.68], [0.58, 0.73], [0.58, 0.67], [0.57, 0.64], [0.54, 0.59], [0.54, 0.61]]}" + } + ], + "images": "RH20T_cfg2_task_0037_user_0007_scene_0006_cfg_0002_104122063678#0", + "gt": "[[289, 199], [359, 229], [360, 237], [382, 232], [383, 246], [372, 263], [374, 242], [366, 229], [345, 213], [343, 221]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122061330#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'place the object on the surface, pick up the object in the container, transfer the object from the container to the surface, place the item on the table.'\nYour starting position is [0.52, 0.57]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.57], [0.5, 0.43], [0.52, 0.36], [0.53, 0.38], [0.52, 0.41], [0.51, 0.35], [0.51, 0.35], [0.5, 0.36], [0.47, 0.45], [0.47, 0.52]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122061330#316.jpg", + "gt": "[[334, 206], [323, 155], [331, 131], [337, 137], [334, 148], [327, 127], [324, 125], [318, 130], [300, 161], [304, 186]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "511_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the box slightly to the center of the table', and now you are performing subtask 'move the box slightly to the center of the table.'\nYou begin at [0.42, 0.36]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.36], [0.45, 0.48], [0.43, 0.67], [0.47, 0.72], [0.48, 0.79], [0.59, 0.79], [0.58, 0.62], [0.57, 0.57], [0.58, 0.58], [0.57, 0.57]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/511_exterior_image_2_left#0.jpg", + "gt": "[[135, 64], [143, 86], [137, 121], [150, 130], [153, 142], [188, 142], [186, 112], [181, 102], [187, 105], [183, 103]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "74509_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the peg from the table and hang it on the black stand', and now you are performing subtask 'take the peg from the table and hang it on the black stand.'\nWith start location [0.85, 0.38], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.85, 0.38], [0.87, 0.48], [0.82, 0.51], [0.78, 0.66], [0.72, 0.73], [0.74, 0.54], [0.74, 0.47], [0.71, 0.51], [0.74, 0.37], [0.78, 0.39]]}" + } + ], + "images": "74509_exterior_image_1_left#0", + "gt": "[[272.0, 68.0], [278.0, 86.0], [263.0, 91.0], [251.0, 119.0], [230.0, 132.0], [238.0, 98.0], [236.0, 84.0], [226.0, 92.0], [236.0, 67.0], [249.0, 71.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "90700_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the towel and the shirt off the back of the chair and put them on the seat', and now you are performing subtask 'throw the towel onto the seat of the chair.'\nFrom start point [0.52, 0.42], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.42], [0.62, 0.21], [0.59, 0.27], [0.7, 0.51], [0.75, 0.54], [0.81, 0.76], [0.81, 0.52], [0.77, 0.38], [0.76, 0.38], [0.77, 0.46]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/90700_exterior_image_2_left#0.jpg", + "gt": "[[165, 76], [199, 38], [188, 49], [223, 91], [240, 97], [259, 136], [258, 94], [246, 68], [242, 68], [245, 83]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "2362_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the wallet from the table and put it in the black bag', and now you are performing subtask 'place the wallet in the black bag.'\nYour initial coordinate is [0.66, 0.68]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.66, 0.68], [0.64, 0.74], [0.67, 0.77], [0.68, 0.7], [0.61, 0.41], [0.49, 0.46], [0.48, 0.55], [0.53, 0.41], [0.61, 0.47], [0.59, 0.49]]}" + } + ], + "images": "2362_exterior_image_1_left#468", + "gt": "[[210.0, 121.5], [206.0, 133.5], [213.0, 138.5], [217.0, 125.5], [196.0, 73.5], [157.0, 83.5], [155.0, 98.5], [171.0, 74.5], [196.0, 85.5], [190.0, 87.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0085_user_0007_scene_0002_cfg_0005_104122063678#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the bubble ring from the assembled bubble ring and ball', and now you are performing subtask 'pick up the bubble ring to separate it from the assembled bubble ring with the ball, place the bubble ring on the table.'\nBegin at [0.54, 0.53] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.53], [0.53, 0.54], [0.54, 0.55], [0.53, 0.55], [0.53, 0.53], [0.53, 0.51], [0.53, 0.5], [0.51, 0.5], [0.51, 0.47], [0.51, 0.47]]}" + } + ], + "images": "RH20T_cfg5_task_0085_user_0007_scene_0002_cfg_0005_104122063678#396", + "gt": "[[344, 190], [342, 196], [343, 199], [339, 198], [338, 192], [341, 184], [340, 181], [328, 180], [324, 169], [324, 169]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "66631_exterior_image_2_left#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the towel to wipe the table', and now you are performing subtask 'wipe the table.'\nYour starting position is [0.59, 0.52]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.52], [0.58, 0.5], [0.57, 0.54], [0.57, 0.59], [0.58, 0.62], [0.66, 0.64], [0.67, 0.63], [0.66, 0.61], [0.64, 0.6], [0.59, 0.61]]}" + } + ], + "images": "66631_exterior_image_2_left#298", + "gt": "[[189, 94], [187, 90], [181, 98], [182, 106], [187, 112], [210, 115], [213, 113], [212, 110], [205, 108], [188, 109]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "31231_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the marker to the left', and now you are performing subtask 'move the marker to the left.'\nStarting at [0.6, 0.0], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.0], [0.49, 0.23], [0.49, 0.24], [0.5, 0.23], [0.52, 0.34], [0.58, 0.39], [0.51, 0.42], [0.46, 0.47], [0.45, 0.39], [0.45, 0.39]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/31231_exterior_image_1_left#0.jpg", + "gt": "[[191, 0], [156, 41], [156, 44], [160, 42], [167, 61], [185, 71], [163, 75], [147, 85], [144, 71], [144, 71]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "33104_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the top drawer of the gray file cabinet', and now you are performing subtask 'close the top drawer of the gray file cabinet.'\nThe trajectory starts at [0.44, 0.72]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.72], [0.44, 0.67], [0.44, 0.66], [0.44, 0.53], [0.44, 0.51], [0.43, 0.48], [0.43, 0.48], [0.43, 0.45], [0.44, 0.43], [0.47, 0.43]]}" + } + ], + "images": "33104_exterior_image_2_left#0", + "gt": "[[142, 130], [142, 121], [141, 119], [142, 95], [140, 92], [139, 87], [137, 86], [136, 81], [140, 77], [149, 78]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "88746_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the grey toy on the plate and put it in the sink', and now you are performing subtask 'pick up the grey toy on the plate and put it in the sink.'\nFrom start point [0.76, 0.0], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.76, 0.0], [0.74, 0.09], [0.71, 0.08], [0.72, 0.18], [0.68, 0.29], [0.73, 0.36], [0.78, 0.26], [0.77, 0.31], [0.72, 0.31], [0.75, 0.31]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/88746_exterior_image_2_left#0.jpg", + "gt": "[[243, 0], [236, 17], [226, 14], [231, 33], [218, 52], [234, 65], [251, 47], [248, 56], [229, 55], [240, 56]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "19986_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the green object in the bowl', and now you are performing subtask 'put the green object in the bowl.'\nFrom start point [0.63, 0.0], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.0], [0.64, 0.1], [0.68, 0.24], [0.67, 0.34], [0.64, 0.39], [0.66, 0.28], [0.7, 0.3], [0.72, 0.37], [0.68, 0.59], [0.67, 0.39]]}" + } + ], + "images": "19986_exterior_image_1_left#0", + "gt": "[[201, 0], [205, 18], [218, 43], [214, 62], [206, 71], [212, 51], [225, 54], [231, 66], [217, 106], [215, 71]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "89639_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the green object from the sink to the left counter', and now you are performing subtask 'move the green object from the sink to the left counter.'\nStarting at [0.3, 0.53], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.3, 0.53], [0.43, 0.63], [0.53, 0.97], [0.56, 1.0], [0.56, 0.89], [0.52, 0.83], [0.47, 0.82], [0.28, 0.88], [0.32, 0.79], [0.37, 0.81]]}" + } + ], + "images": "89639_exterior_image_1_left#0", + "gt": "[[96.0, 95.0], [136.0, 113.0], [169.0, 175.0], [180.0, 180.0], [179.0, 161.0], [167.0, 149.0], [150.0, 148.0], [91.0, 158.0], [102.0, 143.0], [118.0, 145.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "88632_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take one tea bag out of the open box on the left side of the countertop', and now you are performing subtask 'take one tea bag out of the open box on the left side of the countertop.'\nStart at [0.7, 0.51]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.51], [0.71, 0.41], [0.7, 0.48], [0.7, 0.41], [0.6, 0.38], [0.52, 0.57], [0.55, 0.59], [0.53, 0.48], [0.56, 0.61], [0.51, 0.57]]}" + } + ], + "images": "88632_exterior_image_2_left#0", + "gt": "[[223, 92], [226, 74], [225, 86], [223, 74], [192, 68], [168, 103], [177, 107], [170, 87], [180, 110], [162, 103]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "44786_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the top drawer of the white file cabinet', and now you are performing subtask 'close the top drawer of the white file cabinet.'\nBegin at [0.46, 0.68] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.68], [0.46, 0.65], [0.45, 0.58], [0.45, 0.56], [0.44, 0.52], [0.45, 0.44], [0.46, 0.42], [0.5, 0.41], [0.5, 0.41], [0.51, 0.41]]}" + } + ], + "images": "44786_exterior_image_2_left#0", + "gt": "[[147, 122], [148, 117], [145, 105], [145, 100], [142, 93], [143, 80], [146, 75], [159, 74], [159, 73], [163, 74]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "33918_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the mug cup to the right', and now you are performing subtask 'move the mug cup to the right.'\nThe trajectory starts at [0.33, 0.01]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.33, 0.01], [0.41, 0.13], [0.45, 0.22], [0.45, 0.47], [0.5, 0.52], [0.5, 0.57], [0.52, 0.47], [0.51, 0.45], [0.5, 0.5], [0.51, 0.38]]}" + } + ], + "images": "33918_exterior_image_1_left#0", + "gt": "[[104.5, 1.5], [132.5, 23.5], [144.5, 39.5], [144.5, 85.5], [158.5, 94.5], [161.5, 102.5], [165.5, 84.5], [162.5, 81.5], [159.5, 89.5], [162.5, 68.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg3_task_0003_user_0016_scene_0002_cfg_0003_036422060909#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the three buttons in sequence', and now you are performing subtask 'press the green button on the right, press the green button in the middle.'\nThe trajectory starts at [0.65, 0.25]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.25], [0.63, 0.26], [0.61, 0.3], [0.59, 0.32], [0.58, 0.31], [0.55, 0.34], [0.55, 0.35], [0.54, 0.38], [0.53, 0.39], [0.54, 0.37]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg3_task_0003_user_0016_scene_0002_cfg_0003_036422060909#151.jpg", + "gt": "[[413, 89], [401, 94], [389, 108], [380, 114], [371, 113], [355, 122], [352, 125], [347, 136], [342, 140], [345, 132]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "62491_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn on the switch on the socket, open the box', and now you are performing subtask 'pull the lid on the box.'\nThe trajectory starts at [0.65, 0.0]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.0], [0.62, 0.18], [0.63, 0.25], [0.61, 0.45], [0.62, 0.54], [0.61, 0.43], [0.57, 0.35], [0.54, 0.34], [0.51, 0.36], [0.51, 0.36]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/62491_exterior_image_1_left#0.jpg", + "gt": "[[209, 0], [200, 33], [202, 45], [194, 81], [199, 97], [194, 77], [183, 63], [172, 62], [163, 65], [163, 65]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0129_user_0020_scene_0009_cfg_0006_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'plug in the bulb holder with a bulb to the socket', and now you are performing subtask 'plug in the bulb holder with a bulb to the socket.'\nFrom start point [0.53, 0.31], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.31], [0.55, 0.25], [0.57, 0.28], [0.6, 0.26], [0.62, 0.28], [0.6, 0.36], [0.62, 0.24], [0.65, 0.5], [0.65, 0.36], [0.57, 0.35]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0129_user_0020_scene_0009_cfg_0006_104122064161#0.jpg", + "gt": "[[338, 111], [351, 89], [364, 102], [384, 94], [396, 99], [385, 129], [395, 87], [414, 179], [414, 129], [365, 126]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "57351_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the top drawer of the white file cabinet', and now you are performing subtask 'close the top drawer of the white file cabinet.'\nFrom start point [0.78, 1.0], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.78, 1.0], [0.76, 0.91], [0.74, 0.74], [0.72, 0.68], [0.68, 0.64], [0.61, 0.61], [0.57, 0.57], [0.52, 0.56], [0.47, 0.53], [0.44, 0.53]]}" + } + ], + "images": "57351_exterior_image_1_left#0", + "gt": "[[250, 180], [244, 164], [237, 133], [229, 122], [217, 115], [195, 109], [181, 102], [166, 101], [149, 95], [140, 96]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122061602#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the block on the other block, pick up the block on the surface, transfer the block from the surface to the other block, place the block on the other block.'\nYour starting position is [0.62, 0.46]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.46], [0.64, 0.4], [0.62, 0.35], [0.61, 0.38], [0.61, 0.34], [0.6, 0.39], [0.6, 0.37], [0.61, 0.36], [0.63, 0.41], [0.62, 0.42]]}" + } + ], + "images": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122061602#432", + "gt": "[[397, 167], [407, 143], [394, 126], [391, 137], [389, 124], [385, 141], [383, 133], [389, 128], [402, 147], [398, 152]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "20007_exterior_image_1_left#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer on the right, take out three mugs and place them on the countertop', and now you are performing subtask 'place the third cup on the kitchen countertop.'\nYour initial coordinate is [0.65, 0.67]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.67], [0.67, 0.75], [0.64, 0.74], [0.69, 0.69], [0.68, 0.52], [0.64, 0.47], [0.57, 0.61], [0.6, 0.5], [0.68, 0.65], [0.68, 0.65]]}" + } + ], + "images": "20007_exterior_image_1_left#1100", + "gt": "[[209, 121], [214, 135], [205, 133], [222, 124], [217, 93], [206, 85], [182, 109], [193, 90], [219, 117], [219, 117]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "72865_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the middle washing machine door', and now you are performing subtask 'open the middle washing machine door.'\nStarting at [0.34, 0.22], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.34, 0.22], [0.37, 0.29], [0.41, 0.33], [0.44, 0.43], [0.46, 0.46], [0.47, 0.44], [0.48, 0.47], [0.41, 0.58], [0.43, 0.33], [0.38, 0.08]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/72865_exterior_image_1_left#0.jpg", + "gt": "[[109, 40], [119, 52], [132, 60], [141, 77], [147, 83], [151, 80], [154, 85], [132, 104], [137, 59], [123, 14]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "4936_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer, place the black clip in the open drawer', and now you are performing subtask 'place the black clip in the open drawer.'\nFrom start point [0.55, 0.53], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.53], [0.54, 0.47], [0.47, 0.36], [0.44, 0.36], [0.39, 0.41], [0.46, 0.4], [0.43, 0.41], [0.52, 0.47], [0.53, 0.53], [0.53, 0.37]]}" + } + ], + "images": "4936_exterior_image_2_left#192", + "gt": "[[176, 95], [172, 85], [151, 65], [140, 65], [124, 73], [146, 72], [137, 73], [165, 85], [169, 95], [171, 67]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0018_user_0008_scene_0007_cfg_0001_039422060546#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'play jenga', and now you are performing subtask 'play jenga.'\nYou start from [0.42, 0.33]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.33], [0.49, 0.36], [0.53, 0.35], [0.51, 0.34], [0.52, 0.35], [0.51, 0.4], [0.5, 0.42], [0.43, 0.41], [0.42, 0.43], [0.41, 0.39]]}" + } + ], + "images": "RH20T_cfg1_task_0018_user_0008_scene_0007_cfg_0001_039422060546#0", + "gt": "[[266, 120], [315, 128], [339, 126], [328, 124], [331, 127], [327, 144], [318, 152], [273, 147], [271, 154], [265, 140]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "39944_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the mug to the left, then to the right, then back to the left', and now you are performing subtask 'push the cup to the right.'\nYour initial coordinate is [0.48, 0.26]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.26], [0.49, 0.23], [0.54, 0.19], [0.56, 0.21], [0.56, 0.26], [0.53, 0.26], [0.52, 0.33], [0.52, 0.23], [0.51, 0.17], [0.51, 0.17]]}" + } + ], + "images": "39944_exterior_image_1_left#101", + "gt": "[[153, 47], [158, 41], [173, 35], [180, 38], [178, 47], [169, 47], [165, 59], [167, 41], [164, 30], [164, 30]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "52870_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the lemons and an apple out of the glass bowl and place them on the countertop', and now you are performing subtask 'place the third lemon on the tabletop.'\nYour starting position is [0.65, 0.58]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.58], [0.68, 0.51], [0.71, 0.49], [0.77, 0.58], [0.74, 0.66], [0.73, 0.52], [0.71, 0.49], [0.64, 0.53], [0.56, 0.73], [0.56, 0.66]]}" + } + ], + "images": "52870_exterior_image_1_left#225", + "gt": "[[207, 104], [219, 92], [227, 89], [245, 104], [238, 119], [234, 94], [226, 88], [204, 96], [179, 131], [180, 119]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "44868_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the yellow and brown bottle to the right and then move the water bottle backward', and now you are performing subtask 'pick up the water bottle on the table.'\nYour starting position is [0.6, 0.68]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.68], [0.63, 0.73], [0.62, 0.76], [0.48, 0.74], [0.51, 0.75], [0.54, 0.59], [0.67, 0.72], [0.63, 0.65], [0.57, 0.61], [0.57, 0.61]]}" + } + ], + "images": "44868_exterior_image_1_left#0", + "gt": "[[193, 123], [201, 132], [197, 137], [154, 133], [164, 135], [173, 106], [214, 129], [203, 117], [181, 109], [181, 109]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0036_user_0014_scene_0005_cfg_0007_104122061602#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'shake the test tube', and now you are performing subtask 'shake the test tube.'\nYour initial coordinate is [0.7, 0.32]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.32], [0.77, 0.28], [0.72, 0.3], [0.73, 0.36], [0.71, 0.43], [0.71, 0.28], [0.66, 0.22], [0.73, 0.31], [0.73, 0.42], [0.71, 0.3]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0036_user_0014_scene_0005_cfg_0007_104122061602#0.jpg", + "gt": "[[448, 116], [491, 100], [460, 107], [469, 129], [456, 153], [457, 100], [425, 81], [465, 110], [464, 150], [456, 107]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0062_user_0007_scene_0001_cfg_0002_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the toilet paper on its holder', and now you are performing subtask 'put the toilet paper on its holder.'\nStarting at [0.53, 0.3], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.3], [0.68, 0.42], [0.69, 0.49], [0.67, 0.38], [0.63, 0.29], [0.62, 0.31], [0.59, 0.2], [0.62, 0.29], [0.62, 0.35], [0.62, 0.35]]}" + } + ], + "images": "RH20T_cfg2_task_0062_user_0007_scene_0001_cfg_0002_104122061850#0", + "gt": "[[339, 107], [438, 151], [441, 176], [431, 137], [404, 103], [394, 111], [377, 71], [396, 103], [398, 125], [398, 125]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "61866_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move one white object from the right heap to the left, then get the blue-white object and fold it on top of the left heap', and now you are performing subtask 'fold the blue-white object at the top of the left heap.'\nBegin at [0.49, 0.29] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.29], [0.48, 0.3], [0.49, 0.49], [0.53, 0.56], [0.52, 0.58], [0.53, 0.38], [0.52, 0.29], [0.48, 0.25], [0.44, 0.26], [0.42, 0.29]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/61866_exterior_image_2_left#195.jpg", + "gt": "[[157, 53], [153, 54], [158, 89], [170, 100], [167, 105], [170, 69], [166, 52], [153, 45], [140, 46], [133, 52]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "35517_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the red object to the left', and now you are performing subtask 'move the red object to the left.'\nWith start location [0.16, 0.19], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.16, 0.19], [0.45, 0.21], [0.53, 0.3], [0.53, 0.41], [0.43, 0.35], [0.34, 0.35], [0.35, 0.31], [0.37, 0.29], [0.41, 0.32], [0.39, 0.26]]}" + } + ], + "images": "35517_exterior_image_1_left#0", + "gt": "[[51.5, 33.5], [143.5, 37.5], [170.5, 54.5], [168.5, 73.5], [138.5, 63.5], [109.5, 62.5], [110.5, 55.5], [117.5, 52.5], [130.5, 57.5], [124.5, 46.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "57652_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'hang up the cloth on the glass barrier', and now you are performing subtask 'pick up the cloth from the table.'\nWith start location [0.24, 0.26], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.24, 0.26], [0.3, 0.46], [0.42, 0.69], [0.43, 0.73], [0.4, 0.78], [0.4, 0.73], [0.38, 0.68], [0.37, 0.38], [0.37, 0.44], [0.44, 0.62]]}" + } + ], + "images": "57652_exterior_image_2_left#0", + "gt": "[[78, 46], [96, 83], [134, 124], [136, 131], [128, 141], [128, 132], [121, 122], [119, 69], [119, 79], [141, 111]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0048_user_0020_scene_0004_cfg_0007_104122064161#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'rotate the steering wheel clockwise', and now you are performing subtask 'twist the steering wheel clockwise.'\nYou start from [0.6, 0.55]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.55], [0.61, 0.53], [0.6, 0.56], [0.62, 0.57], [0.61, 0.56], [0.61, 0.57], [0.61, 0.56], [0.62, 0.5], [0.61, 0.44], [0.58, 0.42]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0048_user_0020_scene_0004_cfg_0007_104122064161#343.jpg", + "gt": "[[387, 197], [388, 192], [386, 200], [395, 205], [393, 203], [393, 205], [393, 202], [397, 179], [388, 160], [369, 150]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122063633#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the block on the other block, pick up the block on the surface, transfer the block from the surface to the other block, place the block on the other block.'\nYour starting position is [0.28, 0.53]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.28, 0.52], [0.27, 0.58], [0.28, 0.61], [0.28, 0.54], [0.3, 0.48], [0.34, 0.48], [0.37, 0.51], [0.35, 0.44], [0.31, 0.43], [0.29, 0.45]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122063633#743.jpg", + "gt": "[[179, 189], [175, 208], [182, 218], [178, 195], [192, 174], [215, 172], [238, 183], [221, 160], [199, 155], [184, 162]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_035622060973#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'place the second Megablock on the green base, pick up the third Megablock from the left side of the table, place the third Megablock on the green base.'\nYour initial coordinate is [0.38, 0.35]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.35], [0.32, 0.35], [0.29, 0.41], [0.33, 0.37], [0.37, 0.39], [0.37, 0.4], [0.38, 0.38], [0.41, 0.39], [0.41, 0.45], [0.42, 0.41]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_035622060973#333.jpg", + "gt": "[[241, 125], [207, 125], [188, 149], [209, 133], [239, 140], [237, 144], [242, 138], [261, 139], [263, 163], [269, 147]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0064_user_0001_scene_0008_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the roll of paper down from the shelf', and now you are performing subtask 'take the roll of paper down from the shelf.'\nThe trajectory starts at [0.63, 0.37]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.67, 0.43], [0.69, 0.54], [0.69, 0.58], [0.66, 0.61], [0.67, 0.58], [0.67, 0.61], [0.65, 0.64], [0.71, 0.39], [0.66, 0.38]]}" + } + ], + "images": "RH20T_cfg1_task_0064_user_0001_scene_0008_cfg_0001_038522062547#0", + "gt": "[[406, 134], [431, 154], [441, 193], [441, 209], [423, 220], [429, 209], [431, 220], [418, 230], [456, 140], [422, 138]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0008_user_0007_scene_0001_cfg_0005_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grab the block and place it at the designated location', and now you are performing subtask 'grab the block and place it at the designated location.'\nYour starting position is [0.49, 0.14]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.14], [0.49, 0.19], [0.43, 0.36], [0.28, 0.6], [0.36, 0.69], [0.34, 0.65], [0.39, 0.57], [0.49, 0.47], [0.54, 0.49], [0.47, 0.34]]}" + } + ], + "images": "RH20T_cfg5_task_0008_user_0007_scene_0001_cfg_0005_104122061850#0", + "gt": "[[311, 50], [313, 70], [272, 129], [177, 217], [229, 249], [217, 235], [249, 204], [315, 170], [344, 175], [303, 121]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "72827_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'a folded cloth', and now you are performing subtask 'a folded cloth.'\nYour initial coordinate is [0.33, 0.56]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.33, 0.56], [0.32, 0.7], [0.36, 0.73], [0.31, 0.59], [0.31, 0.49], [0.4, 0.47], [0.42, 0.48], [0.41, 0.5], [0.43, 0.53], [0.46, 0.51]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/72827_exterior_image_2_left#0.jpg", + "gt": "[[106, 100], [104, 126], [114, 132], [98, 106], [99, 88], [128, 85], [134, 86], [132, 90], [136, 95], [147, 91]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "5195_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the carrot plush toy and put it on the right side of the table', and now you are performing subtask 'pick up the carrot plush toy and put it on the right side of the table.'\nYour starting position is [0.68, 0.01]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.01], [0.61, 0.25], [0.59, 0.44], [0.6, 0.5], [0.63, 0.5], [0.61, 0.27], [0.58, 0.33], [0.6, 0.41], [0.6, 0.27], [0.58, 0.29]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/5195_exterior_image_2_left#0.jpg", + "gt": "[[219, 1], [194, 45], [189, 80], [193, 90], [202, 90], [195, 48], [187, 59], [191, 73], [191, 48], [187, 52]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "9303_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the cloth', and now you are performing subtask 'slide the cloth to the back side of the table.'\nStarting at [0.75, 0.79], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.75, 0.79], [0.75, 0.81], [0.67, 0.72], [0.61, 0.72], [0.58, 0.74], [0.54, 0.71], [0.48, 0.61], [0.49, 0.6], [0.53, 0.65], [0.51, 0.65]]}" + } + ], + "images": "9303_exterior_image_1_left#236", + "gt": "[[241, 143], [240, 145], [214, 130], [195, 130], [185, 134], [173, 128], [155, 110], [157, 108], [170, 117], [164, 117]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "4908_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull out the top shelf of the nightstand', and now you are performing subtask 'slide the top shelf of the nightstand out.'\nWith start location [0.62, 0.28], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.28], [0.62, 0.26], [0.57, 0.21], [0.59, 0.19], [0.63, 0.2], [0.64, 0.19], [0.65, 0.19], [0.65, 0.16], [0.63, 0.14], [0.63, 0.14]]}" + } + ], + "images": "4908_exterior_image_1_left#193", + "gt": "[[197.0, 49.5], [199.0, 47.5], [184.0, 38.5], [189.0, 34.5], [202.0, 36.5], [204.0, 33.5], [209.0, 33.5], [209.0, 28.5], [203.0, 25.5], [203.0, 25.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_035622060973#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'putting a portion of the Lego blocks on the Lego baseplate', and now you are performing subtask 'pick up the red Lego brick from the table, transfer the red Lego brick from the table to the Lego base, insert the red Lego brick into the Lego base.'\nYou start from [0.39, 0.36]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.36], [0.31, 0.25], [0.27, 0.27], [0.28, 0.3], [0.28, 0.27], [0.3, 0.22], [0.35, 0.27], [0.35, 0.32], [0.34, 0.29], [0.34, 0.29]]}" + } + ], + "images": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_035622060973#0", + "gt": "[[247, 131], [200, 90], [175, 96], [179, 108], [177, 97], [190, 81], [227, 98], [227, 114], [219, 104], [219, 104]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0096_user_0014_scene_0006_cfg_0004_104122062823#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take something out of a drawer', and now you are performing subtask 'pick up the object from the drawer, transfer the object from the drawer to the countertop, place the object on the countertop.'\nBegin at [0.57, 0.0] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.0], [0.64, 0.0], [0.64, 0.04], [0.65, 0.06], [0.63, 0.1], [0.64, 0.11], [0.64, 0.07], [0.61, 0.0], [0.44, 0.0], [0.57, 0.0]]}" + } + ], + "images": "RH20T_cfg4_task_0096_user_0014_scene_0006_cfg_0004_104122062823#186", + "gt": "[[368, 0], [411, 0], [408, 13], [413, 21], [406, 35], [408, 38], [407, 24], [390, 0], [282, 0], [367, 0]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0110_user_0015_scene_0003_cfg_0001_039422060546#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the pencil out from the pencil sharpener', and now you are performing subtask 'take the pencil out from the pencil sharpener.'\nYou start from [0.42, 0.34]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.34], [0.5, 0.31], [0.48, 0.36], [0.56, 0.37], [0.58, 0.34], [0.58, 0.29], [0.55, 0.27], [0.57, 0.26], [0.55, 0.28], [0.52, 0.38]]}" + } + ], + "images": "RH20T_cfg1_task_0110_user_0015_scene_0003_cfg_0001_039422060546#0", + "gt": "[[266, 122], [319, 110], [309, 128], [356, 132], [372, 122], [369, 103], [354, 96], [365, 94], [349, 100], [331, 136]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "61367_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push the faucet to the right', and now you are performing subtask 'push the faucet to the right.'\nYou begin at [0.77, 0.53]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.77, 0.53], [0.62, 0.57], [0.51, 0.56], [0.45, 0.53], [0.47, 0.56], [0.49, 0.54], [0.51, 0.55], [0.57, 0.49], [0.6, 0.41], [0.59, 0.38]]}" + } + ], + "images": "61367_exterior_image_1_left#0", + "gt": "[[246, 96], [198, 102], [162, 101], [143, 96], [151, 101], [156, 97], [164, 99], [182, 88], [192, 73], [189, 68]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0096_user_0014_scene_0001_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take something out of a drawer', and now you are performing subtask 'pull the drawer handle to the outside of the table.'\nThe trajectory starts at [0.63, 0.37]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.64, 0.21], [0.62, 0.11], [0.61, 0.09], [0.6, 0.09], [0.61, 0.17], [0.62, 0.19], [0.62, 0.23], [0.62, 0.21], [0.61, 0.09]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0096_user_0014_scene_0001_cfg_0001_038522062547#0.jpg", + "gt": "[[406, 132], [411, 76], [398, 39], [389, 32], [382, 33], [390, 61], [396, 69], [397, 82], [399, 76], [392, 31]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0016_user_0020_scene_0004_cfg_0007_104122061330#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'hit the pool ball', and now you are performing subtask 'transfer the cue from the top of the table to the position above the pool ball, strike the pool ball with the cue, place the cue on the pool table.'\nWith start location [0.59, 0.37], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.37], [0.6, 0.38], [0.57, 0.41], [0.59, 0.36], [0.57, 0.38], [0.57, 0.34], [0.56, 0.26], [0.54, 0.24], [0.53, 0.26], [0.52, 0.26]]}" + } + ], + "images": "RH20T_cfg7_task_0016_user_0020_scene_0004_cfg_0007_104122061330#275", + "gt": "[[379, 133], [381, 137], [367, 149], [378, 128], [368, 138], [367, 122], [357, 95], [348, 88], [338, 93], [332, 92]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0037_user_0014_scene_0009_cfg_0007_104122061330#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the block on the first block, pick up the fourth block from the desktop, transfer the block from the desktop to the third block, place the block on the third block.'\nYour initial coordinate is [0.57, 0.28]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.28], [0.54, 0.2], [0.54, 0.22], [0.54, 0.24], [0.53, 0.26], [0.54, 0.19], [0.54, 0.21], [0.57, 0.19], [0.57, 0.2], [0.57, 0.25]]}" + } + ], + "images": "RH20T_cfg7_task_0037_user_0014_scene_0009_cfg_0007_104122061330#266", + "gt": "[[365, 101], [347, 73], [347, 81], [343, 85], [341, 92], [343, 68], [347, 76], [362, 70], [365, 71], [368, 89]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0096_user_0015_scene_0005_cfg_0001_039422060546#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the block out of the drawer and put it on the table', and now you are performing subtask 'pick up the block from the drawer, transfer the block from the drawer to the tabletop, place the block on the tabletop.'\nYour starting position is [0.6, 0.17]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.17], [0.57, 0.28], [0.59, 0.17], [0.57, 0.09], [0.54, 0.26], [0.54, 0.43], [0.52, 0.6], [0.53, 0.73], [0.51, 0.62], [0.52, 0.65]]}" + } + ], + "images": "RH20T_cfg1_task_0096_user_0015_scene_0005_cfg_0001_039422060546#217", + "gt": "[[382, 62], [362, 102], [380, 61], [365, 34], [348, 94], [348, 156], [333, 217], [337, 263], [329, 223], [335, 235]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0058_user_0014_scene_0001_cfg_0006_104122063633#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the switches on the power strip in sequence', and now you are performing subtask 'press the fourth button on the power strip.'\nYour initial coordinate is [0.4, 0.55]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.55], [0.4, 0.56], [0.39, 0.56], [0.39, 0.54], [0.39, 0.54], [0.39, 0.5], [0.39, 0.49], [0.4, 0.49], [0.4, 0.49], [0.4, 0.51]]}" + } + ], + "images": "RH20T_cfg6_task_0058_user_0014_scene_0001_cfg_0006_104122063633#229", + "gt": "[[253, 199], [253, 200], [250, 202], [250, 196], [249, 193], [250, 180], [252, 177], [257, 176], [258, 178], [255, 185]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "15661_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the cloth to the left', and now you are performing subtask 'move the cloth to the left.'\nThe trajectory starts at [0.52, 0.02]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.02], [0.61, 0.25], [0.62, 0.31], [0.64, 0.5], [0.63, 0.57], [0.6, 0.59], [0.53, 0.48], [0.51, 0.34], [0.53, 0.39], [0.52, 0.35]]}" + } + ], + "images": "15661_exterior_image_1_left#0", + "gt": "[[167, 4], [194, 45], [198, 56], [204, 90], [201, 102], [193, 106], [170, 87], [164, 62], [169, 71], [167, 63]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "84561_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the brown block in the plate', and now you are performing subtask 'put the brown block in the plate.'\nThe trajectory starts at [0.23, 0.01]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.23, 0.01], [0.3, 0.11], [0.27, 0.09], [0.28, 0.12], [0.47, 0.5], [0.41, 0.45], [0.39, 0.49], [0.39, 0.55], [0.35, 0.45], [0.36, 0.47]]}" + } + ], + "images": "84561_exterior_image_1_left#0", + "gt": "[[74.5, 2.0], [95.5, 19.0], [87.5, 17.0], [90.5, 21.0], [151.5, 90.0], [131.5, 81.0], [124.5, 89.0], [124.5, 99.0], [111.5, 81.0], [116.5, 84.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0077_user_0014_scene_0006_cfg_0002_037522062165#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stacking blocks (small Lego) randomly one at a time', and now you are performing subtask 'place the blue block at the grid center, pick up the red block on the table, transfer the red block from the table to the top of the blue block, place the red block on the blue block.'\nYou begin at [0.3, 0.84]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.3, 0.84], [0.23, 0.91], [0.18, 0.84], [0.21, 0.88], [0.28, 0.82], [0.3, 0.84], [0.32, 0.83], [0.32, 0.86], [0.31, 0.83], [0.31, 0.83]]}" + } + ], + "images": "RH20T_cfg2_task_0077_user_0014_scene_0006_cfg_0002_037522062165#276", + "gt": "[[195, 304], [145, 326], [115, 304], [135, 317], [182, 295], [195, 303], [205, 298], [203, 309], [196, 299], [196, 299]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "70284_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the open drawer', and now you are performing subtask 'close the open drawer.'\nYou begin at [0.66, 0.53]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.66, 0.53], [0.73, 0.49], [0.73, 0.46], [0.67, 0.18], [0.62, 0.14], [0.5, 0.19], [0.45, 0.25], [0.51, 0.26], [0.57, 0.35], [0.56, 0.36]]}" + } + ], + "images": "70284_exterior_image_1_left#0", + "gt": "[[212, 96], [232, 89], [235, 82], [214, 32], [198, 25], [161, 34], [144, 45], [162, 46], [183, 63], [178, 64]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "32646_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the cloth on the table', and now you are performing subtask 'unfold the cloth on the table.'\nYour starting position is [0.38, 0.38]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.38], [0.45, 0.35], [0.54, 0.38], [0.71, 0.56], [0.69, 0.56], [0.68, 0.75], [0.65, 0.63], [0.56, 0.49], [0.6, 0.51], [0.63, 0.58]]}" + } + ], + "images": "32646_exterior_image_2_left#0", + "gt": "[[122, 68], [145, 63], [172, 68], [226, 100], [220, 101], [217, 135], [209, 114], [178, 88], [193, 92], [201, 105]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0044_user_0010_scene_0002_cfg_0005_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up a bag of things', and now you are performing subtask 'pick up a bag of things.'\nYour starting position is [0.45, 0.29]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.29], [0.44, 0.3], [0.43, 0.27], [0.44, 0.27], [0.44, 0.28], [0.44, 0.28], [0.44, 0.29], [0.43, 0.3], [0.44, 0.34], [0.45, 0.2]]}" + } + ], + "images": "RH20T_cfg5_task_0044_user_0010_scene_0002_cfg_0005_036422060215#0", + "gt": "[[286, 106], [284, 109], [276, 97], [280, 96], [280, 100], [284, 101], [283, 106], [278, 108], [279, 122], [286, 71]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "42244_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the two sachets from the box and place them on the upper countertop', and now you are performing subtask 'pick up the object on the plate.'\nStarting at [0.58, 0.39], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.39], [0.49, 0.53], [0.51, 0.51], [0.49, 0.55], [0.53, 0.81], [0.51, 0.83], [0.52, 0.69], [0.47, 0.53], [0.51, 0.63], [0.5, 0.73]]}" + } + ], + "images": "42244_exterior_image_1_left#0", + "gt": "[[186, 70], [157, 95], [162, 91], [158, 99], [171, 146], [164, 149], [165, 125], [150, 96], [163, 113], [159, 131]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0016_user_0003_scene_0009_cfg_0002_037522062165#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'an action of hitting the pool ball', and now you are performing subtask 'strike the pool ball on the green pool table, transfer the cue from in front of the pool ball to the pool table, place the cue on the pool table.'\nThe trajectory starts at [0.46, 0.64]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.64], [0.42, 0.68], [0.44, 0.66], [0.47, 0.59], [0.46, 0.61], [0.46, 0.59], [0.43, 0.59], [0.38, 0.55], [0.4, 0.58], [0.43, 0.43]]}" + } + ], + "images": "RH20T_cfg2_task_0016_user_0003_scene_0009_cfg_0002_037522062165#252", + "gt": "[[295, 230], [271, 245], [282, 239], [300, 213], [297, 218], [295, 212], [272, 213], [246, 199], [256, 208], [272, 156]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0094_user_0007_scene_0006_cfg_0005_104422070044#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'press the part of the box, press the part of the box.'\nYour starting position is [0.65, 0.42]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.42], [0.66, 0.41], [0.68, 0.37], [0.69, 0.33], [0.69, 0.35], [0.69, 0.32], [0.65, 0.38], [0.63, 0.42], [0.62, 0.45], [0.62, 0.44]]}" + } + ], + "images": "RH20T_cfg5_task_0094_user_0007_scene_0006_cfg_0005_104422070044#168", + "gt": "[[415, 151], [420, 149], [435, 134], [443, 119], [444, 126], [442, 117], [416, 137], [405, 150], [399, 161], [398, 160]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "6951_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unhang the cloth and put it on the table', and now you are performing subtask 'unhang the cloth and put it on the table.'\nStart at [0.44, 0.01]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.01], [0.44, 0.08], [0.43, 0.07], [0.43, 0.12], [0.47, 0.07], [0.46, 0.18], [0.43, 0.22], [0.41, 0.23], [0.38, 0.23], [0.36, 0.19]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/6951_exterior_image_2_left#0.jpg", + "gt": "[[140, 1], [140, 14], [138, 13], [137, 22], [151, 12], [148, 32], [139, 40], [131, 42], [121, 42], [115, 34]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0073_user_0014_scene_0003_cfg_0002_105422061350#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the block in the drawer, close the drawer', and now you are performing subtask 'push the drawer to close it.'\nYou begin at [0.5, 0.24]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.24], [0.52, 0.27], [0.54, 0.23], [0.55, 0.23], [0.6, 0.17], [0.55, 0.19], [0.52, 0.24], [0.52, 0.25], [0.53, 0.27], [0.53, 0.27]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg2_task_0073_user_0014_scene_0003_cfg_0002_105422061350#213.jpg", + "gt": "[[317, 87], [330, 97], [347, 83], [355, 84], [386, 60], [351, 69], [336, 85], [334, 91], [339, 98], [339, 98]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "61768_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the marker out of the pot and put it on the table', and now you are performing subtask 'take the marker out of the pot and put it on the table.'\nThe trajectory starts at [0.52, 0.0]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.0], [0.53, 0.38], [0.55, 0.37], [0.55, 0.31], [0.58, 0.28], [0.61, 0.3], [0.66, 0.41], [0.66, 0.45], [0.65, 0.47], [0.65, 0.4]]}" + } + ], + "images": "61768_exterior_image_1_left#0", + "gt": "[[165, 0], [170, 69], [176, 66], [177, 55], [187, 50], [195, 54], [210, 73], [211, 81], [207, 85], [208, 72]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "80108_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick the carrot and put it on the pan', and now you are performing subtask 'pick the carrot and put it on the pan.'\nYou start from [0.33, 0.01]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.01], [0.69, 0.54], [0.64, 0.65], [0.66, 0.75], [0.57, 0.34], [0.57, 0.51], [0.61, 0.49], [0.52, 0.48], [0.54, 0.61], [0.53, 0.53]]}" + } + ], + "images": "80108_exterior_image_1_left#0", + "gt": "[[104.0, 1.0], [222.0, 97.0], [204.0, 117.0], [212.0, 135.0], [184.0, 61.0], [183.0, 92.0], [196.0, 88.0], [167.0, 86.0], [173.0, 110.0], [170.0, 95.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "4810_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove one white plate from the dishwasher and place it on top of the white plate on the countertop', and now you are performing subtask 'remove one white plate from the dishwasher and place it on top of the white plate on the countertop.'\nYou begin at [0.46, 0.36]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.36], [0.55, 0.42], [0.54, 0.57], [0.56, 0.42], [0.61, 0.35], [0.65, 0.36], [0.73, 0.47], [0.69, 0.48], [0.57, 0.39], [0.49, 0.41]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/4810_exterior_image_2_left#0.jpg", + "gt": "[[146, 64], [177, 75], [172, 103], [178, 75], [195, 63], [208, 65], [235, 84], [222, 86], [182, 71], [158, 74]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "25352_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the white towel from right to left', and now you are performing subtask 'fold the white towel from right to left.'\nThe trajectory starts at [0.43, 0.45]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.45], [0.47, 0.59], [0.47, 0.69], [0.46, 0.69], [0.49, 0.73], [0.49, 0.81], [0.47, 0.8], [0.46, 0.83], [0.46, 0.71], [0.37, 0.58]]}" + } + ], + "images": "25352_exterior_image_1_left#0", + "gt": "[[137, 81], [151, 106], [152, 124], [146, 124], [157, 132], [156, 145], [149, 144], [147, 149], [147, 127], [119, 105]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "3657_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push the box into the cabinet', and now you are performing subtask 'push the box into the cabinet.'\nYour starting position is [0.45, 0.58]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.58], [0.49, 0.63], [0.53, 0.63], [0.53, 0.67], [0.55, 0.68], [0.58, 0.68], [0.62, 0.64], [0.54, 0.67], [0.5, 0.63], [0.5, 0.64]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/3657_exterior_image_2_left#0.jpg", + "gt": "[[144, 104], [156, 113], [169, 114], [170, 120], [176, 123], [186, 123], [199, 116], [172, 120], [161, 113], [160, 115]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0003_user_0006_scene_0007_cfg_0002_104122063678#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the three buttons in sequence', and now you are performing subtask 'press the gray base green button, press the gray base red button.'\nYou begin at [0.49, 0.81]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.81], [0.47, 0.81], [0.46, 0.76], [0.44, 0.69], [0.43, 0.66], [0.45, 0.68], [0.43, 0.62], [0.44, 0.64], [0.44, 0.61], [0.45, 0.6]]}" + } + ], + "images": "RH20T_cfg2_task_0003_user_0006_scene_0007_cfg_0002_104122063678#194", + "gt": "[[315, 290], [300, 291], [297, 275], [279, 250], [277, 236], [287, 245], [276, 224], [279, 229], [280, 219], [286, 216]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "37344_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the button to open the rice cooker', and now you are performing subtask 'press the button to open the rice cooker.'\nYour starting position is [0.64, 0.66]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.64, 0.66], [0.54, 0.73], [0.48, 0.67], [0.4, 0.64], [0.4, 0.59], [0.38, 0.59], [0.38, 0.68], [0.41, 0.45], [0.48, 0.42], [0.6, 0.43]]}" + } + ], + "images": "37344_exterior_image_2_left#0", + "gt": "[[204, 118], [172, 131], [154, 120], [128, 115], [127, 106], [120, 106], [121, 122], [130, 81], [155, 75], [191, 78]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0037_user_0007_scene_0006_cfg_0002_104122063678#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the second block on the third block, pick up the third block on the table, transfer the third block from the table to the top of the fourth block, place the third block on the fourth block.'\nYour starting position is [0.55, 0.58]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.58], [0.53, 0.52], [0.53, 0.55], [0.52, 0.53], [0.53, 0.55], [0.53, 0.58], [0.53, 0.57], [0.52, 0.47], [0.55, 0.52], [0.54, 0.54]]}" + } + ], + "images": "RH20T_cfg2_task_0037_user_0007_scene_0006_cfg_0002_104122063678#481", + "gt": "[[349, 209], [337, 187], [338, 197], [334, 191], [340, 199], [340, 209], [338, 205], [333, 170], [351, 186], [347, 196]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "66931_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the plush toy behind the shelf', and now you are performing subtask 'put the plush toy behind the shelf.'\nYou start from [0.99, 0.59]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.99, 0.59], [0.82, 0.56], [0.75, 0.64], [0.72, 0.58], [0.64, 0.52], [0.59, 0.53], [0.56, 0.57], [0.54, 0.66], [0.57, 0.58], [0.55, 0.58]]}" + } + ], + "images": "66931_exterior_image_2_left#0", + "gt": "[[318, 107], [264, 100], [241, 115], [231, 104], [205, 94], [189, 96], [179, 103], [173, 119], [182, 105], [176, 105]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "3681_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer and place an object inside', and now you are performing subtask 'pull the drawer's handle to fully open the drawer.'\nBegin at [0.63, 0.53] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.53], [0.63, 0.55], [0.63, 0.56], [0.64, 0.57], [0.66, 0.58], [0.69, 0.58], [0.7, 0.57], [0.69, 0.53], [0.69, 0.51], [0.65, 0.47]]}" + } + ], + "images": "3681_exterior_image_1_left#158", + "gt": "[[202, 96], [203, 99], [202, 100], [204, 102], [210, 105], [220, 105], [223, 103], [222, 96], [220, 92], [209, 84]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122064161#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'place the block on the other blocks, pick up the block on the flat surface, transfer the block from the flat surface to the other blocks, place the block on the other blocks.'\nYour starting position is [0.55, 0.44]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.44], [0.62, 0.46], [0.64, 0.44], [0.66, 0.48], [0.65, 0.43], [0.6, 0.38], [0.56, 0.4], [0.54, 0.45], [0.56, 0.44], [0.57, 0.36]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122064161#523.jpg", + "gt": "[[354, 160], [397, 167], [409, 157], [420, 172], [414, 153], [387, 136], [359, 145], [345, 163], [360, 158], [363, 130]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0047_user_0014_scene_0005_cfg_0004_104122062823#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the cup off the cup rack', and now you are performing subtask 'push the cup to the edge of the table.'\nYour initial coordinate is [0.38, 0.0]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.31, 0.0], [0.32, 0.01], [0.33, 0.05], [0.33, 0.03], [0.39, 0.01], [0.43, 0.04], [0.43, 0.01], [0.42, 0.0], [0.37, 0.0]]}" + } + ], + "images": "RH20T_cfg4_task_0047_user_0014_scene_0005_cfg_0004_104122062823#247", + "gt": "[[240, 0], [198, 0], [206, 2], [211, 18], [212, 12], [252, 3], [276, 16], [276, 3], [269, 0], [234, 0]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "38503_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the cloth off the glass then put it back', and now you are performing subtask 'take the cloth off the glass then put it back.'\nYour initial coordinate is [0.37, 0.26]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.26], [0.43, 0.31], [0.36, 0.46], [0.37, 0.72], [0.37, 0.39], [0.41, 0.31], [0.41, 0.21], [0.43, 0.17], [0.48, 0.23], [0.5, 0.19]]}" + } + ], + "images": "38503_exterior_image_2_left#0", + "gt": "[[119, 46], [138, 56], [116, 83], [118, 129], [118, 70], [131, 55], [130, 38], [137, 30], [155, 42], [161, 35]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0043_user_0020_scene_0009_cfg_0007_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unscrew the lid from the jar', and now you are performing subtask 'twist the lid counterclockwise.'\nWith start location [0.53, 0.32], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.32], [0.53, 0.31], [0.53, 0.33], [0.55, 0.37], [0.56, 0.36], [0.58, 0.39], [0.59, 0.38], [0.59, 0.43], [0.59, 0.44], [0.59, 0.44]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0043_user_0020_scene_0009_cfg_0007_104122064161#0.jpg", + "gt": "[[337, 115], [338, 112], [338, 118], [350, 132], [361, 129], [371, 139], [380, 137], [377, 153], [380, 160], [375, 159]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "66631_exterior_image_2_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the towel to wipe the table', and now you are performing subtask 'fold the towel.'\nThe trajectory starts at [0.61, 0.54]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.61, 0.54], [0.63, 0.54], [0.63, 0.56], [0.62, 0.52], [0.59, 0.52], [0.57, 0.57], [0.56, 0.61], [0.58, 0.59], [0.59, 0.58], [0.59, 0.52]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/66631_exterior_image_2_left#236.jpg", + "gt": "[[196, 98], [201, 98], [203, 101], [197, 94], [189, 94], [182, 102], [180, 109], [186, 107], [188, 104], [189, 94]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "84706_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the pee plush toy to the counter, then pick up the cloth and place it on the counter, then pick up the toy from the sink and put it in the pot', and now you are performing subtask 'place the pink toy in the pot.'\nYour initial coordinate is [0.36, 0.5]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.5], [0.48, 0.53], [0.6, 0.73], [0.58, 0.73], [0.6, 0.75], [0.59, 0.77], [0.58, 0.71], [0.64, 0.76], [0.73, 0.68], [0.71, 0.66]]}" + } + ], + "images": "84706_exterior_image_1_left#225", + "gt": "[[116.5, 90.0], [154.5, 96.0], [192.5, 132.0], [184.5, 131.0], [192.5, 135.0], [187.5, 138.0], [186.5, 128.0], [203.5, 137.0], [232.5, 122.0], [227.5, 119.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0008_user_0007_scene_0004_cfg_0002_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'swinging the flag', and now you are performing subtask 'swinging the flag.'\nYour starting position is [0.62, 0.53]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.53], [0.61, 0.69], [0.63, 0.57], [0.62, 0.43], [0.59, 0.3], [0.61, 0.33], [0.65, 0.51], [0.6, 0.35], [0.63, 0.48], [0.63, 0.6]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg2_task_0008_user_0007_scene_0004_cfg_0002_104122063678#0.jpg", + "gt": "[[400, 190], [393, 250], [404, 206], [394, 154], [376, 107], [390, 120], [415, 185], [387, 125], [405, 172], [401, 215]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "49706_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the plush toy from the box into the drawer after opening the drawer, then close the drawer', and now you are performing subtask 'pull the drawer handle to the open position.'\nStarting at [0.76, 0.46], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.76, 0.46], [0.7, 0.41], [0.62, 0.43], [0.6, 0.47], [0.61, 0.51], [0.58, 0.56], [0.64, 0.52], [0.58, 0.57], [0.64, 0.53], [0.63, 0.51]]}" + } + ], + "images": "49706_exterior_image_1_left#0", + "gt": "[[243, 82], [223, 74], [200, 78], [191, 84], [195, 91], [187, 101], [206, 94], [187, 102], [206, 96], [203, 92]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0077_user_0015_scene_0009_cfg_0004_036422060909#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the first Lego block at the top left corner of the table to the top left corner of the gray base plate', and now you are performing subtask 'move the first Lego block at the top left corner of the table to the top left corner of the gray base plate.'\nFrom start point [0.52, 0.22], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.22], [0.51, 0.21], [0.45, 0.26], [0.39, 0.36], [0.3, 0.43], [0.28, 0.47], [0.29, 0.49], [0.42, 0.34], [0.42, 0.38], [0.43, 0.33]]}" + } + ], + "images": "RH20T_cfg4_task_0077_user_0015_scene_0009_cfg_0004_036422060909#0", + "gt": "[[334, 80], [324, 75], [286, 92], [252, 129], [194, 154], [180, 170], [186, 177], [266, 123], [270, 135], [275, 118]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "24671_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'rotate the tap nozzle clockwise', and now you are performing subtask 'rotate the tap nozzle clockwise.'\nThe trajectory starts at [0.7, 0.01]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.01], [0.74, 0.34], [0.69, 0.38], [0.72, 0.34], [0.68, 0.32], [0.67, 0.25], [0.68, 0.24], [0.62, 0.24], [0.58, 0.26], [0.6, 0.27]]}" + } + ], + "images": "24671_exterior_image_1_left#0", + "gt": "[[223, 2], [237, 61], [221, 68], [230, 61], [216, 58], [213, 45], [217, 43], [197, 44], [187, 47], [193, 49]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg3_task_0003_user_0016_scene_0006_cfg_0003_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the green button on the right.'\nYou start from [0.38, 0.0]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.46, 0.02], [0.57, 0.15], [0.51, 0.16], [0.57, 0.11], [0.61, 0.18], [0.59, 0.19], [0.6, 0.21], [0.58, 0.25], [0.58, 0.21]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg3_task_0003_user_0016_scene_0006_cfg_0003_104122062823#0.jpg", + "gt": "[[242, 0], [294, 7], [366, 55], [325, 58], [364, 41], [389, 66], [376, 70], [382, 76], [370, 90], [372, 77]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0065_user_0014_scene_0004_cfg_0001_750612070851#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'transfer the blocks to the block building area', and now you are performing subtask 'transfer the blocks to the block building area.'\nStarting at [0.49, 0.47], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.47], [0.38, 0.61], [0.35, 0.68], [0.32, 0.71], [0.3, 0.76], [0.3, 0.8], [0.32, 0.79], [0.3, 0.73], [0.31, 0.68], [0.31, 0.68]]}" + } + ], + "images": "RH20T_cfg1_task_0065_user_0014_scene_0004_cfg_0001_750612070851#0", + "gt": "[[312, 169], [242, 220], [224, 246], [205, 255], [192, 274], [189, 288], [204, 286], [190, 263], [199, 244], [199, 244]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "7364_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the cup in the sink', and now you are performing subtask 'put the cup in the sink.'\nYou start from [0.99, 0.52]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.99, 0.52], [0.94, 0.46], [0.84, 0.47], [0.58, 0.76], [0.65, 0.63], [0.62, 0.63], [0.6, 0.68], [0.62, 0.63], [0.68, 0.68], [0.74, 0.63]]}" + } + ], + "images": "7364_exterior_image_2_left#0", + "gt": "[[318, 93], [301, 83], [269, 85], [185, 136], [208, 113], [198, 114], [193, 123], [200, 114], [217, 123], [237, 113]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0105_user_0007_scene_0014_cfg_0004_104122062823#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the object from the table and move it upward', and now you are performing subtask 'pick up the object from the table.'\nYou begin at [0.44, 0.14]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.14], [0.45, 0.19], [0.43, 0.23], [0.44, 0.32], [0.45, 0.34], [0.45, 0.23], [0.39, 0.04], [0.39, 0.0], [0.09, 0.0], [0.4, 0.0]]}" + } + ], + "images": "RH20T_cfg4_task_0105_user_0007_scene_0014_cfg_0004_104122062823#128", + "gt": "[[279, 50], [287, 70], [278, 82], [279, 115], [289, 122], [289, 83], [248, 14], [248, 0], [57, 0], [256, 0]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0004_user_0010_scene_0001_cfg_0005_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the small block on the left and move it to the right', and now you are performing subtask 'pick up the small block on the left and move it to the right.'\nYour starting position is [0.45, 0.42]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.42], [0.42, 0.44], [0.25, 0.43], [0.21, 0.38], [0.21, 0.47], [0.31, 0.59], [0.45, 0.61], [0.48, 0.68], [0.48, 0.72], [0.48, 0.72]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0004_user_0010_scene_0001_cfg_0005_037522062165#0.jpg", + "gt": "[[289, 152], [269, 160], [159, 155], [137, 137], [136, 171], [201, 214], [290, 219], [310, 243], [309, 260], [309, 260]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0130_user_0020_scene_0004_cfg_0006_104122061018#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb', and now you are performing subtask 'press the switch on the socket to turn on the bulb, press the switch on the bulb.'\nYou start from [0.46, 0.24]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.24], [0.44, 0.34], [0.44, 0.35], [0.45, 0.26], [0.47, 0.31], [0.45, 0.23], [0.55, 0.11], [0.54, 0.11], [0.54, 0.09], [0.53, 0.13]]}" + } + ], + "images": "RH20T_cfg6_task_0130_user_0020_scene_0004_cfg_0006_104122061018#393", + "gt": "[[296, 88], [279, 124], [284, 126], [285, 94], [299, 113], [285, 82], [355, 41], [343, 39], [345, 34], [342, 47]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "18274_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the pen to stir the contents of the mug, pour the contents of the mug on the table, then sweep them', and now you are performing subtask 'place the pen on the table.'\nStart at [0.11, 0.65]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.11, 0.65], [0.2, 0.71], [0.17, 0.85], [0.26, 0.78], [0.31, 0.94], [0.26, 0.77], [0.3, 0.59], [0.36, 0.71], [0.3, 0.66], [0.29, 0.82]]}" + } + ], + "images": "18274_exterior_image_1_left#0", + "gt": "[[36, 117], [65, 128], [55, 153], [82, 140], [99, 169], [83, 138], [96, 107], [115, 128], [96, 119], [93, 147]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "43469_exterior_image_1_left#TrajectoryPrediction_Task#12", + "task": "TrajectoryPrediction_Task", + "p_index": 12, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the sweatshirt twice', and now you are performing subtask 'fold the sweatshirt in half again.'\nYou start from [0.56, 0.34]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.56, 0.34], [0.55, 0.38], [0.55, 0.41], [0.56, 0.43], [0.57, 0.42], [0.59, 0.39], [0.66, 0.36], [0.65, 0.24], [0.66, 0.18], [0.68, 0.18]]}" + } + ], + "images": "43469_exterior_image_1_left#487", + "gt": "[[179, 61], [175, 69], [176, 74], [180, 78], [184, 76], [189, 70], [210, 64], [208, 43], [212, 33], [216, 33]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0101_user_0007_scene_0002_cfg_0002_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'cleaning the table with a cloth', and now you are performing subtask 'cleaning the table with a cloth.'\nStart at [0.46, 0.45]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.45], [0.47, 0.45], [0.47, 0.44], [0.47, 0.42], [0.47, 0.43], [0.46, 0.42], [0.41, 0.46], [0.37, 0.51], [0.41, 0.48], [0.45, 0.5]]}" + } + ], + "images": "RH20T_cfg2_task_0101_user_0007_scene_0002_cfg_0002_036422060215#0", + "gt": "[[294, 162], [303, 163], [299, 158], [301, 151], [303, 155], [292, 152], [260, 165], [234, 183], [260, 173], [288, 180]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0022_user_0014_scene_0006_cfg_0006_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'wave the flag', and now you are performing subtask 'wave the flag.'\nStart at [0.53, 0.31]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.31], [0.6, 0.43], [0.61, 0.51], [0.56, 0.55], [0.58, 0.54], [0.56, 0.42], [0.54, 0.39], [0.55, 0.51], [0.53, 0.38], [0.51, 0.38]]}" + } + ], + "images": "RH20T_cfg6_task_0022_user_0014_scene_0006_cfg_0006_104122064161#0", + "gt": "[[338, 110], [387, 156], [393, 185], [358, 198], [370, 195], [361, 151], [344, 140], [351, 182], [339, 137], [326, 137]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0094_user_0001_scene_0009_cfg_0001_038522062547#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'pull the lid of the box upwards to open the box.'\nBegin at [0.52, 0.86] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.86], [0.52, 0.84], [0.56, 0.81], [0.57, 0.76], [0.61, 0.72], [0.64, 0.59], [0.65, 0.61], [0.63, 0.66], [0.63, 0.63], [0.62, 0.6]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0094_user_0001_scene_0009_cfg_0001_038522062547#212.jpg", + "gt": "[[335, 311], [334, 303], [357, 291], [363, 275], [388, 259], [412, 213], [418, 218], [404, 237], [406, 226], [394, 215]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0061_user_0007_scene_0010_cfg_0005_105422061350#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'pick up the object in the container, transfer the object from the container to the table, place the object on the table.'\nYour initial coordinate is [0.43, 0.13]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.13], [0.41, 0.14], [0.38, 0.24], [0.39, 0.27], [0.4, 0.31], [0.4, 0.28], [0.39, 0.29], [0.37, 0.38], [0.37, 0.49], [0.38, 0.56]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0061_user_0007_scene_0010_cfg_0005_105422061350#0.jpg", + "gt": "[[276, 47], [263, 49], [246, 86], [247, 96], [259, 112], [257, 100], [250, 105], [239, 136], [236, 178], [242, 200]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "69180_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the first drawer from the right, remove the glue stick from it and put the glue stick on the counter', and now you are performing subtask 'place the glue stick on the counter.'\nThe trajectory starts at [0.52, 0.51]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.51], [0.52, 0.52], [0.48, 0.66], [0.53, 0.56], [0.53, 0.49], [0.52, 0.52], [0.52, 0.43], [0.57, 0.24], [0.54, 0.39], [0.59, 0.22]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/69180_exterior_image_2_left#185.jpg", + "gt": "[[165, 92], [168, 94], [154, 118], [169, 100], [171, 89], [165, 94], [167, 78], [182, 43], [174, 70], [189, 40]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0015_user_0020_scene_0007_cfg_0007_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'play the drum', and now you are performing subtask 'play the drum.'\nYour initial coordinate is [0.53, 0.32]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.32], [0.53, 0.37], [0.54, 0.33], [0.57, 0.36], [0.54, 0.31], [0.56, 0.25], [0.59, 0.34], [0.59, 0.25], [0.57, 0.33], [0.55, 0.18]]}" + } + ], + "images": "RH20T_cfg7_task_0015_user_0020_scene_0007_cfg_0007_104122064161#0", + "gt": "[[337, 115], [339, 133], [344, 118], [366, 128], [348, 110], [361, 91], [375, 123], [376, 89], [366, 118], [352, 65]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "53689_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the napkin completely', and now you are performing subtask 'unfold the napkin completely.'\nYou begin at [0.71, 0.56]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.71, 0.56], [0.68, 0.61], [0.71, 0.67], [0.69, 0.67], [0.66, 0.63], [0.66, 0.63], [0.66, 0.49], [0.62, 0.51], [0.61, 0.59], [0.63, 0.63]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/53689_exterior_image_1_left#223.jpg", + "gt": "[[227, 101], [218, 109], [226, 120], [222, 121], [212, 113], [210, 114], [212, 88], [200, 91], [196, 106], [203, 114]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0057_user_0007_scene_0002_cfg_0005_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press a row of four power strip buttons to turn on the power strips', and now you are performing subtask 'press the first button on the power socket.'\nYour starting position is [0.45, 0.42]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.42], [0.48, 0.43], [0.49, 0.46], [0.47, 0.64], [0.5, 0.6], [0.5, 0.57], [0.47, 0.63], [0.47, 0.65], [0.48, 0.64], [0.47, 0.64]]}" + } + ], + "images": "RH20T_cfg5_task_0057_user_0007_scene_0002_cfg_0005_037522062165#0", + "gt": "[[289, 152], [307, 153], [313, 165], [300, 229], [318, 215], [317, 204], [302, 227], [302, 234], [308, 230], [299, 229]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "57764_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the black object to the bottom right corner of the countertop', and now you are performing subtask 'move the black object to the bottom right corner of the countertop.'\nBegin at [0.67, 0.25] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.67, 0.25], [0.17, 0.59], [0.14, 0.63], [0.11, 0.78], [0.09, 0.67], [0.14, 0.52], [0.41, 0.34], [0.47, 0.37], [0.47, 0.51], [0.43, 0.35]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/57764_exterior_image_2_left#0.jpg", + "gt": "[[215, 45], [55, 107], [46, 113], [35, 140], [29, 120], [46, 94], [131, 62], [152, 67], [151, 91], [139, 63]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0016_user_0014_scene_0008_cfg_0001_819112070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'hit the pool ball', and now you are performing subtask 'hit the pool ball.'\nFrom start point [0.62, 0.46], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.46], [0.61, 0.45], [0.59, 0.5], [0.58, 0.48], [0.57, 0.51], [0.57, 0.52], [0.58, 0.54], [0.58, 0.55], [0.58, 0.51], [0.56, 0.53]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0016_user_0014_scene_0008_cfg_0001_819112070044#0.jpg", + "gt": "[[399, 166], [391, 162], [379, 179], [373, 174], [366, 182], [366, 188], [374, 193], [372, 197], [371, 185], [361, 192]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0037_user_0014_scene_0002_cfg_0007_104122063633#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the first block from the table, transfer the first block from the tabletop to the second block, place the first block on the second block.'\nYou start from [0.37, 0.34]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.34], [0.36, 0.41], [0.37, 0.44], [0.37, 0.49], [0.38, 0.56], [0.38, 0.52], [0.37, 0.51], [0.36, 0.5], [0.35, 0.51], [0.32, 0.49]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0037_user_0014_scene_0002_cfg_0007_104122063633#0.jpg", + "gt": "[[234, 121], [232, 146], [237, 160], [235, 178], [244, 200], [242, 188], [238, 182], [228, 179], [221, 185], [208, 178]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "10075_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'slide the black and white brush to the left', and now you are performing subtask 'slide the black and white brush to the left.'\nYou begin at [0.38, 0.44]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.44], [0.48, 0.38], [0.53, 0.38], [0.52, 0.41], [0.59, 0.41], [0.62, 0.43], [0.63, 0.52], [0.68, 0.53], [0.56, 0.53], [0.56, 0.53]]}" + } + ], + "images": "10075_exterior_image_1_left#0", + "gt": "[[123, 79], [153, 68], [169, 68], [168, 74], [190, 73], [198, 77], [202, 93], [216, 96], [178, 95], [178, 95]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "30469_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer, then take out a plush toy from the drawer, then close the drawer', and now you are performing subtask 'place the plush toy on the box on the desk.'\nStarting at [0.52, 0.37], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.37], [0.56, 0.36], [0.62, 0.49], [0.62, 0.63], [0.62, 0.51], [0.58, 0.38], [0.48, 0.28], [0.44, 0.32], [0.44, 0.4], [0.48, 0.34]]}" + } + ], + "images": "30469_exterior_image_1_left#614", + "gt": "[[167, 67], [179, 65], [198, 89], [198, 114], [199, 91], [187, 68], [153, 51], [142, 58], [142, 72], [155, 62]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0002_user_0014_scene_0004_cfg_0001_750612070851#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull out a napkin', and now you are performing subtask 'pull out a napkin.'\nYour initial coordinate is [0.5, 0.43]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.43], [0.5, 0.26], [0.46, 0.66], [0.49, 0.49], [0.54, 0.43], [0.62, 0.45], [0.64, 0.49], [0.74, 0.48], [0.83, 0.51], [0.83, 0.51]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0002_user_0014_scene_0004_cfg_0001_750612070851#0.jpg", + "gt": "[[323, 156], [320, 93], [296, 238], [313, 178], [344, 154], [394, 163], [408, 177], [473, 173], [532, 184], [532, 184]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0083_user_0007_scene_0005_cfg_0005_104422070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'swat with a flyswatter', and now you are performing subtask 'swat with a flyswatter.'\nThe trajectory starts at [0.58, 0.37]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.37], [0.57, 0.37], [0.57, 0.35], [0.55, 0.37], [0.56, 0.34], [0.57, 0.36], [0.52, 0.36], [0.51, 0.39], [0.51, 0.38], [0.57, 0.39]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0083_user_0007_scene_0005_cfg_0005_104422070044#0.jpg", + "gt": "[[374, 132], [366, 133], [362, 125], [355, 134], [356, 124], [364, 131], [335, 129], [326, 142], [325, 137], [368, 142]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0037_user_0007_scene_0006_cfg_0002_105422061350#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the third block on the table, transfer the third block from the table to the top of the fourth block, place the third block on the fourth block, pick up the fourth block on the table.'\nThe trajectory starts at [0.46, 0.49]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.49], [0.45, 0.51], [0.46, 0.5], [0.46, 0.51], [0.46, 0.49], [0.49, 0.38], [0.5, 0.4], [0.51, 0.39], [0.5, 0.41], [0.44, 0.52]]}" + } + ], + "images": "RH20T_cfg2_task_0037_user_0007_scene_0006_cfg_0002_105422061350#499", + "gt": "[[292, 176], [289, 183], [292, 179], [293, 185], [292, 177], [312, 136], [317, 145], [324, 139], [320, 149], [282, 187]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "38943_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the tomato in the pot', and now you are performing subtask 'put the tomato in the pot.'\nThe trajectory starts at [0.39, 0.01]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.01], [0.47, 0.13], [0.53, 0.47], [0.56, 0.51], [0.54, 0.52], [0.55, 0.58], [0.54, 0.28], [0.49, 0.18], [0.48, 0.43], [0.45, 0.18]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/38943_exterior_image_1_left#0.jpg", + "gt": "[[125, 1], [149, 23], [169, 84], [179, 91], [172, 94], [177, 104], [172, 51], [157, 33], [153, 78], [145, 32]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0085_user_0007_scene_0007_cfg_0005_104422070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the bubble ring from the assembled bubble ring and ball', and now you are performing subtask 'remove the bubble ring from the assembled bubble ring and ball.'\nFrom start point [0.58, 0.37], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.37], [0.61, 0.33], [0.62, 0.35], [0.64, 0.43], [0.64, 0.45], [0.65, 0.51], [0.65, 0.49], [0.62, 0.48], [0.61, 0.44], [0.61, 0.43]]}" + } + ], + "images": "RH20T_cfg5_task_0085_user_0007_scene_0007_cfg_0005_104422070044#0", + "gt": "[[374, 132], [388, 118], [398, 125], [410, 153], [408, 162], [415, 183], [413, 176], [395, 173], [389, 157], [391, 154]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0037_user_0020_scene_0006_cfg_0007_037522061512#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'transfer the third block from the table to above the first block, pick up the fourth block from the table, transfer the fourth block from the table to above the third block.'\nYou start from [0.33, 0.32]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.33, 0.32], [0.33, 0.3], [0.32, 0.32], [0.3, 0.48], [0.27, 0.45], [0.27, 0.47], [0.28, 0.41], [0.28, 0.3], [0.31, 0.26], [0.32, 0.27]]}" + } + ], + "images": "RH20T_cfg7_task_0037_user_0020_scene_0006_cfg_0007_037522061512#267", + "gt": "[[211, 115], [211, 107], [204, 117], [189, 172], [174, 163], [171, 168], [178, 147], [180, 107], [197, 95], [206, 96]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0034_user_0020_scene_0001_cfg_0006_104122060811#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'pick up the first square on the right, transfer the first square from the right to the center, place the first square in the center, pick up the second square on the right.'\nThe trajectory starts at [0.6, 0.35]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.35], [0.6, 0.38], [0.63, 0.37], [0.64, 0.34], [0.63, 0.38], [0.65, 0.38], [0.61, 0.38], [0.6, 0.4], [0.63, 0.37], [0.64, 0.32]]}" + } + ], + "images": "RH20T_cfg6_task_0034_user_0020_scene_0001_cfg_0006_104122060811#0", + "gt": "[[385, 127], [381, 136], [404, 134], [412, 123], [406, 136], [415, 138], [391, 135], [381, 144], [404, 132], [411, 116]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "21213_exterior_image_2_left#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the two paper bags out of the wooden crate', and now you are performing subtask 'place the first paper bag on the table.'\nWith start location [0.46, 0.7], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.7], [0.46, 0.73], [0.47, 0.76], [0.46, 0.61], [0.44, 0.56], [0.43, 0.56], [0.42, 0.58], [0.43, 0.62], [0.46, 0.66], [0.47, 0.57]]}" + } + ], + "images": "21213_exterior_image_2_left#489", + "gt": "[[147, 126], [146, 132], [150, 137], [148, 109], [142, 100], [137, 101], [135, 105], [138, 112], [148, 119], [151, 103]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "47386_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the object from the blue plate and place it on the table, then pick up another object from the plate and place it on the table, and then slide the towel to the left', and now you are performing subtask 'slide the towel on the tabletop to the left.'\nStarting at [0.57, 0.28], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.28], [0.57, 0.24], [0.58, 0.23], [0.67, 0.23], [0.68, 0.2], [0.75, 0.27], [0.7, 0.42], [0.62, 0.45], [0.6, 0.43], [0.63, 0.27]]}" + } + ], + "images": "47386_exterior_image_1_left#211", + "gt": "[[184, 50], [181, 43], [185, 41], [214, 41], [219, 36], [241, 48], [225, 75], [197, 81], [193, 77], [201, 48]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "45947_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the tap nozzle to the left', and now you are performing subtask 'twist the tap clockwise.'\nFrom start point [0.51, 0.02], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.02], [0.56, 0.21], [0.57, 0.27], [0.6, 0.34], [0.58, 0.38], [0.59, 0.33], [0.58, 0.32], [0.58, 0.33], [0.58, 0.34], [0.58, 0.34]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/45947_exterior_image_1_left#0.jpg", + "gt": "[[162, 3], [179, 37], [183, 49], [191, 62], [186, 68], [189, 59], [187, 58], [185, 59], [185, 61], [185, 61]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0130_user_0014_scene_0004_cfg_0006_104122061018#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb', and now you are performing subtask 'insert the lamp holder into the socket, press the lamp holder button to turn on the light.'\nYou start from [0.47, 0.29]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.29], [0.47, 0.24], [0.48, 0.21], [0.47, 0.22], [0.52, 0.31], [0.53, 0.3], [0.53, 0.31], [0.54, 0.31], [0.56, 0.33], [0.55, 0.35]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0130_user_0014_scene_0004_cfg_0006_104122061018#315.jpg", + "gt": "[[304, 105], [302, 86], [305, 77], [304, 81], [335, 112], [338, 109], [342, 112], [347, 110], [356, 118], [349, 125]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "89978_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'arrange the water bottles in a straight line', and now you are performing subtask 'place the second bottle of water in the specified position in a straight line.'\nBegin at [0.57, 0.56] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.56], [0.67, 0.57], [0.76, 0.62], [0.75, 0.78], [0.73, 0.78], [0.73, 0.74], [0.69, 0.76], [0.71, 0.83], [0.7, 0.82], [0.72, 0.82]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/89978_exterior_image_2_left#0.jpg", + "gt": "[[182, 100], [215, 102], [242, 112], [239, 140], [235, 140], [232, 133], [221, 136], [226, 149], [223, 148], [231, 148]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "74617_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the fries plush toy and put it on the plate', and now you are performing subtask 'pick up the fourth fries toy on the table.'\nStarting at [0.5, 0.69], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.69], [0.51, 0.72], [0.52, 0.72], [0.51, 0.54], [0.49, 0.52], [0.48, 0.56], [0.5, 0.57], [0.51, 0.55], [0.5, 0.54], [0.51, 0.54]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/74617_exterior_image_1_left#252.jpg", + "gt": "[[160, 125], [164, 130], [168, 130], [163, 97], [158, 93], [155, 100], [161, 103], [162, 99], [159, 98], [162, 97]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0071_user_0014_scene_0003_cfg_0002_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close a sliding window', and now you are performing subtask 'close a sliding window.'\nStart at [0.44, 0.49]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.49], [0.47, 0.48], [0.42, 0.53], [0.3, 0.54], [0.29, 0.65], [0.32, 0.72], [0.31, 0.65], [0.32, 0.59], [0.39, 0.54], [0.39, 0.54]]}" + } + ], + "images": "RH20T_cfg2_task_0071_user_0014_scene_0003_cfg_0002_037522062165#0", + "gt": "[[280, 175], [303, 172], [267, 190], [193, 195], [184, 233], [206, 259], [199, 235], [207, 212], [249, 193], [249, 193]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "81862_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stretch out the ends of the charger at opposite sides', and now you are performing subtask 'pull the charger to the left.'\nThe trajectory starts at [0.96, 0.0]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.96, 0.0], [0.85, 0.05], [0.51, 0.42], [0.52, 0.47], [0.59, 0.38], [0.7, 0.43], [0.68, 0.35], [0.65, 0.36], [0.67, 0.48], [0.62, 0.5]]}" + } + ], + "images": "81862_exterior_image_1_left#0", + "gt": "[[307, 0], [273, 9], [163, 76], [168, 84], [189, 69], [225, 77], [219, 63], [209, 64], [214, 86], [199, 90]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0015_user_0001_scene_0006_cfg_0001_750612070851#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'play the drum', and now you are performing subtask 'play the drum.'\nWith start location [0.43, 0.17], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.17], [0.52, 0.44], [0.58, 0.28], [0.55, 0.35], [0.58, 0.49], [0.52, 0.5], [0.53, 0.54], [0.48, 0.29], [0.43, 0.31], [0.44, 0.38]]}" + } + ], + "images": "RH20T_cfg1_task_0015_user_0001_scene_0006_cfg_0001_750612070851#0", + "gt": "[[273, 61], [333, 160], [371, 99], [355, 125], [369, 178], [333, 180], [337, 196], [309, 103], [278, 112], [283, 135]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "48872_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the mug on the left, then place it under the running faucet, and finally place it on the right side of the countertop', and now you are performing subtask 'pick up the mug on the left, then place it under the running faucet, and finally place it on the right side of the countertop.'\nBegin at [0.55, 0.33] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.33], [0.47, 0.37], [0.62, 0.44], [0.58, 0.5], [0.68, 0.48], [0.69, 0.57], [0.66, 0.56], [0.75, 0.59], [0.77, 0.67], [0.68, 0.52]]}" + } + ], + "images": "48872_exterior_image_2_left#0", + "gt": "[[177, 59], [149, 67], [199, 80], [187, 90], [216, 86], [221, 102], [211, 100], [239, 106], [247, 120], [216, 94]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "3439_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the pen from the mug cup and place it on the table', and now you are performing subtask 'remove the pen from the mug cup and place it on the table.'\nWith start location [0.73, 0.0], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.73, 0.0], [0.63, 0.12], [0.64, 0.15], [0.62, 0.16], [0.62, 0.21], [0.63, 0.17], [0.64, 0.09], [0.61, 0.28], [0.6, 0.28], [0.59, 0.26]]}" + } + ], + "images": "3439_exterior_image_2_left#0", + "gt": "[[235, 0], [201, 22], [206, 27], [200, 29], [197, 37], [202, 30], [205, 16], [194, 50], [191, 51], [189, 46]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0034_user_0014_scene_0007_cfg_0007_104122064161#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'place the first block on the right side of the table, pick up the second block from the desktop, transfer the second block from the desktop to the right side of the table, place the second block on the right side of the table.'\nBegin at [0.58, 0.73] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.73], [0.61, 0.66], [0.52, 0.57], [0.56, 0.55], [0.57, 0.62], [0.6, 0.65], [0.59, 0.71], [0.6, 0.78], [0.6, 0.83], [0.58, 0.82]]}" + } + ], + "images": "RH20T_cfg7_task_0034_user_0014_scene_0007_cfg_0007_104122064161#177", + "gt": "[[372, 263], [389, 239], [334, 206], [357, 197], [367, 224], [381, 233], [376, 255], [386, 281], [381, 298], [369, 294]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0035_user_0010_scene_0001_cfg_0003_104422070011#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up a small block', and now you are performing subtask 'pick up a small block.'\nYour initial coordinate is [0.51, 0.3]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.3], [0.52, 0.31], [0.48, 0.35], [0.48, 0.29], [0.47, 0.38], [0.39, 0.46], [0.4, 0.52], [0.39, 0.64], [0.35, 0.61], [0.34, 0.51]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg3_task_0035_user_0010_scene_0001_cfg_0003_104422070011#0.jpg", + "gt": "[[328, 107], [334, 112], [307, 125], [309, 106], [304, 137], [250, 167], [259, 188], [247, 229], [225, 220], [220, 184]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "37064_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the marker and put it above the table', and now you are performing subtask 'pick up the marker and put it above the table.'\nFrom start point [0.22, 0.05], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.22, 0.05], [0.29, 0.45], [0.35, 0.65], [0.37, 0.76], [0.39, 0.62], [0.4, 0.55], [0.38, 0.36], [0.41, 0.46], [0.42, 0.41], [0.4, 0.39]]}" + } + ], + "images": "37064_exterior_image_1_left#0", + "gt": "[[70.0, 8.5], [93.0, 81.5], [113.0, 116.5], [117.0, 136.5], [126.0, 111.5], [127.0, 99.5], [121.0, 65.5], [131.0, 82.5], [135.0, 74.5], [129.0, 69.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "26372_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the washing machine lid', and now you are performing subtask 'open the washing machine lid.'\nFrom start point [0.77, 0.42], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.77, 0.42], [0.59, 0.62], [0.56, 0.81], [0.54, 0.84], [0.52, 0.82], [0.52, 0.86], [0.63, 0.94], [0.6, 0.87], [0.59, 0.68], [0.68, 0.31]]}" + } + ], + "images": "26372_exterior_image_2_left#0", + "gt": "[[247, 75], [188, 112], [178, 146], [173, 151], [168, 148], [168, 155], [201, 170], [192, 156], [188, 122], [217, 55]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "35261_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull the black rod to the right and rotate it', and now you are performing subtask 'twist the black rod clockwise.'\nYour starting position is [0.65, 0.62]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.62], [0.65, 0.64], [0.66, 0.63], [0.66, 0.63], [0.67, 0.61], [0.66, 0.62], [0.68, 0.61], [0.67, 0.63], [0.68, 0.66], [0.71, 0.62]]}" + } + ], + "images": "35261_exterior_image_1_left#127", + "gt": "[[208, 111], [209, 115], [210, 113], [212, 114], [213, 109], [212, 111], [218, 109], [214, 114], [219, 118], [228, 111]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "77960_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker inside the transparent cup', and now you are performing subtask 'put the marker inside the transparent cup.'\nBegin at [0.68, 0.27] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.68, 0.27], [0.62, 0.18], [0.56, 0.16], [0.52, 0.17], [0.52, 0.31], [0.49, 0.4], [0.5, 0.53], [0.47, 0.39], [0.44, 0.63], [0.4, 0.44]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/77960_exterior_image_2_left#0.jpg", + "gt": "[[218, 49], [197, 33], [180, 28], [167, 30], [167, 55], [158, 72], [160, 96], [149, 71], [141, 114], [129, 80]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "88097_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the pen on the table', and now you are performing subtask 'put the pen on the table.'\nYour starting position is [0.76, 0.0]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.76, 0.0], [0.68, 0.13], [0.69, 0.15], [0.62, 0.29], [0.64, 0.31], [0.59, 0.38], [0.65, 0.24], [0.59, 0.5], [0.57, 0.49], [0.57, 0.46]]}" + } + ], + "images": "88097_exterior_image_2_left#0", + "gt": "[[242, 0], [217, 24], [222, 27], [200, 52], [204, 55], [190, 68], [208, 43], [188, 90], [184, 89], [184, 83]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "68837_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker in the mug', and now you are performing subtask 'put the marker in the mug.'\nStarting at [0.38, 0.19], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.19], [0.45, 0.17], [0.56, 0.44], [0.57, 0.22], [0.63, 0.25], [0.62, 0.39], [0.64, 0.42], [0.61, 0.46], [0.62, 0.38], [0.62, 0.38]]}" + } + ], + "images": "68837_exterior_image_1_left#0", + "gt": "[[122.5, 33.5], [145.5, 30.5], [180.5, 78.5], [182.5, 40.5], [201.5, 45.5], [199.5, 70.5], [203.5, 75.5], [196.5, 83.5], [199.5, 67.5], [199.5, 67.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "70471_exterior_image_1_left#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the white and black mugs from the drawer, push the drawer closed, close the open bottom cabinet', and now you are performing subtask 'push the door of the lower cabinet closed.'\nFrom start point [0.56, 0.71], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.56, 0.71], [0.65, 0.73], [0.65, 0.81], [0.62, 0.83], [0.58, 0.82], [0.59, 0.84], [0.57, 0.84], [0.61, 0.79], [0.73, 0.7], [0.71, 0.66]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/70471_exterior_image_1_left#789.jpg", + "gt": "[[180, 128], [207, 131], [208, 145], [200, 150], [186, 147], [188, 152], [184, 152], [195, 142], [233, 126], [226, 118]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "10286_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the keys from the nightstand and put them on the windowsill then place the keys back on the nightstand', and now you are performing subtask 'transfer the key from the nightstand to the windowsill.'\nYour starting position is [0.27, 0.37]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.27, 0.37], [0.28, 0.43], [0.2, 0.43], [0.26, 0.43], [0.25, 0.47], [0.28, 0.53], [0.32, 0.45], [0.47, 0.49], [0.51, 0.57], [0.51, 0.57]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/10286_exterior_image_2_left#0.jpg", + "gt": "[[85, 67], [88, 78], [65, 78], [84, 78], [79, 84], [88, 96], [101, 81], [151, 89], [163, 102], [163, 102]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg3_task_0032_user_0011_scene_0001_cfg_0003_104122062823#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'slide the outer casing onto the gift box', and now you are performing subtask 'slide the outer casing onto the gift box.'\nBegin at [0.46, 0.44] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.44], [0.42, 0.39], [0.41, 0.33], [0.38, 0.28], [0.32, 0.21], [0.36, 0.26], [0.34, 0.24], [0.37, 0.26], [0.34, 0.22], [0.41, 0.29]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg3_task_0032_user_0011_scene_0001_cfg_0003_104122062823#195.jpg", + "gt": "[[294, 159], [269, 141], [260, 119], [245, 100], [206, 76], [230, 94], [219, 85], [234, 93], [216, 81], [260, 103]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "52362_exterior_image_1_left#TrajectoryPrediction_Task#12", + "task": "TrajectoryPrediction_Task", + "p_index": 12, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'arrange the two small pillows against the right armrest of the sofa', and now you are performing subtask 'pull the left pillow to the right armrest.'\nStart at [0.54, 0.62]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.62], [0.52, 0.68], [0.51, 0.65], [0.53, 0.66], [0.54, 0.59], [0.61, 0.59], [0.66, 0.68], [0.63, 0.63], [0.63, 0.56], [0.57, 0.53]]}" + } + ], + "images": "52362_exterior_image_1_left#610", + "gt": "[[174, 112], [165, 122], [163, 117], [171, 118], [173, 107], [194, 106], [212, 123], [203, 114], [202, 101], [181, 95]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "24992_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the two cushions on the right side of the sofa', and now you are performing subtask 'pull the second cushion to the left side of the first cushion.'\nYou start from [0.66, 0.57]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.66, 0.57], [0.64, 0.53], [0.53, 0.57], [0.48, 0.67], [0.48, 0.7], [0.5, 0.7], [0.49, 0.77], [0.55, 0.72], [0.57, 0.61], [0.6, 0.57]]}" + } + ], + "images": "24992_exterior_image_1_left#189", + "gt": "[[211, 102], [206, 95], [171, 102], [155, 120], [155, 126], [160, 126], [157, 138], [176, 129], [183, 110], [192, 103]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "12154_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the cupboard door', and now you are performing subtask 'open the cupboard door.'\nWith start location [0.77, 0.02], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.77, 0.02], [0.75, 0.09], [0.7, 0.19], [0.57, 0.28], [0.51, 0.43], [0.51, 0.39], [0.5, 0.38], [0.49, 0.41], [0.5, 0.38], [0.51, 0.41]]}" + } + ], + "images": "12154_exterior_image_1_left#0", + "gt": "[[245, 3], [241, 17], [225, 34], [182, 50], [162, 77], [163, 70], [159, 69], [158, 74], [161, 69], [162, 73]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "4309_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull out a tissue square from the pink box', and now you are performing subtask 'pull out a tissue square from the pink box.'\nThe trajectory starts at [0.49, 0.53]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.53], [0.53, 0.67], [0.61, 0.79], [0.59, 0.84], [0.61, 0.9], [0.58, 0.76], [0.58, 0.65], [0.69, 0.63], [0.7, 0.7], [0.65, 0.74]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/4309_exterior_image_1_left#0.jpg", + "gt": "[[156.5, 95.5], [168.5, 120.5], [194.5, 141.5], [187.5, 150.5], [194.5, 161.5], [185.5, 137.5], [186.5, 117.5], [220.5, 113.5], [223.5, 125.5], [208.5, 133.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "42824_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the marker from the cup and put it on the table', and now you are performing subtask 'pick up the marker from the cup and put it on the table.'\nWith start location [0.74, 0.15], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.74, 0.15], [0.69, 0.26], [0.67, 0.33], [0.67, 0.34], [0.64, 0.38], [0.61, 0.39], [0.6, 0.41], [0.57, 0.32], [0.61, 0.53], [0.58, 0.47]]}" + } + ], + "images": "42824_exterior_image_1_left#0", + "gt": "[[237, 27], [220, 46], [213, 59], [215, 62], [206, 69], [195, 70], [191, 74], [183, 58], [196, 96], [186, 85]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0003_user_0010_scene_0001_cfg_0005_104122063678#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the green button on the left.'\nBegin at [0.52, 0.39] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.39], [0.5, 0.36], [0.52, 0.36], [0.51, 0.4], [0.52, 0.33], [0.53, 0.33], [0.51, 0.27], [0.53, 0.36], [0.53, 0.39], [0.52, 0.39]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0003_user_0010_scene_0001_cfg_0005_104122063678#406.jpg", + "gt": "[[331, 142], [322, 130], [331, 129], [327, 143], [332, 120], [338, 118], [326, 98], [339, 129], [339, 139], [332, 140]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "55641_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the cloth from the rack and put it on the table', and now you are performing subtask 'remove the cloth from the rack and put it on the table.'\nFrom start point [0.44, 0.23], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.23], [0.47, 0.23], [0.48, 0.26], [0.47, 0.29], [0.47, 0.28], [0.43, 0.31], [0.43, 0.3], [0.46, 0.27], [0.53, 0.27], [0.56, 0.31]]}" + } + ], + "images": "55641_exterior_image_1_left#0", + "gt": "[[141, 42], [152, 41], [154, 46], [152, 52], [151, 50], [139, 56], [139, 54], [148, 48], [171, 49], [178, 56]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "81924_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the can in the cabinet', and now you are performing subtask 'put the can in the cabinet.'\nYou start from [0.56, 0.4]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.56, 0.4], [0.71, 0.6], [0.67, 0.73], [0.64, 0.73], [0.58, 0.61], [0.57, 0.33], [0.52, 0.07], [0.52, 0.14], [0.53, 0.09], [0.54, 0.12]]}" + } + ], + "images": "81924_exterior_image_2_left#0", + "gt": "[[178, 72], [227, 108], [213, 132], [204, 131], [185, 110], [182, 60], [167, 12], [166, 26], [169, 16], [173, 21]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "21790_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the green item and place it in the box', and now you are performing subtask 'pick up the green item and place it in the box.'\nBegin at [0.45, 0.0] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.0], [0.5, 0.08], [0.47, 0.36], [0.43, 0.22], [0.38, 0.23], [0.39, 0.34], [0.4, 0.26], [0.46, 0.23], [0.46, 0.27], [0.39, 0.24]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/21790_exterior_image_2_left#0.jpg", + "gt": "[[145, 0], [159, 14], [150, 64], [136, 40], [121, 42], [125, 62], [129, 46], [148, 41], [146, 49], [124, 44]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0052_user_0007_scene_0010_cfg_0005_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grab a basketball, release it and shoot it into the basket', and now you are performing subtask 'grab a basketball, release it and shoot it into the basket.'\nFrom start point [0.45, 0.42], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.42], [0.42, 0.54], [0.43, 0.8], [0.38, 0.63], [0.33, 0.27], [0.34, 0.19], [0.27, 0.04], [0.26, 0.16], [0.43, 0.36], [0.52, 0.41]]}" + } + ], + "images": "RH20T_cfg5_task_0052_user_0007_scene_0010_cfg_0005_037522062165#0", + "gt": "[[289, 152], [271, 193], [277, 288], [245, 228], [209, 96], [217, 68], [173, 13], [169, 57], [278, 130], [335, 147]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0121_user_0010_scene_0002_cfg_0001_819112070044#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'taking out the light bulb by rotating it', and now you are performing subtask 'twist the light bulb counterclockwise to loosen it.'\nFrom start point [0.62, 0.37], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.37], [0.62, 0.36], [0.62, 0.35], [0.62, 0.36], [0.62, 0.37], [0.62, 0.37], [0.62, 0.37], [0.62, 0.37], [0.62, 0.37], [0.62, 0.37]]}" + } + ], + "images": "RH20T_cfg1_task_0121_user_0010_scene_0002_cfg_0001_819112070044#203", + "gt": "[[396, 133], [397, 128], [400, 126], [399, 131], [396, 133], [399, 134], [398, 133], [396, 134], [395, 132], [395, 132]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0058_user_0014_scene_0001_cfg_0006_104122063633#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the switches on the power strip in sequence', and now you are performing subtask 'press the third button on the power strip, press the fourth button on the power strip.'\nWith start location [0.37, 0.52], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.52], [0.37, 0.53], [0.37, 0.53], [0.37, 0.54], [0.37, 0.54], [0.38, 0.53], [0.38, 0.54], [0.39, 0.55], [0.39, 0.55], [0.4, 0.55]]}" + } + ], + "images": "RH20T_cfg6_task_0058_user_0014_scene_0001_cfg_0006_104122063633#208", + "gt": "[[236, 188], [238, 192], [238, 191], [237, 193], [236, 193], [242, 191], [243, 194], [247, 198], [249, 199], [253, 199]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "56862_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the bottle of water, the black bag, the dark blue bag, the top plastic and the remaining snack pack in the box respectively', and now you are performing subtask 'place the packaging bag in the box.'\nFrom start point [0.54, 0.59], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.59], [0.4, 0.6], [0.42, 0.69], [0.38, 0.79], [0.36, 0.7], [0.38, 0.52], [0.46, 0.42], [0.51, 0.42], [0.55, 0.51], [0.53, 0.62]]}" + } + ], + "images": "56862_exterior_image_1_left#190", + "gt": "[[172, 107], [128, 108], [133, 124], [123, 143], [116, 126], [123, 93], [147, 76], [164, 76], [176, 91], [170, 112]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0009_user_0010_scene_0006_cfg_0004_104422070011#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take out a Hanoi block and throw it aside', and now you are performing subtask 'take out a Hanoi block and throw it aside.'\nYou begin at [0.53, 0.29]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.29], [0.52, 0.44], [0.48, 0.46], [0.54, 0.44], [0.58, 0.5], [0.58, 0.46], [0.56, 0.45], [0.58, 0.47], [0.58, 0.58], [0.58, 0.58]]}" + } + ], + "images": "RH20T_cfg4_task_0009_user_0010_scene_0006_cfg_0004_104422070011#0", + "gt": "[[340, 104], [330, 158], [307, 165], [347, 159], [369, 180], [371, 164], [357, 162], [372, 168], [369, 208], [369, 208]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "20992_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the candy bar in the sink', and now you are performing subtask 'put the candy bar in the sink.'\nYou start from [0.76, 0.41]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.76, 0.41], [0.62, 0.61], [0.56, 0.83], [0.56, 0.87], [0.57, 0.73], [0.61, 0.65], [0.64, 0.66], [0.68, 0.81], [0.68, 0.74], [0.72, 0.68]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/20992_exterior_image_2_left#0.jpg", + "gt": "[[242, 73], [198, 109], [178, 149], [180, 156], [182, 132], [195, 117], [205, 119], [218, 145], [218, 134], [231, 123]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "17981_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the orange cloth', and now you are performing subtask 'fold the cloth on the table.'\nBegin at [0.65, 0.26] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.26], [0.54, 0.32], [0.58, 0.31], [0.51, 0.46], [0.56, 0.42], [0.53, 0.48], [0.57, 0.48], [0.59, 0.42], [0.53, 0.46], [0.57, 0.4]]}" + } + ], + "images": "17981_exterior_image_1_left#0", + "gt": "[[209, 46], [174, 57], [187, 55], [162, 82], [180, 75], [171, 87], [183, 86], [188, 75], [170, 82], [182, 72]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0031_user_0020_scene_0007_cfg_0007_104122060811#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'cover the box', and now you are performing subtask 'cover the box.'\nStarting at [0.6, 0.36], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.36], [0.6, 0.36], [0.6, 0.37], [0.61, 0.34], [0.61, 0.36], [0.62, 0.33], [0.6, 0.32], [0.52, 0.37], [0.55, 0.34], [0.57, 0.34]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0031_user_0020_scene_0007_cfg_0007_104122060811#0.jpg", + "gt": "[[387, 130], [385, 128], [381, 132], [390, 124], [393, 131], [395, 119], [384, 115], [331, 134], [353, 122], [364, 121]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "2844_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the jar into the box', and now you are performing subtask 'put the jar into the box.'\nStart at [0.55, 0.39]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.39], [0.52, 0.35], [0.58, 0.28], [0.6, 0.53], [0.63, 0.31], [0.53, 0.05], [0.45, 0.12], [0.41, 0.38], [0.47, 0.26], [0.56, 0.24]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/2844_exterior_image_1_left#0.jpg", + "gt": "[[176, 70], [165, 63], [187, 51], [192, 96], [203, 56], [171, 9], [143, 21], [131, 69], [151, 46], [178, 44]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg3_task_0003_user_0016_scene_0005_cfg_0003_104122062823#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons in sequence', and now you are performing subtask 'press the green button on the yellow and white base, press the green button on the gray and white base.'\nWith start location [0.53, 0.29], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.29], [0.52, 0.25], [0.54, 0.19], [0.56, 0.19], [0.6, 0.21], [0.6, 0.24], [0.58, 0.21], [0.57, 0.19], [0.51, 0.17], [0.49, 0.17]]}" + } + ], + "images": "RH20T_cfg3_task_0003_user_0016_scene_0005_cfg_0003_104122062823#228", + "gt": "[[340, 103], [333, 89], [347, 69], [357, 68], [381, 75], [381, 88], [374, 74], [366, 70], [325, 61], [311, 63]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "14083_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'spoon out some contents from the bowl on the left using the silver spoon and place them in the bowl on the right', and now you are performing subtask 'place the spoon in the bowl on the right.'\nYour initial coordinate is [0.3, 0.48]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.3, 0.48], [0.3, 0.52], [0.27, 0.49], [0.31, 0.43], [0.3, 0.52], [0.31, 0.46], [0.31, 0.55], [0.35, 0.62], [0.29, 0.46], [0.32, 0.48]]}" + } + ], + "images": "14083_exterior_image_2_left#97", + "gt": "[[96, 86], [95, 94], [86, 88], [99, 77], [96, 93], [100, 82], [99, 99], [111, 112], [94, 83], [102, 86]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "61454_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the spoon and ladle from the rack, put them on the table, and then move the rack to the right', and now you are performing subtask 'push the rack to the right.'\nYour initial coordinate is [0.86, 0.3]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.86, 0.3], [0.8, 0.21], [0.65, 0.24], [0.67, 0.29], [0.66, 0.16], [0.78, 0.09], [0.86, 0.17], [0.88, 0.25], [0.79, 0.06], [0.79, 0.06]]}" + } + ], + "images": "61454_exterior_image_1_left#380", + "gt": "[[276.0, 53.5], [256.0, 37.5], [208.0, 43.5], [213.0, 51.5], [212.0, 29.5], [250.0, 16.5], [276.0, 31.5], [281.0, 45.5], [252.0, 10.5], [252.0, 10.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0037_user_0014_scene_0002_cfg_0007_104122063633#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the fourth block on the third block, pick up the fifth block from the table, transfer the fifth block from the table to the fourth block, place the fifth block on top of the fourth block.'\nThe trajectory starts at [0.27, 0.43]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.27, 0.43], [0.26, 0.57], [0.26, 0.5], [0.28, 0.43], [0.29, 0.41], [0.29, 0.38], [0.3, 0.3], [0.33, 0.28], [0.35, 0.34], [0.35, 0.19]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0037_user_0014_scene_0002_cfg_0007_104122063633#392.jpg", + "gt": "[[171, 154], [169, 204], [169, 179], [177, 155], [184, 148], [183, 135], [193, 109], [209, 102], [223, 124], [226, 68]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0071_user_0010_scene_0008_cfg_0003_104122063550#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close a sliding window', and now you are performing subtask 'push the sliding window to close.'\nFrom start point [0.38, 0.6], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.6], [0.35, 0.61], [0.38, 0.57], [0.4, 0.56], [0.41, 0.57], [0.41, 0.58], [0.4, 0.61], [0.39, 0.61], [0.4, 0.6], [0.41, 0.6]]}" + } + ], + "images": "RH20T_cfg3_task_0071_user_0010_scene_0008_cfg_0003_104122063550#131", + "gt": "[[241, 216], [224, 219], [246, 204], [254, 202], [260, 204], [265, 209], [258, 218], [252, 220], [255, 215], [262, 215]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "76544_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the green marker from the table and put it in the bowl', and now you are performing subtask 'remove the green marker from the table and put it in the bowl.'\nStarting at [0.67, 0.27], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.67, 0.27], [0.6, 0.23], [0.52, 0.28], [0.45, 0.49], [0.46, 0.58], [0.44, 0.56], [0.5, 0.43], [0.53, 0.43], [0.56, 0.52], [0.57, 0.46]]}" + } + ], + "images": "76544_exterior_image_2_left#0", + "gt": "[[214, 48], [192, 41], [168, 51], [145, 89], [146, 105], [141, 100], [160, 78], [170, 77], [179, 94], [182, 83]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122061602#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the block on the surface, transfer the block from the surface to the other block, place the block on the other block.'\nYour starting position is [0.7, 0.29]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.29], [0.69, 0.37], [0.68, 0.39], [0.69, 0.43], [0.68, 0.57], [0.68, 0.54], [0.66, 0.51], [0.65, 0.43], [0.63, 0.43], [0.62, 0.46]]}" + } + ], + "images": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122061602#0", + "gt": "[[446, 106], [444, 133], [435, 139], [441, 154], [434, 207], [435, 196], [421, 184], [415, 155], [404, 156], [397, 167]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0033_user_0015_scene_0008_cfg_0001_038522063145#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the block', and now you are performing subtask 'push the square in front of the cardboard to the adjacent square in the middle of the cardboard, pick up the square on the cardboard.'\nBegin at [0.69, 0.27] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.27], [0.73, 0.26], [0.73, 0.26], [0.72, 0.29], [0.69, 0.3], [0.68, 0.31], [0.68, 0.33], [0.65, 0.37], [0.67, 0.34], [0.66, 0.36]]}" + } + ], + "images": "RH20T_cfg1_task_0033_user_0015_scene_0008_cfg_0001_038522063145#0", + "gt": "[[443, 98], [470, 92], [470, 95], [463, 103], [444, 107], [438, 113], [437, 120], [418, 132], [428, 122], [424, 128]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "61405_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the plushy to the right', and now you are performing subtask 'move the plushy to the right.'\nStarting at [0.25, 0.02], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.25, 0.02], [0.35, 0.36], [0.39, 0.54], [0.42, 0.6], [0.42, 0.64], [0.44, 0.58], [0.46, 0.59], [0.51, 0.75], [0.53, 0.73], [0.5, 0.66]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/61405_exterior_image_1_left#0.jpg", + "gt": "[[80.5, 3.0], [113.5, 64.0], [125.5, 98.0], [133.5, 108.0], [133.5, 116.0], [139.5, 105.0], [148.5, 106.0], [164.5, 135.0], [168.5, 131.0], [160.5, 119.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "77449_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the washing machine's detergent compartment', and now you are performing subtask 'open the washing machine's detergent compartment.'\nWith start location [0.53, 0.33], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.33], [0.57, 0.35], [0.58, 0.26], [0.54, 0.59], [0.53, 0.63], [0.52, 0.58], [0.54, 0.57], [0.55, 0.63], [0.58, 0.61], [0.54, 0.47]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/77449_exterior_image_2_left#0.jpg", + "gt": "[[171, 60], [183, 63], [186, 47], [174, 106], [169, 114], [165, 104], [173, 103], [175, 113], [186, 109], [174, 84]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0012_user_0010_scene_0005_cfg_0004_104122062823#TrajectoryPrediction_Task#10", + "task": "TrajectoryPrediction_Task", + "p_index": 10, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push the soccer ball into the goal', and now you are performing subtask 'push the soccer ball toward the goal.'\nStarting at [0.39, 0.53], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.53], [0.4, 0.53], [0.36, 0.51], [0.35, 0.57], [0.39, 0.62], [0.43, 0.63], [0.37, 0.48], [0.37, 0.3], [0.4, 0.21], [0.42, 0.18]]}" + } + ], + "images": "RH20T_cfg4_task_0012_user_0010_scene_0005_cfg_0004_104122062823#255", + "gt": "[[247, 192], [257, 190], [231, 183], [225, 204], [251, 222], [275, 226], [235, 172], [239, 107], [255, 77], [270, 64]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "20196_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'arrange the cushions in front of the armrest', and now you are performing subtask 'place the second cushion in front of the first cushion, against the armrest.'\nStart at [0.52, 0.57]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.57], [0.56, 0.63], [0.52, 0.69], [0.52, 0.59], [0.6, 0.52], [0.67, 0.58], [0.65, 0.53], [0.55, 0.55], [0.52, 0.68], [0.74, 0.61]]}" + } + ], + "images": "20196_exterior_image_1_left#172", + "gt": "[[166, 103], [179, 113], [167, 125], [166, 107], [192, 93], [215, 105], [208, 95], [175, 99], [167, 123], [237, 109]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "50242_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the white cup and put it on the orange object, then pick up the black cup and put it on the green object', and now you are performing subtask 'place the black cup on the green object.'\nBegin at [0.38, 0.5] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.5], [0.44, 0.62], [0.44, 0.64], [0.41, 0.62], [0.38, 0.67], [0.4, 0.66], [0.4, 0.67], [0.39, 0.62], [0.37, 0.6], [0.37, 0.6]]}" + } + ], + "images": "50242_exterior_image_2_left#202", + "gt": "[[120, 90], [141, 111], [140, 115], [130, 111], [123, 120], [129, 119], [128, 121], [126, 112], [117, 108], [117, 108]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "44871_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the black cloth and put it on the table', and now you are performing subtask 'pick up the black cloth and put it on the table.'\nYou start from [0.46, 0.29]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.29], [0.54, 0.32], [0.58, 0.38], [0.59, 0.49], [0.54, 0.49], [0.63, 0.26], [0.7, 0.16], [0.66, 0.45], [0.74, 0.64], [0.61, 0.54]]}" + } + ], + "images": "44871_exterior_image_1_left#0", + "gt": "[[146, 53], [174, 58], [186, 68], [189, 88], [172, 89], [201, 46], [223, 28], [212, 81], [238, 116], [194, 97]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "41068_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the drawer and then put the brown object from the bowl on the green cloth', and now you are performing subtask 'pull the drawer handle to open the drawer.'\nFrom start point [0.81, 0.02], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.81, 0.02], [0.8, 0.26], [0.78, 0.38], [0.73, 0.42], [0.67, 0.43], [0.7, 0.49], [0.72, 0.36], [0.69, 0.32], [0.64, 0.32], [0.61, 0.41]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/41068_exterior_image_2_left#0.jpg", + "gt": "[[259, 4], [257, 47], [250, 68], [233, 76], [215, 78], [225, 88], [230, 64], [222, 57], [205, 57], [194, 73]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "32668_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the doors of the second shelf of the cabinet', and now you are performing subtask 'pull the handle of the right door of the second shelf to open the door.'\nYour initial coordinate is [0.18, 0.15]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.18, 0.15], [0.26, 0.23], [0.38, 0.27], [0.36, 0.28], [0.38, 0.28], [0.36, 0.29], [0.37, 0.32], [0.37, 0.19], [0.32, 0.03], [0.31, 0.08]]}" + } + ], + "images": "32668_exterior_image_1_left#0", + "gt": "[[57, 27], [82, 41], [122, 48], [114, 50], [120, 51], [115, 53], [119, 57], [119, 35], [102, 5], [98, 14]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "71515_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the yellow block on its side', and now you are performing subtask 'place the yellow block on its side.'\nBegin at [0.83, 0.0] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.83, 0.0], [0.68, 0.23], [0.57, 0.27], [0.5, 0.34], [0.53, 0.36], [0.49, 0.31], [0.49, 0.36], [0.56, 0.43], [0.57, 0.36], [0.57, 0.26]]}" + } + ], + "images": "71515_exterior_image_2_left#0", + "gt": "[[266, 0], [216, 41], [182, 49], [160, 62], [169, 64], [158, 55], [156, 65], [179, 77], [183, 65], [181, 47]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0022_user_0003_scene_0004_cfg_0002_105422061350#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'wave the flag', and now you are performing subtask 'wave the flag.'\nYour starting position is [0.58, 0.43]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.43], [0.56, 0.44], [0.5, 0.31], [0.5, 0.38], [0.36, 0.53], [0.37, 0.57], [0.49, 0.36], [0.52, 0.23], [0.41, 0.48], [0.5, 0.38]]}" + } + ], + "images": "RH20T_cfg2_task_0022_user_0003_scene_0004_cfg_0002_105422061350#0", + "gt": "[[369, 155], [357, 159], [323, 113], [320, 135], [233, 190], [236, 206], [313, 129], [330, 84], [265, 172], [323, 136]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "68771_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the marker from the cup and put it on the left side of the table', and now you are performing subtask 'remove the marker from the cup and put it on the left side of the table.'\nYour initial coordinate is [0.25, 0.0]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.25, 0.0], [0.57, 0.12], [0.56, 0.22], [0.47, 0.37], [0.48, 0.48], [0.53, 0.46], [0.54, 0.37], [0.5, 0.39], [0.53, 0.19], [0.38, 0.22]]}" + } + ], + "images": "68771_exterior_image_2_left#0", + "gt": "[[80, 0], [182, 22], [179, 39], [152, 67], [153, 87], [170, 83], [173, 67], [159, 70], [169, 35], [121, 40]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "23683_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the lid from the table and put it on the pot', and now you are performing subtask 'pick up the lid from the table and put it on the pot.'\nYour starting position is [0.89, 0.0]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.89, 0.0], [0.81, 0.33], [0.73, 0.45], [0.74, 0.59], [0.72, 0.44], [0.62, 0.44], [0.69, 0.36], [0.7, 0.3], [0.67, 0.3], [0.67, 0.3]]}" + } + ], + "images": "23683_exterior_image_2_left#0", + "gt": "[[284, 0], [260, 59], [233, 81], [236, 107], [229, 79], [198, 79], [221, 65], [224, 54], [215, 54], [215, 54]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "66608_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the spoon with the other cutlery then put the red ball on the plate', and now you are performing subtask 'place the red ball on the plate.'\nStart at [0.61, 0.54]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.61, 0.54], [0.73, 0.81], [0.66, 0.72], [0.62, 0.76], [0.61, 0.68], [0.58, 0.67], [0.55, 0.84], [0.58, 1.0], [0.55, 1.0], [0.62, 0.77]]}" + } + ], + "images": "66608_exterior_image_1_left#272", + "gt": "[[195, 97], [234, 145], [212, 129], [198, 137], [195, 123], [185, 121], [177, 152], [186, 180], [176, 180], [198, 138]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0101_user_0007_scene_0007_cfg_0005_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'clean the table with a cloth', and now you are performing subtask 'clean the table with a cloth.'\nYou begin at [0.48, 0.14]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.14], [0.54, 0.12], [0.62, 0.21], [0.62, 0.24], [0.56, 0.48], [0.59, 0.31], [0.62, 0.33], [0.58, 0.39], [0.55, 0.39], [0.55, 0.26]]}" + } + ], + "images": "RH20T_cfg5_task_0101_user_0007_scene_0007_cfg_0005_104122061850#0", + "gt": "[[305, 50], [343, 43], [394, 74], [399, 86], [358, 172], [377, 111], [399, 118], [371, 141], [355, 142], [352, 93]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "10952_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick the wooden object on the table, put it on the case cover, take the container out of the case and put it on the box then place the dumbbell at the edge of the box', and now you are performing subtask 'pick up the dumbbell on the table.'\nStart at [0.64, 0.24]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.64, 0.24], [0.62, 0.8], [0.57, 0.78], [0.57, 0.35], [0.53, 0.48], [0.58, 0.52], [0.56, 0.44], [0.49, 0.5], [0.52, 0.47], [0.53, 0.57]]}" + } + ], + "images": "10952_exterior_image_1_left#0", + "gt": "[[205.5, 44.0], [197.5, 144.0], [182.5, 141.0], [181.5, 63.0], [169.5, 86.0], [185.5, 93.0], [178.5, 80.0], [157.5, 90.0], [167.5, 84.0], [170.5, 103.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "62551_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the faucet, open the faucet, close the faucet', and now you are performing subtask 'pull the faucet up.'\nYou begin at [0.29, 0.44]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.29, 0.44], [0.28, 0.48], [0.26, 0.48], [0.27, 0.48], [0.26, 0.51], [0.25, 0.48], [0.23, 0.48], [0.22, 0.45], [0.22, 0.47], [0.22, 0.47]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/62551_exterior_image_1_left#178.jpg", + "gt": "[[92, 80], [89, 86], [82, 86], [85, 87], [82, 91], [79, 86], [75, 87], [69, 81], [71, 85], [71, 85]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "64566_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'switch off the toaster', and now you are performing subtask 'switch off the toaster.'\nWith start location [0.66, 0.0], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.66, 0.0], [0.63, 0.02], [0.48, 0.25], [0.42, 0.29], [0.41, 0.34], [0.43, 0.36], [0.41, 0.37], [0.41, 0.34], [0.47, 0.36], [0.55, 0.32]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/64566_exterior_image_2_left#0.jpg", + "gt": "[[212, 0], [202, 4], [154, 45], [133, 53], [131, 61], [136, 65], [132, 66], [132, 62], [152, 65], [175, 58]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0121_user_0010_scene_0002_cfg_0001_819112070044#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'taking out the light bulb by rotating it', and now you are performing subtask 'twist the light bulb counterclockwise to loosen it.'\nFrom start point [0.62, 0.36], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.36], [0.62, 0.34], [0.62, 0.34], [0.62, 0.34], [0.62, 0.35], [0.62, 0.35], [0.62, 0.36], [0.62, 0.36], [0.62, 0.37], [0.62, 0.37]]}" + } + ], + "images": "RH20T_cfg1_task_0121_user_0010_scene_0002_cfg_0001_819112070044#384", + "gt": "[[398, 131], [397, 124], [399, 124], [399, 123], [399, 125], [397, 126], [398, 130], [398, 129], [395, 132], [395, 132]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0015_user_0001_scene_0006_cfg_0001_038522063145#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'play the drum', and now you are performing subtask 'play the drum.'\nStarting at [0.69, 0.27], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.27], [0.67, 0.32], [0.63, 0.32], [0.62, 0.36], [0.61, 0.28], [0.62, 0.24], [0.7, 0.17], [0.73, 0.2], [0.7, 0.22], [0.7, 0.31]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0015_user_0001_scene_0006_cfg_0001_038522063145#0.jpg", + "gt": "[[442, 98], [428, 114], [403, 117], [397, 131], [391, 100], [398, 87], [447, 62], [464, 71], [445, 80], [447, 112]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "69215_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the plush toy out of the blue pan and place it on the right side of the table', and now you are performing subtask 'take the plush toy out of the blue pan and place it on the right side of the table.'\nYour initial coordinate is [0.64, 0.01]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.64, 0.01], [0.64, 0.33], [0.57, 0.32], [0.57, 0.35], [0.54, 0.26], [0.57, 0.16], [0.68, 0.11], [0.78, 0.27], [0.73, 0.53], [0.72, 0.34]]}" + } + ], + "images": "69215_exterior_image_1_left#0", + "gt": "[[204, 1], [206, 59], [181, 57], [183, 63], [173, 46], [182, 28], [219, 20], [249, 48], [232, 96], [229, 61]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "86836_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the towel', and now you are performing subtask 'fold the towel on the table.'\nYour starting position is [0.21, 0.91]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.21, 0.91], [0.38, 0.67], [0.44, 0.77], [0.44, 0.93], [0.42, 0.82], [0.45, 0.56], [0.43, 0.86], [0.44, 0.74], [0.47, 0.85], [0.4, 0.94]]}" + } + ], + "images": "86836_exterior_image_1_left#236", + "gt": "[[67, 164], [123, 120], [141, 138], [140, 167], [133, 147], [144, 100], [137, 155], [142, 133], [149, 153], [127, 170]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122061018#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'place the block on the block board, pick up the block on the flat surface, transfer the block from the flat surface to the block board, place the block on the block board.'\nYou start from [0.47, 0.26]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.26], [0.45, 0.26], [0.41, 0.31], [0.37, 0.41], [0.39, 0.33], [0.43, 0.27], [0.49, 0.33], [0.55, 0.35], [0.56, 0.4], [0.56, 0.42]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122061018#333.jpg", + "gt": "[[301, 95], [285, 93], [261, 110], [234, 148], [248, 119], [274, 97], [315, 119], [349, 127], [358, 144], [356, 151]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "53827_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the pen and place it in the green cup', and now you are performing subtask 'pick up the pen and place it in the green cup.'\nYour starting position is [0.66, 0.0]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.66, 0.0], [0.6, 0.15], [0.65, 0.35], [0.64, 0.49], [0.58, 0.28], [0.54, 0.3], [0.57, 0.32], [0.54, 0.39], [0.52, 0.26], [0.57, 0.22]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/53827_exterior_image_1_left#0.jpg", + "gt": "[[211.0, 0.5], [192.0, 27.5], [208.0, 62.5], [206.0, 87.5], [185.0, 49.5], [173.0, 54.5], [181.0, 58.5], [173.0, 70.5], [168.0, 46.5], [182.0, 39.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "10958_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the green doll in the pot', and now you are performing subtask 'put the green doll in the pot.'\nYour initial coordinate is [0.38, 0.01]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.01], [0.34, 0.2], [0.37, 0.44], [0.3, 0.34], [0.24, 0.39], [0.28, 0.46], [0.3, 0.44], [0.27, 0.43], [0.28, 0.49], [0.28, 0.43]]}" + } + ], + "images": "10958_exterior_image_1_left#0", + "gt": "[[122.0, 1.5], [108.0, 36.5], [117.0, 78.5], [97.0, 61.5], [77.0, 69.5], [91.0, 82.5], [95.0, 79.5], [86.0, 77.5], [88.0, 87.5], [91.0, 77.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "91226_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the green marker inside the yellow mug', and now you are performing subtask 'put the green marker inside the yellow mug.'\nYou begin at [0.6, 0.0]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.0], [0.57, 0.02], [0.57, 0.27], [0.6, 0.17], [0.59, 0.1], [0.58, 0.12], [0.59, 0.2], [0.57, 0.27], [0.55, 0.25], [0.56, 0.26]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/91226_exterior_image_1_left#0.jpg", + "gt": "[[191, 0], [181, 4], [181, 48], [191, 31], [190, 18], [186, 21], [189, 36], [182, 48], [175, 45], [179, 47]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_104122063633#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the fourth block on the table, transfer the fourth block from the table to above the third block, place the fourth block on top of the third block, pick up the fifth block on the table.'\nWith start location [0.35, 0.36], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.36], [0.28, 0.36], [0.27, 0.43], [0.26, 0.53], [0.28, 0.38], [0.33, 0.25], [0.38, 0.26], [0.38, 0.29], [0.34, 0.22], [0.33, 0.36]]}" + } + ], + "images": "RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_104122063633#1024", + "gt": "[[222, 131], [182, 128], [170, 153], [168, 191], [178, 135], [213, 89], [242, 95], [246, 105], [220, 80], [213, 128]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0003_user_0006_scene_0007_cfg_0002_037522062165#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the three buttons in sequence', and now you are performing subtask 'press the gray base red button.'\nYou start from [0.35, 0.37]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.37], [0.38, 0.37], [0.39, 0.39], [0.42, 0.42], [0.43, 0.47], [0.42, 0.49], [0.42, 0.49], [0.43, 0.43], [0.43, 0.47], [0.43, 0.47]]}" + } + ], + "images": "RH20T_cfg2_task_0003_user_0006_scene_0007_cfg_0002_037522062165#216", + "gt": "[[226, 133], [240, 134], [247, 142], [267, 151], [273, 171], [266, 176], [267, 178], [275, 156], [275, 169], [275, 169]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "68447_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'transfer the lid from the black pot to the blue pot', and now you are performing subtask 'transfer the lid from the black pot to the blue pot.'\nYou begin at [0.92, 0.54]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.92, 0.54], [0.91, 0.67], [0.86, 0.76], [0.7, 0.82], [0.59, 0.97], [0.59, 0.85], [0.61, 0.76], [0.6, 0.7], [0.72, 0.71], [0.75, 0.65]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/68447_exterior_image_1_left#0.jpg", + "gt": "[[294.5, 97.5], [290.5, 120.5], [276.5, 137.5], [222.5, 148.5], [189.5, 174.5], [187.5, 152.5], [196.5, 136.5], [190.5, 125.5], [229.5, 128.5], [241.5, 116.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "47509_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the gray object inside the pot', and now you are performing subtask 'put the gray object inside the pot.'\nWith start location [0.46, 0.0], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.0], [0.69, 0.24], [0.73, 0.44], [0.7, 0.54], [0.74, 0.25], [0.77, 0.21], [0.78, 0.26], [0.75, 0.37], [0.73, 0.39], [0.74, 0.3]]}" + } + ], + "images": "47509_exterior_image_1_left#0", + "gt": "[[147, 0], [220, 44], [232, 79], [224, 97], [236, 45], [245, 38], [249, 46], [239, 67], [232, 71], [238, 54]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "25615_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the cloth off the stand and put it on the table', and now you are performing subtask 'take the cloth off the stand and put it on the table.'\nFrom start point [0.54, 0.01], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.01], [0.52, 0.14], [0.51, 0.06], [0.5, 0.06], [0.5, 0.07], [0.5, 0.23], [0.52, 0.26], [0.51, 0.19], [0.49, 0.12], [0.49, 0.06]]}" + } + ], + "images": "25615_exterior_image_2_left#0", + "gt": "[[174, 2], [167, 25], [163, 11], [160, 10], [159, 13], [161, 42], [165, 46], [163, 35], [157, 21], [157, 11]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0078_user_0007_scene_0010_cfg_0005_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the microwave door', and now you are performing subtask 'close the microwave door.'\nYou start from [0.48, 0.14]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.14], [0.52, 0.16], [0.54, 0.25], [0.59, 0.29], [0.64, 0.46], [0.6, 0.43], [0.53, 0.26], [0.43, 0.19], [0.43, 0.11], [0.41, 0.04]]}" + } + ], + "images": "RH20T_cfg5_task_0078_user_0007_scene_0010_cfg_0005_104122061850#0", + "gt": "[[305, 50], [331, 56], [344, 89], [378, 103], [412, 165], [387, 154], [339, 92], [277, 68], [276, 41], [263, 14]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0004_user_0005_scene_0003_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the small block on the left and move it to the right', and now you are performing subtask 'pick up the small block on the left and move it to the right.'\nFrom start point [0.63, 0.38], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.38], [0.64, 0.46], [0.59, 0.57], [0.62, 0.59], [0.62, 0.58], [0.63, 0.64], [0.6, 0.66], [0.59, 0.56], [0.69, 0.41], [0.68, 0.45]]}" + } + ], + "images": "RH20T_cfg1_task_0004_user_0005_scene_0003_cfg_0001_038522062547#0", + "gt": "[[404, 135], [409, 164], [379, 207], [396, 212], [399, 208], [404, 229], [381, 237], [377, 202], [444, 149], [434, 162]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122063633#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the block on the other block, pick up the block on the surface, transfer the block from the surface to the other block, place the block on the other block.'\nWith start location [0.29, 0.45], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.29, 0.45], [0.28, 0.47], [0.3, 0.46], [0.29, 0.46], [0.3, 0.48], [0.31, 0.44], [0.35, 0.44], [0.37, 0.48], [0.35, 0.41], [0.32, 0.38]]}" + } + ], + "images": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122063633#997", + "gt": "[[184, 162], [182, 171], [194, 166], [185, 164], [191, 174], [201, 157], [225, 158], [238, 173], [226, 146], [205, 135]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0058_user_0015_scene_0003_cfg_0001_819112071037#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn off the power strips', and now you are performing subtask 'press the third button on the power socket.'\nYou begin at [0.13, 0.42]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.13, 0.42], [0.16, 0.43], [0.16, 0.46], [0.15, 0.46], [0.15, 0.47], [0.15, 0.46], [0.15, 0.41], [0.15, 0.4], [0.16, 0.43], [0.16, 0.43]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0058_user_0015_scene_0003_cfg_0001_819112071037#229.jpg", + "gt": "[[82, 151], [103, 155], [103, 166], [97, 167], [97, 171], [95, 165], [96, 149], [98, 145], [100, 155], [100, 155]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_036422060215#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the first block on top of the second block, pick up the third block on the table, transfer the third block from the table to the position above the first block, place the third block on top of the first block.'\nWith start location [0.47, 0.37], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.37], [0.47, 0.35], [0.44, 0.32], [0.44, 0.33], [0.43, 0.35], [0.43, 0.37], [0.45, 0.26], [0.46, 0.24], [0.47, 0.27], [0.49, 0.31]]}" + } + ], + "images": "RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_036422060215#388", + "gt": "[[303, 134], [302, 126], [284, 115], [279, 118], [275, 125], [275, 134], [287, 95], [297, 88], [300, 98], [312, 113]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "28128_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the canned drink and put it in the sink', and now you are performing subtask 'pick up the canned drink and put it in the sink.'\nYou start from [0.29, 0.55]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.29, 0.55], [0.32, 0.59], [0.31, 0.71], [0.34, 0.64], [0.39, 0.63], [0.43, 0.66], [0.48, 0.74], [0.47, 0.84], [0.47, 0.76], [0.37, 0.76]]}" + } + ], + "images": "28128_exterior_image_1_left#0", + "gt": "[[94, 99], [101, 106], [100, 128], [108, 116], [124, 114], [139, 119], [154, 133], [152, 152], [150, 137], [118, 136]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "77695_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the watermelon plush toy from the table and move it to the box', and now you are performing subtask 'pick up the watermelon plush toy from the table and move it to the box.'\nStarting at [0.42, 0.38], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.38], [0.51, 0.34], [0.69, 0.5], [0.65, 0.56], [0.73, 0.63], [0.7, 0.58], [0.73, 0.72], [0.71, 0.76], [0.68, 0.57], [0.56, 0.37]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/77695_exterior_image_2_left#0.jpg", + "gt": "[[136, 68], [164, 62], [222, 90], [207, 100], [234, 114], [223, 105], [235, 129], [228, 137], [216, 103], [178, 67]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0001_user_0001_scene_0002_cfg_0001_819112070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the button from top to bottom', and now you are performing subtask 'press the button from top to bottom.'\nYou begin at [0.61, 0.3]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.61, 0.3], [0.6, 0.32], [0.6, 0.33], [0.58, 0.32], [0.6, 0.34], [0.59, 0.34], [0.59, 0.39], [0.58, 0.4], [0.59, 0.41], [0.61, 0.38]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0001_user_0001_scene_0002_cfg_0001_819112070044#0.jpg", + "gt": "[[389, 108], [381, 116], [384, 118], [373, 114], [382, 124], [378, 124], [378, 140], [374, 143], [377, 147], [391, 135]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0029_user_0014_scene_0003_cfg_0006_104122060811#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp the handle and open the drawer', and now you are performing subtask 'grasp the handle and open the drawer.'\nThe trajectory starts at [0.6, 0.35]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.35], [0.54, 0.36], [0.5, 0.35], [0.51, 0.36], [0.54, 0.36], [0.56, 0.38], [0.6, 0.38], [0.61, 0.38], [0.63, 0.36], [0.63, 0.36]]}" + } + ], + "images": "RH20T_cfg6_task_0029_user_0014_scene_0003_cfg_0006_104122060811#0", + "gt": "[[385, 127], [348, 130], [323, 126], [324, 130], [347, 130], [360, 135], [386, 135], [388, 138], [401, 129], [401, 129]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "56466_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the red lid on the lunchbox, move the cup to the left, and then move the cup to the right', and now you are performing subtask 'place the mug in the middle of the table.'\nThe trajectory starts at [0.64, 0.54]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.64, 0.54], [0.68, 0.49], [0.73, 0.48], [0.75, 0.51], [0.77, 0.63], [0.87, 0.65], [0.85, 0.59], [0.81, 0.55], [0.77, 0.54], [0.73, 0.57]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/56466_exterior_image_1_left#311.jpg", + "gt": "[[205, 97], [219, 88], [233, 86], [239, 91], [246, 113], [279, 117], [271, 106], [258, 99], [245, 97], [233, 102]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0034_user_0007_scene_0006_cfg_0005_104122063678#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a line', and now you are performing subtask 'place the second block beside the first block, pick up the third block from the table, transfer the third block from the table to beside the first block, place the third block beside the first block.'\nWith start location [0.53, 0.58], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.58], [0.54, 0.55], [0.63, 0.68], [0.66, 0.79], [0.66, 0.81], [0.59, 0.72], [0.56, 0.73], [0.56, 0.72], [0.57, 0.69], [0.66, 0.67]]}" + } + ], + "images": "RH20T_cfg5_task_0034_user_0007_scene_0006_cfg_0005_104122063678#597", + "gt": "[[341, 209], [348, 198], [406, 244], [424, 283], [423, 293], [379, 259], [357, 261], [356, 258], [366, 248], [424, 240]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0094_user_0014_scene_0003_cfg_0004_038522062288#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'push the lid on the bottom of the box.'\nYour initial coordinate is [0.27, 0.38]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.27, 0.38], [0.22, 0.38], [0.26, 0.41], [0.32, 0.26], [0.33, 0.28], [0.34, 0.24], [0.35, 0.28], [0.31, 0.22], [0.28, 0.26], [0.29, 0.33]]}" + } + ], + "images": "RH20T_cfg4_task_0094_user_0014_scene_0003_cfg_0004_038522062288#373", + "gt": "[[172, 138], [142, 135], [169, 147], [202, 92], [213, 101], [220, 87], [221, 99], [199, 79], [180, 94], [185, 118]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "81550_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the mug from the coffee maker and put it on the counter, and pour the content from the transparent cup into the mug', and now you are performing subtask 'place the cup on the countertop.'\nYou begin at [0.53, 0.6]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.6], [0.54, 0.68], [0.52, 0.66], [0.52, 0.71], [0.54, 0.68], [0.51, 0.75], [0.55, 0.67], [0.56, 0.75], [0.58, 0.68], [0.58, 0.68]]}" + } + ], + "images": "81550_exterior_image_1_left#405", + "gt": "[[170, 108], [174, 123], [166, 118], [168, 128], [172, 123], [162, 135], [175, 121], [178, 135], [185, 123], [185, 123]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0073_user_0007_scene_0001_cfg_0005_105422061350#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the things in the drawer', and now you are performing subtask 'grab the handle to pull the drawer open.'\nYour initial coordinate is [0.46, 0.08]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.08], [0.46, 0.08], [0.44, 0.11], [0.45, 0.14], [0.44, 0.21], [0.43, 0.19], [0.42, 0.19], [0.39, 0.21], [0.37, 0.22], [0.37, 0.24]]}" + } + ], + "images": "RH20T_cfg5_task_0073_user_0007_scene_0001_cfg_0005_105422061350#400", + "gt": "[[293, 29], [292, 27], [284, 40], [286, 52], [281, 77], [275, 70], [266, 69], [251, 77], [239, 80], [234, 85]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "39560_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the topmost drawer, take out the towel and place it on the counter then close the drawer', and now you are performing subtask 'pull the handle of the top drawer to the outside of the countertop.'\nYour starting position is [0.08, 0.45]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.08, 0.45], [0.24, 0.54], [0.31, 0.77], [0.35, 0.78], [0.33, 0.79], [0.37, 0.85], [0.45, 0.85], [0.45, 0.83], [0.27, 0.84], [0.22, 0.81]]}" + } + ], + "images": "39560_exterior_image_1_left#0", + "gt": "[[27.0, 80.5], [77.0, 96.5], [100.0, 138.5], [111.0, 140.5], [106.0, 141.5], [118.0, 153.5], [144.0, 153.5], [144.0, 149.5], [86.0, 151.5], [72.0, 145.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "58558_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the clothes', and now you are performing subtask 'fold the garment again on the table.'\nYou begin at [0.69, 0.55]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.55], [0.65, 0.6], [0.78, 0.56], [0.78, 0.76], [0.87, 0.87], [0.88, 0.99], [0.86, 0.84], [0.79, 0.68], [0.69, 0.58], [0.53, 0.6]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/58558_exterior_image_2_left#223.jpg", + "gt": "[[220, 99], [207, 108], [249, 100], [250, 137], [277, 157], [280, 178], [276, 151], [253, 122], [222, 104], [171, 108]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0031_user_0020_scene_0007_cfg_0007_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'cover the box', and now you are performing subtask 'cover the box.'\nYour starting position is [0.53, 0.32]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.32], [0.54, 0.34], [0.56, 0.47], [0.58, 0.51], [0.58, 0.62], [0.6, 0.48], [0.59, 0.39], [0.59, 0.32], [0.59, 0.33], [0.57, 0.28]]}" + } + ], + "images": "RH20T_cfg7_task_0031_user_0020_scene_0007_cfg_0007_104122064161#0", + "gt": "[[337, 115], [344, 122], [360, 168], [373, 184], [372, 225], [383, 172], [375, 140], [376, 116], [380, 119], [364, 100]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "67833_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the cylindrical object inside the box', and now you are performing subtask 'place the cylindrical object into the box.'\nWith start location [0.55, 0.43], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.43], [0.56, 0.48], [0.55, 0.53], [0.54, 0.51], [0.54, 0.54], [0.53, 0.44], [0.51, 0.4], [0.48, 0.41], [0.45, 0.46], [0.45, 0.37]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/67833_exterior_image_2_left#138.jpg", + "gt": "[[177, 77], [179, 87], [176, 95], [172, 92], [174, 97], [171, 79], [163, 72], [154, 74], [143, 83], [144, 66]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "65809_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the lid on the pot', and now you are performing subtask 'place the lid on the pot.'\nYou begin at [0.86, 0.04]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.86, 0.04], [0.64, 0.06], [0.58, 0.12], [0.53, 0.28], [0.51, 0.52], [0.47, 0.54], [0.52, 0.49], [0.57, 0.64], [0.47, 0.72], [0.48, 0.66]]}" + } + ], + "images": "65809_exterior_image_2_left#0", + "gt": "[[274, 8], [206, 10], [187, 21], [171, 50], [162, 93], [151, 97], [168, 89], [181, 116], [152, 129], [155, 119]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0066_user_0016_scene_0003_cfg_0004_104422070011#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'building blocks', and now you are performing subtask 'pick up the green block from the table, transfer the green block from the table to the block board, place the green block on the block board.'\nStarting at [0.53, 0.28], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.28], [0.56, 0.37], [0.52, 0.42], [0.48, 0.57], [0.43, 0.68], [0.45, 0.73], [0.42, 0.72], [0.4, 0.59], [0.37, 0.57], [0.39, 0.6]]}" + } + ], + "images": "RH20T_cfg4_task_0066_user_0016_scene_0003_cfg_0004_104422070011#0", + "gt": "[[336, 102], [360, 134], [330, 151], [310, 207], [278, 245], [285, 264], [270, 260], [255, 212], [234, 204], [249, 216]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "71404_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the cloth from the stand and put it on the table', and now you are performing subtask 'take the cloth from the stand and put it on the table.'\nYour starting position is [0.32, 0.0]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.0], [0.48, 0.14], [0.5, 0.2], [0.48, 0.27], [0.47, 0.11], [0.5, 0.14], [0.54, 0.27], [0.53, 0.46], [0.5, 0.35], [0.5, 0.35]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/71404_exterior_image_2_left#0.jpg", + "gt": "[[102, 0], [153, 26], [159, 36], [153, 49], [149, 19], [161, 25], [172, 48], [171, 83], [161, 63], [161, 63]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "40415_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the banana from the pot', and now you are performing subtask 'remove the banana from the pot.'\nBegin at [0.77, 0.02] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.77, 0.02], [0.52, 0.39], [0.48, 0.48], [0.47, 0.59], [0.45, 0.61], [0.44, 0.58], [0.42, 0.64], [0.43, 0.69], [0.42, 0.64], [0.41, 0.64]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/40415_exterior_image_2_left#0.jpg", + "gt": "[[246, 3], [167, 70], [155, 87], [149, 106], [145, 109], [140, 105], [133, 115], [137, 124], [135, 115], [131, 115]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg3_task_0057_user_0010_scene_0002_cfg_0003_104122062823#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn on the power strips', and now you are performing subtask 'press the third button on the power strip, press the fourth button on the power strip.'\nYour starting position is [0.45, 0.29]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.29], [0.45, 0.28], [0.45, 0.27], [0.44, 0.26], [0.44, 0.27], [0.44, 0.28], [0.44, 0.28], [0.44, 0.27], [0.45, 0.27], [0.46, 0.26]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg3_task_0057_user_0010_scene_0002_cfg_0003_104122062823#255.jpg", + "gt": "[[290, 106], [287, 99], [287, 96], [282, 95], [280, 97], [280, 99], [281, 99], [281, 98], [285, 97], [292, 93]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0039_user_0014_scene_0003_cfg_0006_104122061602#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pour the water from one cup into another empty cup', and now you are performing subtask 'pour the water from one cup into another empty cup.'\nWith start location [0.7, 0.29], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.29], [0.65, 0.28], [0.63, 0.34], [0.66, 0.29], [0.69, 0.33], [0.67, 0.34], [0.71, 0.31], [0.77, 0.39], [0.72, 0.26], [0.64, 0.31]]}" + } + ], + "images": "RH20T_cfg6_task_0039_user_0014_scene_0003_cfg_0006_104122061602#0", + "gt": "[[446, 106], [413, 100], [403, 122], [423, 103], [444, 119], [427, 122], [457, 113], [492, 142], [459, 94], [408, 113]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "13970_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the silver object on the sink drain', and now you are performing subtask 'put the silver object on the sink drain.'\nStarting at [0.28, 0.55], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.28, 0.55], [0.78, 0.71], [0.85, 0.77], [0.84, 0.84], [0.84, 0.77], [0.76, 0.71], [0.72, 0.74], [0.75, 0.78], [0.69, 0.74], [0.69, 0.74]]}" + } + ], + "images": "13970_exterior_image_1_left#0", + "gt": "[[89.0, 99.0], [249.0, 128.0], [272.0, 139.0], [268.0, 151.0], [269.0, 139.0], [244.0, 128.0], [231.0, 133.0], [241.0, 140.0], [220.0, 134.0], [220.0, 134.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "90036_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the white cloth from the glass panel onto the table', and now you are performing subtask 'place the white cloth from the glass panel onto the table.'\nFrom start point [0.61, 0.17], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.61, 0.17], [0.57, 0.19], [0.47, 0.22], [0.46, 0.28], [0.42, 0.29], [0.44, 0.33], [0.56, 0.44], [0.57, 0.51], [0.57, 0.49], [0.59, 0.49]]}" + } + ], + "images": "90036_exterior_image_1_left#0", + "gt": "[[196, 30], [182, 35], [151, 39], [148, 51], [134, 52], [141, 60], [178, 79], [181, 91], [182, 88], [189, 89]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "64414_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker inside the white bowl then remove it and put it on the table', and now you are performing subtask 'pick up the marker from the white bowl.'\nWith start location [0.86, 0.09], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.86, 0.09], [0.76, 0.1], [0.68, 0.13], [0.5, 0.38], [0.48, 0.42], [0.49, 0.45], [0.51, 0.32], [0.54, 0.21], [0.57, 0.17], [0.57, 0.26]]}" + } + ], + "images": "64414_exterior_image_2_left#0", + "gt": "[[275, 16], [242, 18], [216, 24], [160, 69], [155, 76], [157, 81], [164, 57], [174, 38], [183, 30], [182, 47]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0003_user_0014_scene_0010_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the orange button on the vertical base.'\nThe trajectory starts at [0.5, 0.37]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.52, 0.38], [0.62, 0.38], [0.65, 0.36], [0.65, 0.34], [0.59, 0.29], [0.65, 0.4], [0.66, 0.45], [0.66, 0.62], [0.65, 0.68]]}" + } + ], + "images": "RH20T_cfg5_task_0003_user_0014_scene_0010_cfg_0005_104122063678#0", + "gt": "[[323, 132], [331, 136], [400, 136], [417, 129], [417, 123], [376, 103], [418, 145], [424, 161], [424, 223], [417, 246]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0078_user_0014_scene_0008_cfg_0006_104122061330#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the microwave door', and now you are performing subtask 'close the microwave door.'\nBegin at [0.42, 0.27] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.27], [0.46, 0.22], [0.5, 0.26], [0.48, 0.36], [0.42, 0.44], [0.27, 0.59], [0.24, 0.67], [0.3, 0.6], [0.33, 0.53], [0.32, 0.5]]}" + } + ], + "images": "RH20T_cfg6_task_0078_user_0014_scene_0008_cfg_0006_104122061330#0", + "gt": "[[266, 97], [294, 79], [317, 92], [305, 130], [270, 159], [171, 211], [153, 242], [190, 216], [211, 192], [203, 180]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "45684_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the pen from inside the mug and place it on the desk', and now you are performing subtask 'remove the pen from inside the mug and place it on the desk.'\nStart at [0.47, 0.0]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.0], [0.44, 0.26], [0.46, 0.36], [0.43, 0.51], [0.36, 0.33], [0.34, 0.61], [0.32, 0.66], [0.3, 0.61], [0.38, 0.39], [0.4, 0.39]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/45684_exterior_image_2_left#0.jpg", + "gt": "[[150, 0], [141, 47], [146, 65], [137, 91], [116, 60], [108, 110], [101, 118], [96, 110], [121, 71], [128, 71]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "18069_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the carrot plush toy and put it on the counter', and now you are performing subtask 'pick up the carrot plush toy from the plate.'\nWith start location [0.38, 0.49], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.49], [0.38, 0.48], [0.37, 0.46], [0.37, 0.44], [0.38, 0.46], [0.35, 0.43], [0.35, 0.43], [0.36, 0.43], [0.36, 0.44], [0.38, 0.44]]}" + } + ], + "images": "18069_exterior_image_1_left#141", + "gt": "[[123, 89], [122, 86], [119, 83], [119, 80], [123, 83], [113, 78], [113, 77], [114, 77], [116, 79], [120, 80]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_104422070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the block on the plane, transfer the block from the plane to the other blocks, place the block on the other blocks.'\nYour starting position is [0.58, 0.37]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.37], [0.63, 0.36], [0.62, 0.38], [0.66, 0.42], [0.66, 0.51], [0.66, 0.54], [0.65, 0.5], [0.6, 0.46], [0.57, 0.47], [0.57, 0.45]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_104422070044#0.jpg", + "gt": "[[374, 132], [401, 130], [400, 137], [420, 150], [424, 182], [420, 193], [413, 179], [383, 166], [366, 169], [367, 161]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "37843_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the blackboard eraser to wipe the whiteboard and place it on the table', and now you are performing subtask 'use the blackboard eraser to wipe the whiteboard and place it on the table.'\nYou start from [0.71, 0.06]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.71, 0.06], [0.61, 0.24], [0.62, 0.18], [0.57, 0.11], [0.45, 0.07], [0.37, 0.11], [0.47, 0.1], [0.5, 0.17], [0.56, 0.75], [0.59, 0.38]]}" + } + ], + "images": "37843_exterior_image_2_left#0", + "gt": "[[227, 11], [195, 44], [200, 33], [181, 19], [144, 12], [118, 19], [151, 18], [161, 31], [178, 135], [190, 68]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "37914_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the door of the microwave', and now you are performing subtask 'close the door of the microwave.'\nWith start location [0.65, 0.52], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.52], [0.63, 0.47], [0.62, 0.36], [0.61, 0.31], [0.6, 0.31], [0.56, 0.32], [0.53, 0.35], [0.55, 0.38], [0.62, 0.39], [0.61, 0.38]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/37914_exterior_image_1_left#0.jpg", + "gt": "[[207, 93], [203, 85], [200, 64], [195, 56], [191, 55], [180, 58], [170, 63], [176, 69], [200, 71], [194, 68]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0032_user_0016_scene_0006_cfg_0004_104422070011#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'slide the outer casing onto the gift box', and now you are performing subtask 'push the gift box lid on the desktop.'\nBegin at [0.51, 0.39] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.39], [0.53, 0.43], [0.6, 0.33], [0.59, 0.23], [0.56, 0.26], [0.6, 0.37], [0.52, 0.36], [0.51, 0.44], [0.49, 0.41], [0.48, 0.44]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0032_user_0016_scene_0006_cfg_0004_104422070011#138.jpg", + "gt": "[[329, 142], [338, 154], [382, 120], [376, 83], [358, 92], [381, 134], [334, 130], [327, 160], [314, 148], [310, 159]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "30710_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the white cloth from the glass panel', and now you are performing subtask 'remove the white cloth from the glass panel.'\nBegin at [0.17, 0.29] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.17, 0.29], [0.27, 0.27], [0.38, 0.12], [0.39, 0.23], [0.32, 0.26], [0.28, 0.32], [0.3, 0.43], [0.31, 0.46], [0.33, 0.47], [0.31, 0.47]]}" + } + ], + "images": "30710_exterior_image_2_left#0", + "gt": "[[55, 53], [85, 48], [122, 22], [125, 42], [102, 47], [91, 57], [95, 77], [99, 82], [106, 84], [98, 84]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0081_user_0010_scene_0001_cfg_0001_039422060546#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the leg of the glasses (with one hand)', and now you are performing subtask 'pull the leg of the glasses outward.'\nYou start from [0.42, 0.29]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.29], [0.4, 0.3], [0.42, 0.3], [0.45, 0.28], [0.47, 0.28], [0.47, 0.27], [0.47, 0.28], [0.48, 0.26], [0.51, 0.24], [0.51, 0.24]]}" + } + ], + "images": "RH20T_cfg1_task_0081_user_0010_scene_0001_cfg_0001_039422060546#169", + "gt": "[[271, 104], [255, 107], [267, 108], [290, 99], [302, 101], [300, 98], [302, 100], [310, 93], [328, 86], [328, 86]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "87466_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the packet from the sink and place it on the left side of the counter', and now you are performing subtask 'remove the packet from the sink and place it on the left side of the counter.'\nYour initial coordinate is [0.69, 0.56]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.56], [0.71, 0.61], [0.69, 0.8], [0.7, 0.86], [0.66, 0.69], [0.62, 0.71], [0.6, 0.74], [0.6, 0.78], [0.6, 0.66], [0.64, 0.62]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/87466_exterior_image_2_left#0.jpg", + "gt": "[[220, 101], [226, 110], [220, 144], [223, 154], [212, 125], [197, 128], [192, 133], [192, 140], [191, 119], [205, 111]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0034_user_0007_scene_0005_cfg_0005_036422060215#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares in a row', and now you are performing subtask 'pick up the third square from the table.'\nWith start location [0.4, 0.43], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.43], [0.42, 0.41], [0.52, 0.43], [0.56, 0.46], [0.55, 0.47], [0.54, 0.48], [0.54, 0.49], [0.53, 0.47], [0.48, 0.41], [0.48, 0.43]]}" + } + ], + "images": "RH20T_cfg5_task_0034_user_0007_scene_0005_cfg_0005_036422060215#608", + "gt": "[[259, 154], [271, 148], [331, 155], [357, 165], [353, 171], [346, 172], [346, 177], [342, 168], [306, 146], [306, 156]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "52357_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up a vial and put it in the top compartment of the tray', and now you are performing subtask 'pick up a vial and put it in the top compartment of the tray.'\nYou start from [0.83, 0.0]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.83, 0.0], [0.82, 0.24], [0.77, 0.35], [0.76, 0.26], [0.68, 0.23], [0.61, 0.28], [0.64, 0.27], [0.64, 0.34], [0.58, 0.32], [0.7, 0.11]]}" + } + ], + "images": "52357_exterior_image_1_left#0", + "gt": "[[265, 0], [264, 43], [245, 63], [243, 46], [218, 41], [196, 50], [206, 48], [205, 61], [186, 57], [223, 20]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "43647_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'wipe the table with the yellow towel', and now you are performing subtask 'wipe the table with the yellow towel.'\nStart at [0.33, 0.17]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.17], [0.45, 0.31], [0.4, 0.46], [0.52, 0.44], [0.41, 0.53], [0.57, 0.51], [0.47, 0.44], [0.56, 0.57], [0.45, 0.46], [0.44, 0.24]]}" + } + ], + "images": "43647_exterior_image_2_left#0", + "gt": "[[104, 30], [143, 55], [129, 83], [166, 79], [130, 95], [182, 91], [149, 79], [180, 102], [143, 83], [141, 44]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0037_user_0014_scene_0002_cfg_0007_104122063633#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the first block on the second block, pick up the third block from the tabletop, transfer the third block from the tabletop to the first block, place the third block on the first block, pick up the fourth block from the tabletop.'\nYou begin at [0.33, 0.49]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.32, 0.49], [0.31, 0.52], [0.31, 0.56], [0.29, 0.5], [0.31, 0.43], [0.33, 0.41], [0.35, 0.45], [0.33, 0.42], [0.29, 0.43], [0.28, 0.45]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0037_user_0014_scene_0002_cfg_0007_104122063633#144.jpg", + "gt": "[[208, 178], [196, 187], [199, 201], [185, 179], [200, 153], [211, 149], [225, 163], [212, 150], [185, 153], [177, 162]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0037_user_0007_scene_0007_cfg_0002_104422070044#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the first block on the surface, pick up the second block from the surface, transfer the second block from the surface to the top of the first block, place the second block on top of the first block.'\nYou start from [0.51, 0.56]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.56], [0.51, 0.54], [0.53, 0.54], [0.54, 0.55], [0.54, 0.56], [0.54, 0.54], [0.51, 0.53], [0.51, 0.53], [0.51, 0.55], [0.51, 0.54]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg2_task_0037_user_0007_scene_0007_cfg_0002_104422070044#176.jpg", + "gt": "[[326, 201], [327, 195], [340, 195], [346, 197], [347, 203], [344, 196], [329, 190], [325, 192], [325, 197], [326, 194]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "15165_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push in the first rod on the foosball table', and now you are performing subtask 'push in the first rod on the foosball table.'\nStart at [0.58, 0.38]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.38], [0.62, 0.4], [0.62, 0.41], [0.64, 0.46], [0.65, 0.5], [0.67, 0.48], [0.73, 0.44], [0.7, 0.37], [0.67, 0.32], [0.67, 0.32]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/15165_exterior_image_1_left#0.jpg", + "gt": "[[186.0, 69.0], [199.0, 72.0], [197.0, 73.0], [206.0, 82.0], [208.0, 90.0], [213.0, 86.0], [232.0, 79.0], [224.0, 66.0], [213.0, 57.0], [213.0, 57.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_035622060973#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'place the third square beside the second square, pick up the fourth square on the table, pick up the fourth square from the right side of the table.'\nStart at [0.4, 0.45]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.4, 0.45], [0.4, 0.43], [0.39, 0.43], [0.38, 0.43], [0.35, 0.44], [0.33, 0.47], [0.32, 0.49], [0.31, 0.5], [0.32, 0.5], [0.32, 0.5]]}" + } + ], + "images": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_035622060973#487", + "gt": "[[256, 163], [253, 156], [250, 155], [244, 154], [226, 158], [210, 170], [203, 177], [201, 180], [202, 179], [202, 181]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "4672_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the yellow towel to wipe the table', and now you are performing subtask 'use the yellow towel to wipe the table.'\nYour initial coordinate is [0.5, 0.02]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.02], [0.47, 0.18], [0.49, 0.34], [0.39, 0.29], [0.33, 0.36], [0.42, 0.44], [0.48, 0.35], [0.36, 0.37], [0.25, 0.27], [0.29, 0.27]]}" + } + ], + "images": "4672_exterior_image_2_left#0", + "gt": "[[160, 4], [151, 33], [157, 61], [125, 52], [106, 65], [134, 79], [153, 63], [116, 66], [80, 49], [93, 49]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0078_user_0007_scene_0010_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the microwave door', and now you are performing subtask 'close the microwave door.'\nFrom start point [0.5, 0.37], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.7, 0.43], [0.77, 0.49], [0.76, 0.54], [0.72, 0.55], [0.61, 0.45], [0.48, 0.43], [0.5, 0.35], [0.48, 0.29], [0.48, 0.29]]}" + } + ], + "images": "RH20T_cfg5_task_0078_user_0007_scene_0010_cfg_0005_104122063678#0", + "gt": "[[323, 132], [450, 156], [491, 178], [488, 195], [463, 197], [392, 161], [310, 153], [319, 125], [305, 103], [305, 103]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "90987_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put a black pen on top of the table, then move another black pen and place it into the white cup, then, open the grey towel on the table', and now you are performing subtask 'place the pen into the white cup.'\nBegin at [0.67, 0.49] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.67, 0.49], [0.63, 0.48], [0.57, 0.52], [0.56, 0.56], [0.57, 0.59], [0.58, 0.47], [0.62, 0.32], [0.61, 0.24], [0.63, 0.36], [0.62, 0.38]]}" + } + ], + "images": "90987_exterior_image_1_left#112", + "gt": "[[213, 89], [203, 86], [183, 93], [180, 100], [183, 106], [185, 84], [197, 58], [195, 44], [201, 64], [198, 68]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "84743_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'folded shirt', and now you are performing subtask 'folded shirt.'\nThe trajectory starts at [0.65, 0.54]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.54], [0.57, 0.53], [0.5, 0.66], [0.62, 0.42], [0.45, 0.71], [0.57, 0.47], [0.62, 0.56], [0.56, 0.66], [0.59, 0.52], [0.59, 0.59]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/84743_exterior_image_2_left#0.jpg", + "gt": "[[208, 98], [182, 96], [161, 119], [200, 76], [143, 128], [183, 84], [197, 101], [178, 118], [188, 94], [190, 106]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0090_user_0007_scene_0001_cfg_0001_750612070851#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the wooden block', and now you are performing subtask 'press the wooden block.'\nFrom start point [0.52, 0.44], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.44], [0.49, 0.43], [0.44, 0.46], [0.38, 0.47], [0.37, 0.45], [0.36, 0.49], [0.31, 0.5], [0.28, 0.55], [0.3, 0.52], [0.38, 0.49]]}" + } + ], + "images": "RH20T_cfg1_task_0090_user_0007_scene_0001_cfg_0001_750612070851#0", + "gt": "[[333, 157], [316, 153], [282, 165], [244, 168], [237, 162], [231, 175], [199, 181], [182, 197], [192, 187], [244, 176]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122061018#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the block on the other block, pick up the block on the surface, transfer the block from the surface to the other block, place the block on the other block.'\nWith start location [0.59, 0.31], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.31], [0.56, 0.24], [0.51, 0.25], [0.43, 0.33], [0.58, 0.23], [0.6, 0.23], [0.6, 0.27], [0.57, 0.18], [0.52, 0.17], [0.52, 0.17]]}" + } + ], + "images": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122061018#1008", + "gt": "[[380, 110], [357, 87], [326, 90], [275, 120], [369, 83], [381, 82], [384, 98], [366, 66], [330, 62], [330, 62]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "11460_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the door', and now you are performing subtask 'pull the door handle to open the door.'\nYou start from [0.54, 0.56]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.56], [0.52, 0.56], [0.5, 0.59], [0.55, 0.72], [0.53, 0.68], [0.57, 0.7], [0.82, 0.47], [0.82, 0.44], [0.81, 0.43], [0.77, 0.44]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/11460_exterior_image_1_left#208.jpg", + "gt": "[[173, 101], [165, 101], [161, 106], [176, 130], [171, 122], [182, 126], [261, 85], [263, 80], [259, 78], [248, 79]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0061_user_0007_scene_0010_cfg_0005_104122063678#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'place the object on the table, pick up the object in the container, transfer the object from the container to the table, place the object on the table.'\nStart at [0.47, 0.39]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.39], [0.5, 0.46], [0.49, 0.57], [0.51, 0.45], [0.48, 0.31], [0.47, 0.3], [0.46, 0.32], [0.46, 0.38], [0.47, 0.35], [0.5, 0.41]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0061_user_0007_scene_0010_cfg_0005_104122063678#374.jpg", + "gt": "[[300, 140], [317, 165], [312, 204], [327, 161], [307, 111], [301, 107], [297, 115], [293, 137], [299, 125], [323, 147]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "9809_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the black basket backwards', and now you are performing subtask 'move the black basket backwards.'\nBegin at [0.64, 0.37] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.64, 0.37], [0.48, 0.46], [0.5, 0.61], [0.36, 0.97], [0.27, 0.69], [0.29, 0.56], [0.41, 0.58], [0.42, 0.68], [0.38, 0.7], [0.45, 0.64]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/9809_exterior_image_1_left#0.jpg", + "gt": "[[204, 67], [155, 82], [161, 110], [115, 174], [87, 125], [93, 100], [131, 105], [135, 122], [121, 126], [145, 116]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0013_user_0020_scene_0010_cfg_0007_104122061602#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the block on the scale', and now you are performing subtask 'place the big block on the scale, pick up the small block from the table, transfer the small block from the table to the big block, place the small block on the big block.'\nThe trajectory starts at [0.74, 0.47]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.74, 0.47], [0.77, 0.46], [0.78, 0.52], [0.78, 0.59], [0.76, 0.64], [0.79, 0.57], [0.79, 0.52], [0.75, 0.45], [0.79, 0.47], [0.79, 0.49]]}" + } + ], + "images": "RH20T_cfg7_task_0013_user_0020_scene_0010_cfg_0007_104122061602#248", + "gt": "[[473, 171], [492, 167], [501, 188], [502, 214], [489, 229], [503, 207], [504, 187], [482, 163], [507, 168], [504, 176]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0065_user_0015_scene_0012_cfg_0004_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with small Lego blocks', and now you are performing subtask 'pick up black block on the left side of the table, transfer black block from the left side of the table to the central base, place black block on the base, push black block to the right, push black block to the left, push black block to the left again, press black block down, press black block down again, press black block down, press black block down, press black block down, pick up red small block on the left side of the table.'\nThe trajectory starts at [0.45, 0.0]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.0], [0.39, 0.0], [0.4, 0.03], [0.35, 0.07], [0.37, 0.14], [0.37, 0.08], [0.43, 0.14], [0.5, 0.25], [0.4, 0.17], [0.4, 0.17]]}" + } + ], + "images": "RH20T_cfg4_task_0065_user_0015_scene_0012_cfg_0004_104122062823#0", + "gt": "[[291, 0], [249, 0], [259, 11], [225, 24], [238, 49], [234, 30], [277, 51], [317, 89], [259, 60], [259, 60]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "86272_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the white bowl and place the white bowl on the right side of the table', and now you are performing subtask 'pick up the white bowl and place the white bowl on the right side of the table.'\nYour initial coordinate is [0.72, 0.0]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.72, 0.0], [0.63, 0.19], [0.59, 0.38], [0.58, 0.48], [0.63, 0.33], [0.67, 0.29], [0.68, 0.34], [0.66, 0.44], [0.67, 0.36], [0.63, 0.21]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/86272_exterior_image_2_left#0.jpg", + "gt": "[[231, 0], [203, 34], [188, 68], [187, 87], [201, 59], [214, 53], [219, 61], [212, 80], [215, 64], [201, 38]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0061_user_0007_scene_0001_cfg_0005_104122061850#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take everything out of the gift box', and now you are performing subtask 'pick up the second piece from the gift box, transfer the second piece from the gift box to the table, place the second piece on the table.'\nStart at [0.5, 0.29]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.29], [0.51, 0.46], [0.5, 0.42], [0.47, 0.1], [0.55, 0.13], [0.59, 0.2], [0.55, 0.15], [0.5, 0.11], [0.47, 0.2], [0.47, 0.2]]}" + } + ], + "images": "RH20T_cfg5_task_0061_user_0007_scene_0001_cfg_0005_104122061850#320", + "gt": "[[323, 103], [326, 166], [319, 151], [303, 37], [349, 48], [377, 73], [353, 53], [320, 41], [299, 73], [299, 73]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "71515_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the yellow block on its side', and now you are performing subtask 'place the yellow block on its side.'\nWith start location [0.7, 0.01], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.01], [0.6, 0.14], [0.53, 0.17], [0.54, 0.19], [0.57, 0.16], [0.54, 0.2], [0.55, 0.28], [0.54, 0.18], [0.56, 0.13], [0.57, 0.13]]}" + } + ], + "images": "71515_exterior_image_1_left#0", + "gt": "[[225, 2], [193, 26], [170, 30], [172, 34], [183, 28], [174, 36], [177, 51], [174, 32], [179, 24], [184, 24]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "62596_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the brown object to the right', and now you are performing subtask 'move the brown object to the right.'\nStart at [0.0, 0.04]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.0, 0.04], [0.45, 0.47], [0.5, 0.59], [0.47, 0.76], [0.55, 0.48], [0.69, 0.36], [0.81, 0.64], [0.82, 0.82], [0.69, 0.57], [0.69, 0.57]]}" + } + ], + "images": "62596_exterior_image_1_left#0", + "gt": "[[1.0, 7.5], [144.0, 85.5], [159.0, 106.5], [149.0, 136.5], [175.0, 86.5], [221.0, 64.5], [258.0, 114.5], [261.0, 147.5], [222.0, 103.5], [222.0, 103.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0011_user_0013_scene_0001_cfg_0002_104422070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the watering can to the center of the table', and now you are performing subtask 'move the watering can to the center of the table.'\nThe trajectory starts at [0.49, 0.53]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.53], [0.44, 0.43], [0.39, 0.38], [0.4, 0.41], [0.47, 0.43], [0.5, 0.47], [0.49, 0.44], [0.45, 0.4], [0.41, 0.4], [0.41, 0.41]]}" + } + ], + "images": "RH20T_cfg2_task_0011_user_0013_scene_0001_cfg_0002_104422070044#0", + "gt": "[[312, 190], [281, 156], [251, 138], [258, 148], [298, 153], [321, 168], [316, 160], [290, 145], [262, 145], [263, 149]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_038522063145#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'putting a portion of the Lego blocks on the Lego baseplate', and now you are performing subtask 'pick up the purple Lego brick from the table, transfer the purple Lego brick from the table to the Lego base, insert the purple Lego brick into the Lego base.'\nStarting at [0.79, 0.4], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.79, 0.4], [0.84, 0.42], [0.82, 0.47], [0.82, 0.42], [0.81, 0.38], [0.77, 0.33], [0.73, 0.35], [0.7, 0.46], [0.72, 0.42], [0.75, 0.39]]}" + } + ], + "images": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_038522063145#148", + "gt": "[[506, 144], [537, 152], [524, 171], [526, 151], [520, 136], [493, 120], [467, 125], [448, 166], [461, 151], [482, 140]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0081_user_0007_scene_0003_cfg_0005_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the leg of the glasses (with one hand)', and now you are performing subtask 'unfold the leg of the glasses (with one hand).'\nYou begin at [0.5, 0.37]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.37], [0.49, 0.37], [0.48, 0.46], [0.53, 0.55], [0.54, 0.45], [0.58, 0.39], [0.54, 0.53], [0.53, 0.49], [0.55, 0.41], [0.58, 0.4]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0081_user_0007_scene_0003_cfg_0005_104122063678#0.jpg", + "gt": "[[323, 132], [314, 134], [308, 165], [340, 197], [346, 163], [372, 140], [343, 192], [337, 178], [352, 149], [369, 144]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0029_user_0003_scene_0005_cfg_0002_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp the handle and open the drawer', and now you are performing subtask 'grasp the handle and open the drawer.'\nStart at [0.53, 0.29]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.29], [0.53, 0.54], [0.52, 0.57], [0.53, 0.59], [0.52, 0.58], [0.52, 0.63], [0.5, 0.62], [0.51, 0.65], [0.57, 0.41], [0.56, 0.3]]}" + } + ], + "images": "RH20T_cfg2_task_0029_user_0003_scene_0005_cfg_0002_104122061850#0", + "gt": "[[340, 105], [340, 196], [331, 206], [339, 214], [332, 209], [331, 227], [320, 222], [325, 234], [363, 147], [361, 108]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0002_user_0013_scene_0006_cfg_0002_104122063678#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull out a napkin, place the napkin on the table', and now you are performing subtask 'pick up a napkin on the table, transfer the tissue from the tissue bag to above the table, place the tissue on the table.'\nFrom start point [0.7, 0.69], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.69], [0.7, 0.73], [0.7, 0.72], [0.71, 0.72], [0.72, 0.68], [0.71, 0.64], [0.64, 0.61], [0.62, 0.57], [0.57, 0.52], [0.53, 0.46]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg2_task_0002_user_0013_scene_0006_cfg_0002_104122063678#134.jpg", + "gt": "[[449, 250], [448, 261], [445, 259], [452, 258], [461, 245], [455, 230], [409, 219], [395, 206], [362, 187], [337, 167]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0064_user_0020_scene_0009_cfg_0007_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the roll of paper down from the shelf', and now you are performing subtask 'take the roll of paper down from the shelf.'\nYou begin at [0.53, 0.32]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.32], [0.55, 0.33], [0.56, 0.21], [0.59, 0.22], [0.6, 0.29], [0.55, 0.36], [0.55, 0.44], [0.55, 0.32], [0.54, 0.28], [0.52, 0.3]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0064_user_0020_scene_0009_cfg_0007_104122064161#0.jpg", + "gt": "[[337, 115], [350, 118], [360, 74], [377, 81], [385, 104], [353, 128], [355, 160], [351, 117], [343, 102], [336, 108]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "37401_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'folding the towel on the table', and now you are performing subtask 'pick up the bottom right corner of the towel on the table.'\nYour starting position is [0.65, 0.63]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.63], [0.69, 0.61], [0.7, 0.64], [0.69, 0.67], [0.61, 0.64], [0.5, 0.72], [0.51, 0.73], [0.52, 0.71], [0.56, 0.72], [0.64, 0.79]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/37401_exterior_image_1_left#183.jpg", + "gt": "[[209, 114], [220, 109], [223, 116], [221, 121], [195, 116], [161, 130], [164, 132], [168, 128], [180, 129], [205, 142]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "91717_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the orange cloth', and now you are performing subtask 'fold the orange cloth on the tabletop.'\nStart at [0.62, 0.54]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.54], [0.61, 0.54], [0.62, 0.62], [0.6, 0.64], [0.67, 0.63], [0.89, 0.68], [0.69, 0.47], [0.65, 0.42], [0.64, 0.39], [0.65, 0.4]]}" + } + ], + "images": "91717_exterior_image_2_left#106", + "gt": "[[198, 98], [194, 98], [198, 111], [191, 115], [215, 113], [285, 123], [220, 85], [208, 76], [204, 70], [209, 72]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0094_user_0014_scene_0003_cfg_0004_104422070011#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'push the lid on the left side of the box, push the lid on the top of the box, push the lid on the bottom of the box.'\nYou start from [0.42, 0.47]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.47], [0.42, 0.54], [0.46, 0.54], [0.51, 0.68], [0.51, 0.6], [0.45, 0.52], [0.48, 0.6], [0.49, 0.7], [0.5, 0.59], [0.46, 0.5]]}" + } + ], + "images": "RH20T_cfg4_task_0094_user_0014_scene_0003_cfg_0004_104422070011#237", + "gt": "[[271, 171], [271, 195], [295, 195], [327, 244], [324, 216], [288, 189], [309, 217], [313, 252], [318, 212], [292, 181]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0044_user_0014_scene_0005_cfg_0007_037522061512#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up a bag of things', and now you are performing subtask 'pick up a bag of things.'\nYour initial coordinate is [0.41, 0.24]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.24], [0.4, 0.26], [0.41, 0.23], [0.41, 0.22], [0.44, 0.24], [0.44, 0.26], [0.43, 0.25], [0.41, 0.2], [0.42, 0.16], [0.41, 0.12]]}" + } + ], + "images": "RH20T_cfg7_task_0044_user_0014_scene_0005_cfg_0007_037522061512#0", + "gt": "[[261, 86], [259, 93], [260, 83], [264, 81], [280, 85], [281, 92], [275, 89], [265, 72], [269, 58], [265, 45]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0014_user_0014_scene_0001_cfg_0001_750612070851#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the block from the scale', and now you are performing subtask 'remove the block from the scale.'\nYou start from [0.49, 0.45]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.45], [0.53, 0.49], [0.53, 0.47], [0.53, 0.49], [0.57, 0.47], [0.66, 0.52], [0.64, 0.61], [0.64, 0.58], [0.52, 0.52], [0.48, 0.53]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0014_user_0014_scene_0001_cfg_0001_750612070851#0.jpg", + "gt": "[[315, 161], [339, 177], [338, 171], [339, 176], [365, 170], [421, 186], [412, 220], [411, 208], [331, 188], [305, 192]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0081_user_0015_scene_0002_cfg_0004_104422070011#TrajectoryPrediction_Task#8", + "task": "TrajectoryPrediction_Task", + "p_index": 8, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the leg of the glasses (with one hand)', and now you are performing subtask 'pick up the glasses on the table.'\nStart at [0.44, 0.51]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.51], [0.44, 0.52], [0.45, 0.49], [0.45, 0.45], [0.45, 0.47], [0.45, 0.48], [0.44, 0.47], [0.45, 0.47], [0.44, 0.46], [0.45, 0.44]]}" + } + ], + "images": "RH20T_cfg4_task_0081_user_0015_scene_0002_cfg_0004_104422070011#285", + "gt": "[[283, 183], [280, 186], [286, 176], [290, 162], [290, 171], [286, 172], [283, 169], [285, 168], [284, 165], [287, 159]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_038522063145#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'fold the fourth lid on the box on the table.'\nStart at [0.74, 0.42]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.74, 0.43], [0.77, 0.37], [0.77, 0.41], [0.77, 0.37], [0.84, 0.34], [0.87, 0.38], [0.84, 0.41], [0.84, 0.31], [0.83, 0.28], [0.8, 0.31]]}" + } + ], + "images": "RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_038522063145#191", + "gt": "[[472, 153], [495, 133], [492, 146], [492, 134], [536, 121], [554, 135], [535, 146], [539, 110], [529, 102], [509, 110]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0108_user_0014_scene_0004_cfg_0001_039422060546#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'sharpen the pencil with a pencil sharpener', and now you are performing subtask 'sharpen the pencil with a pencil sharpener.'\nStart at [0.41, 0.34]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.34], [0.45, 0.29], [0.5, 0.26], [0.47, 0.22], [0.48, 0.33], [0.5, 0.38], [0.51, 0.38], [0.49, 0.28], [0.5, 0.22], [0.5, 0.17]]}" + } + ], + "images": "RH20T_cfg1_task_0108_user_0014_scene_0004_cfg_0001_039422060546#0", + "gt": "[[265, 124], [286, 104], [318, 92], [304, 80], [305, 120], [323, 137], [327, 135], [311, 102], [320, 78], [318, 63]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "23502_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the orange cuboid on the stack', and now you are performing subtask 'put the orange cuboid on the stack.'\nThe trajectory starts at [0.67, 0.01]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.67, 0.01], [0.71, 0.46], [0.77, 0.57], [0.75, 0.62], [0.74, 0.38], [0.66, 0.16], [0.59, 0.11], [0.56, 0.17], [0.57, 0.28], [0.58, 0.04]]}" + } + ], + "images": "23502_exterior_image_1_left#0", + "gt": "[[215.0, 2.0], [226.0, 83.0], [247.0, 103.0], [240.0, 112.0], [238.0, 68.0], [210.0, 29.0], [188.0, 20.0], [179.0, 31.0], [182.0, 50.0], [186.0, 8.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "53339_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the lid and close it on the jar', and now you are performing subtask 'pick up the lid and close it on the jar.'\nYour initial coordinate is [0.37, 0.61]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.37, 0.61], [0.54, 0.48], [0.73, 0.45], [0.94, 0.56], [0.98, 0.74], [0.88, 0.56], [0.74, 0.58], [0.72, 0.67], [0.75, 0.69], [0.7, 0.57]]}" + } + ], + "images": "53339_exterior_image_1_left#0", + "gt": "[[119, 110], [174, 86], [233, 81], [301, 101], [312, 133], [282, 101], [237, 105], [229, 120], [240, 124], [225, 102]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122063633#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'place the block on the block board, pick up the block on the flat surface, transfer the block from the flat surface to the block board, place the block on the block board, pick up the block on the flat surface.'\nStarting at [0.27, 0.42], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.27, 0.42], [0.24, 0.44], [0.23, 0.45], [0.26, 0.39], [0.31, 0.41], [0.35, 0.49], [0.35, 0.52], [0.31, 0.44], [0.3, 0.42], [0.28, 0.43]]}" + } + ], + "images": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122063633#220", + "gt": "[[172, 150], [152, 157], [148, 163], [167, 142], [196, 148], [224, 175], [222, 187], [200, 158], [189, 152], [180, 154]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "69043_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the green pen from the maroon mug', and now you are performing subtask 'remove the green pen from the maroon mug.'\nStart at [0.22, 0.0]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.22, 0.0], [0.38, 0.18], [0.43, 0.31], [0.47, 0.34], [0.51, 0.48], [0.48, 0.3], [0.43, 0.28], [0.41, 0.42], [0.43, 0.48], [0.44, 0.22]]}" + } + ], + "images": "69043_exterior_image_1_left#0", + "gt": "[[70, 0], [120, 32], [138, 55], [151, 62], [162, 87], [153, 54], [136, 50], [132, 75], [136, 86], [140, 39]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122061602#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the block on the other block, pick up the block on the surface, transfer the block from the surface to the other block, place the block on the other block.'\nYou begin at [0.65, 0.37]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.37], [0.68, 0.34], [0.73, 0.39], [0.74, 0.54], [0.71, 0.38], [0.66, 0.31], [0.64, 0.36], [0.68, 0.32], [0.72, 0.33], [0.72, 0.33]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122061602#1009.jpg", + "gt": "[[415, 132], [436, 123], [467, 141], [475, 195], [452, 137], [420, 113], [407, 129], [435, 114], [462, 120], [462, 120]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_038522063145#TrajectoryPrediction_Task#14", + "task": "TrajectoryPrediction_Task", + "p_index": 14, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with small Lego blocks for the nth time with n blocks', and now you are performing subtask 'pick up the red small Lego block from the desktop, transfer the red small Lego block from the desktop to the blue small Lego block, place the red small Lego block on the blue small Lego block.'\nStarting at [0.69, 0.37], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.37], [0.67, 0.32], [0.68, 0.35], [0.69, 0.33], [0.7, 0.34], [0.71, 0.42], [0.7, 0.43], [0.71, 0.43], [0.7, 0.46], [0.71, 0.42]]}" + } + ], + "images": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_038522063145#1478", + "gt": "[[442, 134], [431, 116], [434, 125], [440, 120], [445, 122], [454, 152], [448, 153], [452, 156], [447, 167], [452, 150]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "77489_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the carrot in the pot, move the banana to the left of the counter, then slide the spout to the right', and now you are performing subtask 'place the object in the bowl.'\nStarting at [0.36, 0.01], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.01], [0.36, 0.41], [0.37, 0.47], [0.38, 0.49], [0.36, 0.49], [0.37, 0.4], [0.39, 0.38], [0.4, 0.41], [0.4, 0.44], [0.38, 0.4]]}" + } + ], + "images": "77489_exterior_image_1_left#0", + "gt": "[[115, 2], [116, 74], [117, 84], [123, 88], [116, 89], [118, 72], [124, 68], [128, 74], [128, 79], [121, 72]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "57515_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove a tissue from the tissue box and put it on the table', and now you are performing subtask 'remove a tissue from the tissue box and put it on the table.'\nStarting at [0.5, 0.08], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.08], [0.54, 0.33], [0.52, 0.44], [0.54, 0.43], [0.6, 0.68], [0.61, 0.66], [0.5, 0.56], [0.5, 0.4], [0.66, 0.44], [0.63, 0.51]]}" + } + ], + "images": "57515_exterior_image_1_left#0", + "gt": "[[161.5, 14.5], [172.5, 59.5], [167.5, 79.5], [172.5, 76.5], [191.5, 122.5], [196.5, 119.5], [161.5, 101.5], [161.5, 71.5], [210.5, 78.5], [202.5, 91.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "25857_exterior_image_1_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the second drawer from the right then put the left straw in the drawer', and now you are performing subtask 'push the drawer.'\nStart at [0.41, 0.61]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.61], [0.35, 0.62], [0.33, 0.66], [0.33, 0.69], [0.35, 0.68], [0.34, 0.66], [0.38, 0.68], [0.42, 0.59], [0.45, 0.57], [0.34, 0.54]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/25857_exterior_image_1_left#359.jpg", + "gt": "[[131.5, 109.5], [113.5, 112.5], [105.5, 118.5], [106.5, 124.5], [112.5, 121.5], [109.5, 119.5], [122.5, 122.5], [135.5, 106.5], [145.5, 102.5], [109.5, 96.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "17711_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the white paper in the black bin then put the orange object in the open drawer then close the drawer', and now you are performing subtask 'push the drawer door.'\nFrom start point [0.55, 0.13], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.55, 0.13], [0.58, 0.14], [0.51, 0.14], [0.42, 0.23], [0.4, 0.31], [0.41, 0.36], [0.35, 0.39], [0.39, 0.52], [0.35, 0.48], [0.32, 0.49]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/17711_exterior_image_1_left#98.jpg", + "gt": "[[176.5, 23.5], [184.5, 25.5], [164.5, 24.5], [134.5, 41.5], [126.5, 55.5], [130.5, 65.5], [111.5, 69.5], [123.5, 94.5], [111.5, 86.5], [102.5, 88.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0038_user_0020_scene_0006_cfg_0007_104122061018#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the cup', and now you are performing subtask 'pick up the cup.'\nBegin at [0.57, 0.09] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.09], [0.54, 0.16], [0.52, 0.32], [0.48, 0.35], [0.47, 0.41], [0.45, 0.38], [0.46, 0.35], [0.45, 0.27], [0.48, 0.19], [0.48, 0.16]]}" + } + ], + "images": "RH20T_cfg7_task_0038_user_0020_scene_0006_cfg_0007_104122061018#0", + "gt": "[[363, 31], [343, 56], [330, 116], [308, 127], [302, 146], [288, 136], [294, 125], [289, 96], [307, 69], [306, 58]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122061330#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a sliding window', and now you are performing subtask 'slide the left window, slide the right window.'\nYour starting position is [0.51, 0.51]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.51], [0.48, 0.49], [0.45, 0.51], [0.45, 0.54], [0.48, 0.57], [0.45, 0.52], [0.47, 0.56], [0.45, 0.53], [0.45, 0.48], [0.45, 0.43]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122061330#393.jpg", + "gt": "[[327, 183], [308, 176], [288, 184], [289, 193], [305, 207], [291, 186], [299, 200], [291, 192], [286, 172], [289, 155]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0108_user_0015_scene_0009_cfg_0004_104122062823#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'sharpen the pencil with a pencil sharpener', and now you are performing subtask 'twist the handle of the pencil sharpener clockwise.'\nYour initial coordinate is [0.52, 0.31]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.31], [0.57, 0.41], [0.56, 0.44], [0.53, 0.44], [0.56, 0.43], [0.56, 0.46], [0.52, 0.46], [0.51, 0.37], [0.48, 0.28], [0.48, 0.28]]}" + } + ], + "images": "RH20T_cfg4_task_0108_user_0015_scene_0009_cfg_0004_104122062823#191", + "gt": "[[332, 113], [365, 147], [358, 157], [338, 157], [358, 154], [356, 167], [332, 166], [327, 133], [306, 99], [306, 99]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0029_user_0020_scene_0010_cfg_0007_104122061330#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grasp the handle and open the drawer', and now you are performing subtask 'grasp the handle and open the drawer.'\nThe trajectory starts at [0.42, 0.29]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.29], [0.47, 0.34], [0.47, 0.32], [0.53, 0.3], [0.58, 0.33], [0.59, 0.37], [0.52, 0.55], [0.49, 0.46], [0.49, 0.38], [0.47, 0.28]]}" + } + ], + "images": "RH20T_cfg7_task_0029_user_0020_scene_0010_cfg_0007_104122061330#0", + "gt": "[[266, 103], [301, 123], [303, 115], [337, 107], [369, 120], [380, 134], [336, 198], [313, 164], [314, 137], [298, 99]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0058_user_0010_scene_0006_cfg_0003_104122062823#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn off the power strips', and now you are performing subtask 'press the third button of the power strip.'\nBegin at [0.21, 0.13] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.21, 0.13], [0.21, 0.12], [0.21, 0.12], [0.21, 0.13], [0.21, 0.15], [0.21, 0.13], [0.21, 0.12], [0.21, 0.12], [0.21, 0.13], [0.2, 0.14]]}" + } + ], + "images": "RH20T_cfg3_task_0058_user_0010_scene_0006_cfg_0003_104122062823#193", + "gt": "[[136, 46], [134, 44], [134, 43], [136, 46], [136, 53], [135, 48], [135, 45], [136, 45], [132, 46], [129, 50]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0111_user_0010_scene_0006_cfg_0001_819112070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the object with the corresponding shape into the corresponding hole', and now you are performing subtask 'put the object with the corresponding shape into the corresponding hole.'\nYou begin at [0.62, 0.47]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.47], [0.61, 0.49], [0.59, 0.47], [0.56, 0.53], [0.52, 0.49], [0.5, 0.49], [0.47, 0.54], [0.49, 0.62], [0.54, 0.51], [0.62, 0.49]]}" + } + ], + "images": "RH20T_cfg1_task_0111_user_0010_scene_0006_cfg_0001_819112070044#0", + "gt": "[[400, 168], [391, 177], [378, 171], [357, 191], [335, 175], [321, 177], [304, 194], [314, 224], [347, 183], [397, 176]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "61794_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the mug with the marker to the left', and now you are performing subtask 'transfer the screwdriver from the cup to the tabletop.'\nYou begin at [0.43, 0.36]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.36], [0.46, 0.36], [0.47, 0.37], [0.48, 0.39], [0.5, 0.46], [0.52, 0.49], [0.52, 0.52], [0.51, 0.44], [0.51, 0.41], [0.51, 0.41]]}" + } + ], + "images": "61794_exterior_image_1_left#0", + "gt": "[[139, 65], [148, 65], [151, 67], [154, 70], [159, 83], [168, 89], [168, 93], [163, 80], [162, 73], [162, 73]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0042_user_0016_scene_0007_cfg_0001_819112070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'screw the lid onto the jar (needs to be rotated)', and now you are performing subtask 'screw the lid onto the jar (needs to be rotated).'\nYour initial coordinate is [0.62, 0.47]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.47], [0.53, 0.45], [0.43, 0.51], [0.42, 0.55], [0.4, 0.5], [0.4, 0.52], [0.4, 0.49], [0.44, 0.48], [0.49, 0.57], [0.49, 0.57]]}" + } + ], + "images": "RH20T_cfg1_task_0042_user_0016_scene_0007_cfg_0001_819112070044#0", + "gt": "[[400, 168], [340, 162], [275, 184], [269, 199], [253, 180], [258, 188], [259, 178], [280, 173], [313, 205], [313, 205]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "42174_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the right white towel to wipe the pan', and now you are performing subtask 'use the right white towel to wipe the pan.'\nYou begin at [0.39, 0.15]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.15], [0.62, 0.48], [0.66, 0.82], [0.68, 0.52], [0.59, 0.5], [0.63, 0.7], [0.58, 0.61], [0.66, 0.49], [0.65, 0.79], [0.65, 0.63]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/42174_exterior_image_1_left#0.jpg", + "gt": "[[124.0, 26.5], [200.0, 86.5], [211.0, 147.5], [216.0, 93.5], [188.0, 90.5], [203.0, 125.5], [185.0, 110.5], [211.0, 88.5], [209.0, 142.5], [209.0, 113.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "80082_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker in the mug', and now you are performing subtask 'put the marker in the mug.'\nBegin at [0.83, 0.0] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.83, 0.0], [0.6, 0.5], [0.59, 0.58], [0.63, 0.59], [0.62, 0.53], [0.69, 0.33], [0.74, 0.28], [0.76, 0.35], [0.7, 0.38], [0.69, 0.35]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/80082_exterior_image_1_left#0.jpg", + "gt": "[[267.0, 0.5], [192.0, 90.5], [190.0, 104.5], [203.0, 105.5], [198.0, 95.5], [220.0, 59.5], [237.0, 50.5], [242.0, 63.5], [225.0, 68.5], [222.0, 62.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0036_user_0014_scene_0001_cfg_0007_104122061330#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take a test tube from the rightmost side of the test tube rack, shake it in the air, and then place it back on the test tube rack', and now you are performing subtask 'take a test tube from the rightmost side of the test tube rack, shake it in the air, and then place it back on the test tube rack.'\nStart at [0.42, 0.29]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.29], [0.45, 0.46], [0.44, 0.57], [0.43, 0.28], [0.4, 0.43], [0.43, 0.29], [0.41, 0.41], [0.44, 0.3], [0.44, 0.54], [0.43, 0.44]]}" + } + ], + "images": "RH20T_cfg7_task_0036_user_0014_scene_0001_cfg_0007_104122061330#0", + "gt": "[[266, 103], [290, 166], [283, 204], [274, 99], [254, 156], [278, 105], [262, 149], [281, 108], [282, 195], [274, 159]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "34753_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the box from the table and move it to the right', and now you are performing subtask 'pick up the box from the table and move it to the right.'\nYour starting position is [0.6, 0.34]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.34], [0.4, 0.38], [0.22, 0.74], [0.36, 0.53], [0.48, 0.52], [0.47, 0.42], [0.43, 0.52], [0.46, 0.57], [0.48, 0.33], [0.48, 0.33]]}" + } + ], + "images": "34753_exterior_image_1_left#0", + "gt": "[[192, 62], [128, 68], [72, 133], [115, 95], [153, 93], [149, 76], [136, 93], [147, 103], [154, 60], [154, 60]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0104_user_0007_scene_0007_cfg_0002_104122063678#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'drag the plate back after holding it down', and now you are performing subtask 'drag the plate back after holding it down.'\nYou begin at [0.63, 0.53]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.53], [0.62, 0.76], [0.65, 0.79], [0.63, 0.75], [0.62, 0.79], [0.63, 0.75], [0.62, 0.78], [0.62, 0.62], [0.62, 0.68], [0.64, 0.62]]}" + } + ], + "images": "RH20T_cfg2_task_0104_user_0007_scene_0007_cfg_0002_104122063678#0", + "gt": "[[401, 192], [397, 275], [414, 285], [402, 270], [399, 284], [401, 269], [397, 281], [395, 224], [399, 243], [407, 223]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0061_user_0020_scene_0001_cfg_0007_104122061330#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the lid of the gift box on the table, then take everything out of the gift box and place it all on the table', and now you are performing subtask 'place the cube on the tabletop, pick up the object from the gift box, transfer the object from the gift box to the tabletop, place the object on the tabletop.'\nYour initial coordinate is [0.52, 0.57]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.57], [0.54, 0.36], [0.52, 0.39], [0.53, 0.45], [0.53, 0.38], [0.52, 0.38], [0.49, 0.47], [0.47, 0.52], [0.47, 0.57], [0.48, 0.54]]}" + } + ], + "images": "RH20T_cfg7_task_0061_user_0020_scene_0001_cfg_0007_104122061330#398", + "gt": "[[332, 205], [345, 130], [336, 141], [341, 163], [339, 138], [333, 138], [316, 170], [301, 187], [301, 205], [307, 193]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0016_user_0020_scene_0004_cfg_0007_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'hit the pool ball', and now you are performing subtask 'hit the pool ball.'\nBegin at [0.53, 0.32] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.32], [0.54, 0.39], [0.57, 0.38], [0.58, 0.42], [0.58, 0.37], [0.62, 0.32], [0.65, 0.35], [0.63, 0.41], [0.62, 0.28], [0.6, 0.28]]}" + } + ], + "images": "RH20T_cfg7_task_0016_user_0020_scene_0004_cfg_0007_104122064161#0", + "gt": "[[337, 115], [348, 139], [368, 136], [372, 150], [372, 133], [399, 114], [414, 126], [406, 149], [396, 101], [383, 100]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0013_user_0007_scene_0001_cfg_0005_104422070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the block on the scale', and now you are performing subtask 'place the block on the scale.'\nFrom start point [0.59, 0.37], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.59, 0.37], [0.57, 0.36], [0.56, 0.33], [0.58, 0.36], [0.58, 0.38], [0.58, 0.36], [0.59, 0.37], [0.64, 0.48], [0.64, 0.52], [0.6, 0.37]]}" + } + ], + "images": "RH20T_cfg5_task_0013_user_0007_scene_0001_cfg_0005_104422070044#0", + "gt": "[[375, 134], [364, 130], [361, 118], [373, 130], [372, 138], [372, 131], [379, 134], [408, 172], [409, 186], [385, 134]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "79070_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the pen inside the white coffee cup', and now you are performing subtask 'put the pen inside the white coffee cup.'\nStart at [0.48, 0.03]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.03], [0.54, 0.16], [0.57, 0.29], [0.59, 0.43], [0.58, 0.57], [0.64, 0.26], [0.69, 0.28], [0.67, 0.23], [0.64, 0.34], [0.67, 0.24]]}" + } + ], + "images": "79070_exterior_image_1_left#0", + "gt": "[[153.5, 5.0], [172.5, 28.0], [182.5, 53.0], [188.5, 78.0], [185.5, 102.0], [204.5, 47.0], [219.5, 50.0], [215.5, 42.0], [204.5, 61.0], [214.5, 43.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0085_user_0007_scene_0006_cfg_0005_104422070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the bubble ring from the assembled bubble ring and ball and place it on the table', and now you are performing subtask 'remove the bubble ring from the assembled bubble ring and ball and place it on the table.'\nStarting at [0.58, 0.37], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.37], [0.57, 0.36], [0.56, 0.3], [0.56, 0.38], [0.57, 0.41], [0.57, 0.44], [0.56, 0.47], [0.55, 0.44], [0.59, 0.45], [0.59, 0.45]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg5_task_0085_user_0007_scene_0006_cfg_0005_104422070044#0.jpg", + "gt": "[[374, 132], [367, 129], [361, 108], [357, 138], [363, 149], [364, 159], [359, 171], [355, 160], [376, 162], [376, 162]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "88538_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'folding the towel', and now you are performing subtask 'fold the towel on the desk.'\nStart at [0.38, 0.2]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.2], [0.54, 0.18], [0.62, 0.5], [0.6, 0.67], [0.61, 0.73], [0.59, 0.73], [0.61, 0.76], [0.58, 0.52], [0.52, 0.48], [0.41, 0.52]]}" + } + ], + "images": "88538_exterior_image_1_left#0", + "gt": "[[121, 36], [174, 33], [199, 90], [191, 121], [196, 132], [188, 131], [194, 136], [187, 94], [166, 87], [132, 94]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_038522063145#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with small Lego blocks for the nth time with n blocks', and now you are performing subtask 'press the black small Lego block, press the black small Lego block, pick up the white small Lego block from the desktop.'\nStart at [0.69, 0.44]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.44], [0.7, 0.44], [0.7, 0.43], [0.69, 0.47], [0.68, 0.46], [0.69, 0.45], [0.68, 0.46], [0.69, 0.48], [0.7, 0.39], [0.73, 0.37]]}" + } + ], + "images": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_038522063145#676", + "gt": "[[443, 160], [450, 158], [445, 155], [442, 168], [436, 165], [442, 163], [437, 165], [440, 173], [450, 142], [465, 133]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0001_user_0010_scene_0004_cfg_0004_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the button from top to bottom', and now you are performing subtask 'press the button on the surface from top to bottom.'\nYour initial coordinate is [0.38, 0.0]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.6, 0.0], [0.61, 0.05], [0.56, 0.1], [0.55, 0.17], [0.55, 0.14], [0.57, 0.17], [0.58, 0.13], [0.57, 0.15], [0.57, 0.15]]}" + } + ], + "images": "RH20T_cfg4_task_0001_user_0010_scene_0004_cfg_0004_104122062823#0", + "gt": "[[243, 0], [384, 1], [389, 17], [359, 35], [352, 60], [353, 52], [366, 61], [373, 48], [362, 55], [362, 55]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0127_user_0020_scene_0006_cfg_0006_104122064161#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'plug in the power cord of the desk lamp, turn on the socket, and light up the desk lamp', and now you are performing subtask 'transfer the power cable from the table to the socket on the desktop, insert the power cable into the socket on the desktop, press the switch on the desk lamp to turn it on.'\nThe trajectory starts at [0.6, 0.41]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.6, 0.41], [0.59, 0.44], [0.59, 0.41], [0.61, 0.39], [0.61, 0.41], [0.62, 0.39], [0.61, 0.39], [0.6, 0.43], [0.61, 0.32], [0.63, 0.29]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0127_user_0020_scene_0006_cfg_0006_104122064161#319.jpg", + "gt": "[[385, 147], [378, 160], [380, 146], [388, 141], [388, 148], [394, 142], [392, 139], [387, 156], [388, 116], [405, 105]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_035622060973#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'place the first square in the center of the cardboard, pick up the second square from the right side of the table, transfer the second square from the right side of the table to the top of the first square, place the second square beside the first square.'\nStart at [0.38, 0.41]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.41], [0.37, 0.39], [0.31, 0.4], [0.3, 0.42], [0.29, 0.46], [0.29, 0.43], [0.33, 0.4], [0.39, 0.42], [0.38, 0.44], [0.39, 0.43]]}" + } + ], + "images": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_035622060973#200", + "gt": "[[244, 149], [235, 139], [201, 145], [190, 151], [187, 167], [188, 156], [209, 145], [247, 151], [246, 157], [248, 154]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "12799_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the cup', and now you are performing subtask 'move the cup.'\nFrom start point [0.76, 0.03], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.76, 0.03], [0.54, 0.13], [0.52, 0.16], [0.55, 0.13], [0.55, 0.17], [0.47, 0.15], [0.44, 0.31], [0.48, 0.3], [0.51, 0.39], [0.53, 0.34]]}" + } + ], + "images": "12799_exterior_image_1_left#0", + "gt": "[[242, 6], [173, 24], [165, 28], [177, 24], [175, 30], [149, 27], [142, 55], [154, 54], [164, 70], [169, 62]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "88868_exterior_image_2_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the blue shirt', and now you are performing subtask 'fold the blue shirt on the table.'\nYou begin at [0.65, 0.53]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.53], [0.58, 0.51], [0.54, 0.53], [0.55, 0.57], [0.59, 0.48], [0.6, 0.54], [0.58, 0.58], [0.48, 0.67], [0.44, 0.62], [0.52, 0.55]]}" + } + ], + "images": "88868_exterior_image_2_left#234", + "gt": "[[207, 96], [186, 91], [173, 96], [176, 102], [188, 86], [191, 97], [186, 105], [155, 121], [140, 112], [166, 99]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "84471_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the lid from the pot and place it on the table', and now you are performing subtask 'remove the lid from the pot and place it on the table.'\nFrom start point [0.99, 0.21], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.99, 0.21], [0.94, 0.27], [0.81, 0.48], [0.77, 0.63], [0.78, 0.61], [0.77, 0.58], [0.74, 0.62], [0.68, 0.8], [0.68, 0.86], [0.73, 0.78]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/84471_exterior_image_2_left#0.jpg", + "gt": "[[318, 38], [302, 48], [259, 87], [246, 113], [249, 109], [248, 105], [238, 112], [217, 144], [218, 155], [235, 141]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "56042_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the mug to the left', and now you are performing subtask 'move the mug to the left.'\nYou begin at [0.54, 0.01]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.54, 0.01], [0.54, 0.17], [0.49, 0.27], [0.5, 0.32], [0.39, 0.3], [0.34, 0.34], [0.32, 0.41], [0.31, 0.58], [0.33, 0.54], [0.35, 0.39]]}" + } + ], + "images": "56042_exterior_image_1_left#0", + "gt": "[[172.0, 1.5], [173.0, 31.5], [156.0, 48.5], [161.0, 58.5], [126.0, 54.5], [108.0, 60.5], [101.0, 73.5], [100.0, 104.5], [107.0, 97.5], [113.0, 69.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "41728_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the cup from the countertop and pour its contents into the bowl', and now you are performing subtask 'pick up the cup from the countertop and pour its contents into the bowl.'\nYou start from [0.56, 0.59]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.56, 0.59], [0.48, 0.53], [0.42, 0.52], [0.4, 0.56], [0.41, 0.61], [0.44, 0.57], [0.43, 0.57], [0.41, 0.51], [0.39, 0.51], [0.39, 0.46]]}" + } + ], + "images": "41728_exterior_image_1_left#0", + "gt": "[[179, 107], [153, 96], [134, 94], [127, 101], [131, 109], [141, 103], [137, 103], [130, 92], [126, 92], [125, 82]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "70193_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the lid from the blue pot and place it on the table', and now you are performing subtask 'remove the lid from the blue pot and place it on the table.'\nFrom start point [0.74, 0.15], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.74, 0.15], [0.61, 0.1], [0.58, 0.05], [0.52, 0.18], [0.54, 0.25], [0.5, 0.35], [0.47, 0.36], [0.48, 0.32], [0.54, 0.33], [0.61, 0.17]]}" + } + ], + "images": "70193_exterior_image_1_left#0", + "gt": "[[237.0, 27.0], [196.0, 18.0], [185.0, 9.0], [167.0, 32.0], [174.0, 45.0], [159.0, 63.0], [149.0, 65.0], [155.0, 58.0], [172.0, 59.0], [194.0, 30.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0044_user_0010_scene_0007_cfg_0005_105422061350#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up a bag of things', and now you are performing subtask 'pick up a bag of things.'\nStarting at [0.42, 0.14], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.14], [0.42, 0.19], [0.39, 0.17], [0.39, 0.19], [0.37, 0.19], [0.37, 0.21], [0.38, 0.03], [0.38, 0.0], [0.45, 0.0], [0.45, 0.03]]}" + } + ], + "images": "RH20T_cfg5_task_0044_user_0010_scene_0007_cfg_0005_105422061350#0", + "gt": "[[272, 50], [269, 70], [252, 63], [250, 68], [239, 68], [239, 75], [245, 12], [240, 0], [286, 0], [291, 9]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0016_user_0001_scene_0004_cfg_0001_819112070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'the action of hitting the pool ball', and now you are performing subtask 'the action of hitting the pool ball.'\nWith start location [0.61, 0.3], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.61, 0.3], [0.62, 0.35], [0.7, 0.36], [0.68, 0.42], [0.59, 0.35], [0.58, 0.33], [0.64, 0.31], [0.61, 0.32], [0.62, 0.38], [0.62, 0.38]]}" + } + ], + "images": "RH20T_cfg1_task_0016_user_0001_scene_0004_cfg_0001_819112070044#0", + "gt": "[[388, 108], [395, 127], [449, 131], [438, 152], [378, 127], [372, 119], [410, 111], [390, 115], [400, 136], [400, 136]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "7347_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the marker from the right side of the table to the left side of the table', and now you are performing subtask 'move the marker from the right side of the table to the left side of the table.'\nThe trajectory starts at [0.8, 0.02]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.8, 0.02], [0.81, 0.1], [0.73, 0.3], [0.73, 0.44], [0.65, 0.49], [0.69, 0.38], [0.62, 0.48], [0.61, 0.49], [0.62, 0.45], [0.62, 0.45]]}" + } + ], + "images": "7347_exterior_image_1_left#0", + "gt": "[[255.0, 3.5], [259.0, 18.5], [234.0, 53.5], [233.0, 79.5], [207.0, 88.5], [221.0, 67.5], [200.0, 86.5], [194.0, 88.5], [199.0, 81.5], [199.0, 81.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0014_user_0016_scene_0003_cfg_0004_104422070011#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the block on the table next to the scale', and now you are performing subtask 'place the block on the table next to the scale.'\nYou begin at [0.53, 0.29]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.29], [0.5, 0.12], [0.54, 0.15], [0.56, 0.3], [0.5, 0.39], [0.5, 0.45], [0.55, 0.41], [0.57, 0.43], [0.55, 0.43], [0.55, 0.46]]}" + } + ], + "images": "RH20T_cfg4_task_0014_user_0016_scene_0003_cfg_0004_104422070011#0", + "gt": "[[340, 104], [320, 45], [343, 53], [360, 108], [320, 142], [318, 162], [351, 147], [365, 155], [351, 153], [354, 164]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "43239_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the towel', and now you are performing subtask 'shake the towel on the table.'\nThe trajectory starts at [0.86, 0.16]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.86, 0.16], [0.84, 0.2], [0.66, 0.25], [0.64, 0.29], [0.65, 0.32], [0.57, 0.5], [0.63, 0.45], [0.71, 0.53], [0.65, 0.5], [0.6, 0.53]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/43239_exterior_image_1_left#0.jpg", + "gt": "[[276, 29], [270, 36], [211, 45], [205, 52], [208, 58], [182, 90], [202, 81], [226, 96], [209, 90], [191, 95]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "90950_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the blue marker out of the white bowl', and now you are performing subtask 'take the blue marker out of the white bowl.'\nYou start from [0.85, 0.36]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.85, 0.36], [0.66, 0.34], [0.55, 0.43], [0.48, 0.58], [0.48, 0.81], [0.45, 0.82], [0.47, 0.86], [0.51, 0.66], [0.58, 0.79], [0.62, 0.55]]}" + } + ], + "images": "90950_exterior_image_2_left#0", + "gt": "[[272, 65], [210, 62], [177, 77], [155, 104], [155, 146], [145, 148], [152, 154], [163, 118], [185, 143], [200, 99]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "34215_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the small white bottle, keep it upright and move it to the right', and now you are performing subtask 'place the small white bottle upright on the table.'\nYour starting position is [0.73, 0.34]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.73, 0.34], [0.66, 0.56], [0.58, 0.57], [0.5, 0.64], [0.51, 0.69], [0.52, 0.6], [0.5, 0.6], [0.51, 0.56], [0.53, 0.75], [0.56, 0.71]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/34215_exterior_image_1_left#0.jpg", + "gt": "[[232.5, 61.5], [210.5, 100.5], [185.5, 103.5], [160.5, 115.5], [162.5, 123.5], [166.5, 108.5], [158.5, 107.5], [164.5, 101.5], [168.5, 134.5], [179.5, 128.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "52926_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the two pillows and move them to the right side of the couch', and now you are performing subtask 'transfer the second pillow from the sofa to the right side of the sofa.'\nWith start location [0.44, 0.56], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.44, 0.56], [0.4, 0.66], [0.37, 0.61], [0.4, 0.59], [0.42, 0.68], [0.47, 0.73], [0.38, 0.63], [0.42, 0.56], [0.49, 0.61], [0.49, 0.8]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/52926_exterior_image_2_left#231.jpg", + "gt": "[[140, 101], [129, 118], [117, 110], [129, 106], [133, 122], [151, 131], [122, 113], [134, 101], [157, 110], [158, 144]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "61454_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the spoon and ladle from the rack, put them on the table, and then move the rack to the right', and now you are performing subtask 'place the spoon on the table.'\nBegin at [0.43, 0.16] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.16], [0.49, 0.17], [0.55, 0.25], [0.61, 0.24], [0.62, 0.31], [0.56, 0.25], [0.61, 0.3], [0.63, 0.35], [0.61, 0.32], [0.6, 0.26]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/61454_exterior_image_1_left#0.jpg", + "gt": "[[137.0, 29.5], [157.0, 31.5], [175.0, 44.5], [194.0, 43.5], [198.0, 56.5], [180.0, 44.5], [196.0, 53.5], [201.0, 63.5], [194.0, 57.5], [192.0, 47.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "44464_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the plush toy out of the box and put it on the black tray', and now you are performing subtask 'throw the plush toy on the black tray.'\nThe trajectory starts at [0.3, 0.69]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.3, 0.69], [0.36, 0.58], [0.4, 0.43], [0.45, 0.47], [0.5, 0.55], [0.52, 0.59], [0.51, 0.65], [0.52, 0.66], [0.47, 0.62], [0.39, 0.63]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/44464_exterior_image_2_left#0.jpg", + "gt": "[[97, 124], [114, 105], [129, 77], [145, 84], [159, 99], [165, 107], [163, 117], [166, 118], [149, 111], [126, 114]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0024_user_0014_scene_0003_cfg_0004_104122063550#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn off the power strip by pressing the button', and now you are performing subtask 'turn off the power strip by pressing the button.'\nThe trajectory starts at [0.43, 0.52]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.43, 0.52], [0.48, 0.61], [0.44, 0.59], [0.42, 0.7], [0.41, 0.54], [0.43, 0.56], [0.47, 0.7], [0.46, 0.72], [0.44, 0.58], [0.43, 0.51]]}" + } + ], + "images": "RH20T_cfg4_task_0024_user_0014_scene_0003_cfg_0004_104122063550#0", + "gt": "[[275, 186], [305, 220], [281, 213], [269, 251], [264, 196], [274, 200], [304, 253], [296, 259], [279, 210], [277, 182]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0014_user_0016_scene_0003_cfg_0004_036422060909#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the block on the table next to the scale', and now you are performing subtask 'place the block on the table next to the scale.'\nYou begin at [0.53, 0.22]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.22], [0.6, 0.28], [0.56, 0.28], [0.5, 0.2], [0.5, 0.37], [0.47, 0.34], [0.43, 0.39], [0.46, 0.32], [0.44, 0.33], [0.44, 0.33]]}" + } + ], + "images": "RH20T_cfg4_task_0014_user_0016_scene_0003_cfg_0004_036422060909#0", + "gt": "[[337, 78], [383, 99], [357, 102], [317, 72], [323, 133], [303, 123], [275, 140], [295, 116], [284, 119], [284, 119]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "83270_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the square from the left side of the yellow plate and place it on the right side of the plate', and now you are performing subtask 'take the square from the left side of the yellow plate and place it on the right side of the plate.'\nYour initial coordinate is [0.31, 0.0]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.31, 0.0], [0.32, 0.44], [0.37, 0.48], [0.35, 0.46], [0.36, 0.42], [0.42, 0.44], [0.44, 0.5], [0.44, 0.59], [0.45, 0.57], [0.4, 0.42]]}" + } + ], + "images": "83270_exterior_image_1_left#0", + "gt": "[[98, 0], [103, 79], [117, 86], [112, 82], [114, 76], [135, 80], [140, 90], [140, 107], [143, 103], [128, 75]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_038522062547#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the third block from the table, transfer the third block from the table to the position above the fourth block, place the third block on top of the second block.'\nYour starting position is [0.65, 0.34]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.34], [0.64, 0.44], [0.62, 0.44], [0.63, 0.47], [0.65, 0.28], [0.6, 0.12], [0.64, 0.17], [0.65, 0.22], [0.65, 0.31], [0.62, 0.35]]}" + } + ], + "images": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_038522062547#253", + "gt": "[[413, 124], [407, 157], [395, 158], [405, 170], [413, 101], [387, 42], [410, 63], [415, 80], [416, 111], [399, 126]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "65770_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push the blocks on the table, put the orange block in the middle of the marked spot', and now you are performing subtask 'push the small object on the table.'\nYour initial coordinate is [0.24, 0.17]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.24, 0.17], [0.34, 0.12], [0.32, 0.16], [0.47, 0.19], [0.5, 0.27], [0.46, 0.29], [0.47, 0.37], [0.44, 0.34], [0.5, 0.44], [0.42, 0.41]]}" + } + ], + "images": "65770_exterior_image_1_left#0", + "gt": "[[78, 30], [109, 21], [104, 29], [152, 34], [159, 48], [147, 52], [152, 66], [142, 61], [159, 80], [133, 74]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0094_user_0007_scene_0009_cfg_0005_105422061350#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'press the cardboard box on the side, press the cardboard box on the side.'\nStart at [0.38, 0.07]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.07], [0.37, 0.1], [0.37, 0.15], [0.36, 0.17], [0.36, 0.19], [0.35, 0.19], [0.34, 0.22], [0.34, 0.28], [0.34, 0.24], [0.36, 0.18]]}" + } + ], + "images": "RH20T_cfg5_task_0094_user_0007_scene_0009_cfg_0005_105422061350#213", + "gt": "[[243, 25], [239, 36], [237, 53], [231, 61], [230, 68], [227, 69], [220, 81], [216, 99], [218, 87], [229, 65]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "85110_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the black button on the right side of the board', and now you are performing subtask 'press the black button on the right side of the board.'\nWith start location [0.47, 0.01], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.01], [0.64, 0.18], [0.62, 0.38], [0.68, 0.47], [0.73, 0.78], [0.77, 0.79], [0.73, 0.78], [0.74, 0.83], [0.79, 0.63], [0.75, 0.44]]}" + } + ], + "images": "85110_exterior_image_1_left#0", + "gt": "[[150, 1], [206, 33], [198, 68], [219, 85], [234, 140], [247, 142], [232, 141], [238, 150], [253, 113], [239, 79]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "46919_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the door of the washing machine and push the tray back in', and now you are performing subtask 'push the tray back into the washing machine.'\nYou start from [0.39, 0.4]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.4], [0.42, 0.45], [0.4, 0.56], [0.43, 0.61], [0.43, 0.66], [0.36, 0.72], [0.34, 0.71], [0.38, 0.66], [0.44, 0.63], [0.48, 0.79]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/46919_exterior_image_1_left#0.jpg", + "gt": "[[126, 72], [133, 81], [127, 100], [137, 109], [139, 118], [115, 130], [110, 128], [121, 118], [141, 114], [154, 142]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "19783_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the button on the adaptor and push the switch on the electric kettle', and now you are performing subtask 'press the button on the adaptor.'\nStart at [0.3, 0.56]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.3, 0.56], [0.35, 0.55], [0.4, 0.62], [0.43, 0.74], [0.51, 0.89], [0.52, 0.98], [0.48, 1.0], [0.49, 0.82], [0.49, 0.74], [0.5, 0.77]]}" + } + ], + "images": "19783_exterior_image_2_left#0", + "gt": "[[96, 101], [111, 99], [129, 112], [138, 133], [164, 161], [167, 177], [153, 180], [158, 147], [156, 134], [159, 138]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0037_user_0007_scene_0007_cfg_0002_104422070044#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the second block on top of the first block, pick up the third block from the surface, transfer the third block from the surface to the top of the second block, place the third block on top of the second block.'\nStart at [0.51, 0.54]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.54], [0.51, 0.53], [0.52, 0.52], [0.52, 0.51], [0.53, 0.52], [0.52, 0.52], [0.52, 0.48], [0.52, 0.51], [0.51, 0.52], [0.51, 0.52]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg2_task_0037_user_0007_scene_0007_cfg_0002_104422070044#262.jpg", + "gt": "[[326, 194], [326, 192], [332, 189], [335, 185], [337, 188], [334, 187], [335, 172], [330, 183], [325, 187], [325, 189]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "10276_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the tap, put the bowl on the dish rack, and open the sink plug', and now you are performing subtask 'place the bowl on the dish rack.'\nYour starting position is [0.66, 0.43]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.66, 0.43], [0.63, 0.51], [0.71, 0.48], [0.72, 0.36], [0.51, 0.4], [0.49, 0.51], [0.49, 0.44], [0.51, 0.41], [0.6, 0.33], [0.63, 0.36]]}" + } + ], + "images": "10276_exterior_image_2_left#161", + "gt": "[[211, 77], [203, 91], [226, 87], [231, 64], [164, 72], [157, 91], [157, 80], [164, 74], [192, 59], [202, 64]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0068_user_0007_scene_0005_cfg_0004_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the red button on the desktop', and now you are performing subtask 'press the red button on the desktop.'\nYou start from [0.45, 0.0]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.0], [0.83, 0.0], [0.85, 0.11], [0.81, 0.34], [0.86, 0.38], [0.76, 0.31], [0.72, 0.24], [0.82, 0.0], [0.46, 0.0], [0.5, 0.05]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0068_user_0007_scene_0005_cfg_0004_104122062823#0.jpg", + "gt": "[[291, 0], [534, 0], [547, 40], [521, 123], [551, 137], [486, 112], [459, 85], [523, 0], [292, 0], [320, 18]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0073_user_0010_scene_0010_cfg_0003_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the things in the drawer', and now you are performing subtask 'pick up the small square from the table, transfer the small square from the table to the open drawer, place the small square in the drawer.'\nFrom start point [0.38, 0.0], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.55, 0.0], [0.54, 0.15], [0.51, 0.17], [0.57, 0.39], [0.54, 0.43], [0.53, 0.59], [0.49, 0.29], [0.43, 0.11], [0.43, 0.11]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg3_task_0073_user_0010_scene_0010_cfg_0003_104122062823#0.jpg", + "gt": "[[246, 0], [354, 0], [348, 54], [329, 60], [366, 141], [348, 154], [339, 211], [316, 104], [277, 39], [277, 39]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0016_user_0014_scene_0008_cfg_0001_039422060546#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'hit the pool ball', and now you are performing subtask 'strike the pool ball with a tool.'\nFrom start point [0.52, 0.51], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.51], [0.52, 0.51], [0.52, 0.51], [0.52, 0.51], [0.52, 0.5], [0.52, 0.5], [0.52, 0.5], [0.52, 0.5], [0.52, 0.5], [0.52, 0.49]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0016_user_0014_scene_0008_cfg_0001_039422060546#284.jpg", + "gt": "[[335, 184], [335, 182], [334, 183], [332, 182], [331, 179], [332, 179], [331, 180], [333, 179], [331, 179], [331, 177]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0038_user_0014_scene_0004_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the cup', and now you are performing subtask 'pick up the cup.'\nBegin at [0.63, 0.37] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.64, 0.34], [0.64, 0.26], [0.6, 0.17], [0.59, 0.12], [0.57, 0.14], [0.58, 0.17], [0.59, 0.19], [0.61, 0.16], [0.59, 0.15]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0038_user_0014_scene_0004_cfg_0001_038522062547#0.jpg", + "gt": "[[406, 134], [410, 124], [407, 93], [383, 60], [379, 44], [366, 49], [370, 63], [379, 68], [388, 57], [375, 53]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "17981_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the orange cloth', and now you are performing subtask 'push the cloth to the right side of the table.'\nYour initial coordinate is [0.49, 0.41]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.49, 0.41], [0.56, 0.41], [0.54, 0.43], [0.56, 0.46], [0.68, 0.56], [0.66, 0.58], [0.73, 0.72], [0.72, 0.8], [0.71, 0.7], [0.72, 0.75]]}" + } + ], + "images": "17981_exterior_image_2_left#0", + "gt": "[[158, 74], [180, 74], [174, 78], [179, 83], [217, 101], [212, 105], [235, 130], [231, 144], [226, 126], [229, 135]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "81019_exterior_image_1_left#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the fruits on the bowl and put them on the counter', and now you are performing subtask 'transfer the fourth fruit from the bowl to the countertop.'\nYou start from [0.78, 0.64]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.78, 0.64], [0.74, 0.57], [0.69, 0.55], [0.63, 0.6], [0.62, 0.63], [0.61, 0.67], [0.63, 0.59], [0.68, 0.53], [0.73, 0.56], [0.72, 0.62]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/81019_exterior_image_1_left#406.jpg", + "gt": "[[250, 116], [236, 102], [221, 99], [202, 108], [197, 113], [196, 120], [201, 107], [218, 95], [233, 101], [231, 112]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0094_user_0007_scene_0009_cfg_0005_104122061850#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'press the cardboard box on the side.'\nYour starting position is [0.52, 0.38]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.38], [0.57, 0.4], [0.62, 0.38], [0.62, 0.41], [0.62, 0.43], [0.62, 0.39], [0.58, 0.35], [0.54, 0.32], [0.52, 0.32], [0.51, 0.32]]}" + } + ], + "images": "RH20T_cfg5_task_0094_user_0007_scene_0009_cfg_0005_104122061850#487", + "gt": "[[335, 138], [366, 144], [396, 138], [400, 147], [400, 153], [399, 142], [372, 126], [347, 117], [335, 114], [329, 115]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "77225_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the pen from the mug and place it on the table', and now you are performing subtask 'remove the pen from the mug and place it on the table.'\nThe trajectory starts at [0.62, 0.01]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.01], [0.43, 0.21], [0.46, 0.22], [0.39, 0.34], [0.38, 0.16], [0.41, 0.09], [0.45, 0.09], [0.52, 0.16], [0.59, 0.33], [0.6, 0.24]]}" + } + ], + "images": "77225_exterior_image_2_left#0", + "gt": "[[199, 1], [139, 38], [147, 39], [125, 62], [122, 28], [130, 16], [143, 17], [167, 28], [189, 59], [192, 44]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0037_user_0020_scene_0009_cfg_0006_104122061602#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the first block on the second block, pick up the third block from the table, transfer the third block from the table to the first block, place the third block on the first block, pick up the fourth block from the table.'\nYou start from [0.65, 0.35]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.65, 0.35], [0.64, 0.25], [0.62, 0.34], [0.63, 0.28], [0.67, 0.28], [0.69, 0.32], [0.7, 0.41], [0.72, 0.25], [0.75, 0.26], [0.76, 0.33]]}" + } + ], + "images": "RH20T_cfg6_task_0037_user_0020_scene_0009_cfg_0006_104122061602#553", + "gt": "[[416, 126], [408, 91], [399, 124], [406, 100], [427, 100], [443, 116], [448, 147], [458, 90], [481, 95], [489, 119]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "25145_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'hang the yellow peg on the wooden object', and now you are performing subtask 'hang the yellow peg on the wooden object.'\nBegin at [0.91, 0.52] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.91, 0.52], [0.72, 0.56], [0.62, 0.7], [0.57, 0.69], [0.55, 0.86], [0.62, 0.67], [0.54, 0.54], [0.47, 0.58], [0.52, 0.67], [0.56, 0.64]]}" + } + ], + "images": "25145_exterior_image_1_left#0", + "gt": "[[291, 93], [230, 101], [199, 126], [183, 124], [175, 155], [199, 121], [174, 97], [149, 105], [167, 120], [180, 116]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "29263_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the button on the power strip', and now you are performing subtask 'press the button on the power strip.'\nThe trajectory starts at [0.18, 0.0]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.18, 0.0], [0.24, 0.14], [0.25, 0.37], [0.23, 0.4], [0.32, 0.42], [0.34, 0.49], [0.34, 0.56], [0.35, 0.54], [0.36, 0.59], [0.35, 0.6]]}" + } + ], + "images": "29263_exterior_image_2_left#0", + "gt": "[[59, 0], [77, 26], [80, 67], [74, 72], [101, 76], [109, 89], [109, 100], [112, 97], [114, 106], [111, 108]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_750612070851#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn them on', and now you are performing subtask 'press the second button on the right side of the power strip.'\nWith start location [0.21, 0.55], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.21, 0.55], [0.22, 0.57], [0.21, 0.57], [0.21, 0.58], [0.21, 0.59], [0.21, 0.58], [0.2, 0.56], [0.2, 0.52], [0.2, 0.53], [0.2, 0.53]]}" + } + ], + "images": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_750612070851#179", + "gt": "[[135, 197], [140, 207], [133, 207], [137, 210], [135, 211], [132, 209], [128, 203], [126, 186], [125, 192], [128, 192]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0081_user_0010_scene_0001_cfg_0001_035622060973#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the leg of the glasses (with one hand)', and now you are performing subtask 'pull the leg of the glasses outward.'\nBegin at [0.38, 0.37] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.37], [0.39, 0.42], [0.41, 0.45], [0.42, 0.51], [0.43, 0.5], [0.43, 0.51], [0.43, 0.49], [0.42, 0.49], [0.42, 0.49], [0.42, 0.49]]}" + } + ], + "images": "RH20T_cfg1_task_0081_user_0010_scene_0001_cfg_0001_035622060973#0", + "gt": "[[242, 133], [249, 151], [262, 161], [269, 182], [273, 179], [275, 183], [273, 176], [269, 175], [266, 178], [268, 176]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122061602#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with large Megabloks for the nth time with n blocks', and now you are performing subtask 'place the block on the block board, pick up the block on the flat surface, transfer the block from the flat surface to the block board, place the block on the block board.'\nStart at [0.73, 0.53]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.73, 0.53], [0.75, 0.5], [0.78, 0.5], [0.8, 0.5], [0.82, 0.54], [0.81, 0.47], [0.73, 0.45], [0.69, 0.47], [0.68, 0.51], [0.68, 0.49]]}" + } + ], + "images": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122061602#224", + "gt": "[[469, 191], [482, 180], [501, 179], [515, 180], [526, 193], [518, 169], [468, 161], [439, 171], [433, 182], [434, 177]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0037_user_0014_scene_0001_cfg_0007_037522061512#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line', and now you are performing subtask 'transfer the third block from the table to the stacking position, place the third block on top of the second block, pick up the fourth block from the table, transfer the fourth block from the table to the stacking position, place the fourth block on top of the third block.'\nThe trajectory starts at [0.45, 0.26]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.26], [0.44, 0.24], [0.4, 0.29], [0.39, 0.28], [0.38, 0.32], [0.4, 0.21], [0.42, 0.21], [0.45, 0.24], [0.43, 0.21], [0.4, 0.25]]}" + } + ], + "images": "RH20T_cfg7_task_0037_user_0014_scene_0001_cfg_0007_037522061512#327", + "gt": "[[291, 93], [279, 87], [258, 105], [247, 102], [246, 116], [253, 77], [268, 74], [287, 85], [275, 75], [257, 89]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "16392_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take a bottle from the tray and place it on the table', and now you are performing subtask 'take a bottle from the tray and place it on the table.'\nThe trajectory starts at [0.36, 0.96]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.36, 0.96], [0.6, 0.17], [0.76, 0.45], [0.78, 0.38], [0.78, 0.41], [0.74, 0.41], [0.73, 0.46], [0.84, 0.49], [0.88, 0.47], [0.74, 0.4]]}" + } + ], + "images": "16392_exterior_image_2_left#0", + "gt": "[[116, 172], [191, 31], [242, 81], [249, 68], [250, 74], [238, 74], [232, 83], [268, 89], [280, 84], [237, 72]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0013_user_0016_scene_0007_cfg_0004_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the block on the scale', and now you are performing subtask 'place the block on the scale.'\nYour starting position is [0.38, 0.0]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.31, 0.0], [0.41, 0.01], [0.39, 0.08], [0.42, 0.12], [0.43, 0.2], [0.41, 0.21], [0.43, 0.18], [0.41, 0.15], [0.41, 0.09]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg4_task_0013_user_0016_scene_0007_cfg_0004_104122062823#0.jpg", + "gt": "[[243, 0], [198, 0], [261, 5], [252, 30], [266, 43], [272, 73], [265, 74], [274, 66], [262, 53], [260, 34]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0037_user_0020_scene_0009_cfg_0006_104122061330#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'place the first block on the second block, pick up the third block from the table, transfer the third block from the table to the first block, place the third block on the first block, pick up the fourth block from the table.'\nYour initial coordinate is [0.47, 0.43]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.47, 0.43], [0.5, 0.3], [0.6, 0.2], [0.58, 0.28], [0.58, 0.22], [0.52, 0.25], [0.46, 0.35], [0.46, 0.39], [0.55, 0.27], [0.59, 0.46]]}" + } + ], + "images": "RH20T_cfg6_task_0037_user_0020_scene_0009_cfg_0006_104122061330#459", + "gt": "[[299, 154], [322, 107], [382, 73], [372, 101], [371, 81], [336, 90], [293, 127], [296, 140], [352, 96], [375, 165]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "18908_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the marker from the mug and place it on the table', and now you are performing subtask 'remove the marker from the mug and place it on the table.'\nBegin at [0.51, 0.26] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.51, 0.26], [0.44, 0.27], [0.39, 0.33], [0.37, 0.56], [0.37, 0.51], [0.35, 0.47], [0.32, 0.51], [0.34, 0.64], [0.34, 0.48], [0.34, 0.48]]}" + } + ], + "images": "18908_exterior_image_2_left#0", + "gt": "[[162, 46], [142, 49], [125, 59], [117, 100], [118, 91], [112, 85], [104, 91], [109, 115], [110, 86], [110, 86]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "86543_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the glass lid from the black pot', and now you are performing subtask 'remove the glass lid from the black pot.'\nWith start location [0.35, 0.01], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.01], [0.4, 0.22], [0.36, 0.44], [0.36, 0.35], [0.44, 0.31], [0.48, 0.37], [0.46, 0.48], [0.45, 0.49], [0.45, 0.46], [0.45, 0.47]]}" + } + ], + "images": "86543_exterior_image_1_left#0", + "gt": "[[113.5, 2.0], [127.5, 40.0], [116.5, 79.0], [115.5, 63.0], [141.5, 56.0], [153.5, 66.0], [147.5, 86.0], [144.5, 89.0], [144.5, 82.0], [142.5, 85.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_038522062547#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn them on', and now you are performing subtask 'press the first button on the right side of the power strip.'\nWith start location [0.63, 0.37], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.63, 0.37], [0.62, 0.38], [0.62, 0.58], [0.6, 0.66], [0.59, 0.77], [0.57, 0.87], [0.56, 0.91], [0.56, 0.88], [0.55, 0.88], [0.6, 0.82]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_038522062547#0.jpg", + "gt": "[[405, 132], [399, 135], [399, 210], [381, 239], [375, 279], [362, 314], [356, 326], [358, 318], [354, 318], [384, 296]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "78597_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the towel', and now you are performing subtask 'fold the towel on the countertop.'\nStarting at [0.93, 0.31], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.93, 0.31], [0.65, 0.59], [0.69, 0.63], [0.68, 0.54], [0.62, 0.49], [0.58, 0.49], [0.58, 0.63], [0.57, 0.57], [0.56, 0.61], [0.56, 0.61]]}" + } + ], + "images": "78597_exterior_image_1_left#0", + "gt": "[[296, 55], [209, 106], [220, 113], [218, 97], [199, 88], [187, 89], [185, 114], [182, 102], [179, 110], [179, 110]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "55462_exterior_image_1_left#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'set the two pillows against the armrest', and now you are performing subtask 'pull the second pillow from the sofa to the backrest position.'\nWith start location [0.58, 0.58], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.58], [0.59, 0.57], [0.59, 0.57], [0.6, 0.57], [0.6, 0.56], [0.61, 0.56], [0.61, 0.57], [0.59, 0.56], [0.59, 0.54], [0.59, 0.54]]}" + } + ], + "images": "55462_exterior_image_1_left#475", + "gt": "[[187, 104], [190, 103], [190, 102], [192, 102], [192, 101], [195, 101], [194, 102], [189, 101], [189, 98], [190, 98]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "62246_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the marker in the bowl', and now you are performing subtask 'put the marker in the bowl.'\nStarting at [0.0, 0.13], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.0, 0.13], [0.16, 0.31], [0.22, 0.46], [0.3, 0.56], [0.37, 0.7], [0.38, 0.62], [0.43, 0.57], [0.5, 0.54], [0.52, 0.56], [0.52, 0.61]]}" + } + ], + "images": "62246_exterior_image_1_left#0", + "gt": "[[0.5, 23.0], [52.5, 56.0], [70.5, 82.0], [97.5, 100.0], [118.5, 126.0], [122.5, 112.0], [137.5, 102.0], [161.5, 98.0], [165.5, 100.0], [165.5, 110.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "66766_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the open drawer', and now you are performing subtask 'close the open drawer.'\nYou start from [0.91, 0.03]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.91, 0.03], [0.86, 0.25], [0.81, 0.36], [0.75, 0.41], [0.73, 0.4], [0.72, 0.47], [0.69, 0.47], [0.68, 0.42], [0.68, 0.44], [0.66, 0.39]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/66766_exterior_image_2_left#0.jpg", + "gt": "[[291, 5], [274, 45], [258, 65], [240, 74], [232, 72], [230, 85], [220, 84], [219, 76], [217, 79], [211, 71]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "230_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the pen and put it in the cup', and now you are performing subtask 'pick up the pen and put it in the cup.'\nBegin at [0.52, 0.26] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.26], [0.46, 0.32], [0.4, 0.56], [0.32, 0.64], [0.33, 0.68], [0.37, 0.43], [0.42, 0.34], [0.46, 0.38], [0.45, 0.43], [0.39, 0.39]]}" + } + ], + "images": "230_exterior_image_2_left#0", + "gt": "[[166, 46], [146, 58], [128, 101], [102, 115], [105, 122], [118, 77], [134, 62], [148, 68], [144, 77], [126, 71]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "89414_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the marker out of the cup and place it on the table', and now you are performing subtask 'take the marker out of the cup and place it on the table.'\nYou start from [0.5, 0.12]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.5, 0.12], [0.67, 0.29], [0.66, 0.32], [0.71, 0.42], [0.71, 0.36], [0.66, 0.36], [0.64, 0.48], [0.65, 0.48], [0.65, 0.46], [0.63, 0.32]]}" + } + ], + "images": "89414_exterior_image_1_left#0", + "gt": "[[160.5, 21.0], [214.5, 53.0], [212.5, 58.0], [226.5, 75.0], [226.5, 64.0], [211.5, 65.0], [204.5, 87.0], [208.5, 87.0], [208.5, 82.0], [202.5, 58.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0073_user_0007_scene_0001_cfg_0005_037522062165#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the things in the drawer', and now you are performing subtask 'grab the handle to pull the drawer open.'\nFrom start point [0.45, 0.42], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.42], [0.43, 0.47], [0.37, 0.48], [0.34, 0.47], [0.32, 0.41], [0.3, 0.4], [0.31, 0.44], [0.38, 0.42], [0.38, 0.46], [0.37, 0.42]]}" + } + ], + "images": "RH20T_cfg5_task_0073_user_0007_scene_0001_cfg_0005_037522062165#0", + "gt": "[[289, 152], [273, 171], [238, 173], [220, 169], [204, 148], [190, 145], [201, 159], [240, 151], [241, 166], [237, 151]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "55098_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the snack pack in the cabinet', and now you are performing subtask 'put the snack pack in the cabinet.'\nWith start location [0.25, 0.56], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.25, 0.56], [0.4, 0.79], [0.4, 0.73], [0.32, 0.65], [0.29, 0.41], [0.33, 0.09], [0.43, 0.19], [0.37, 0.05], [0.33, 0.15], [0.32, 0.52]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/55098_exterior_image_1_left#0.jpg", + "gt": "[[81.0, 101.5], [128.0, 142.5], [128.0, 131.5], [104.0, 117.5], [94.0, 74.5], [107.0, 15.5], [139.0, 34.5], [118.0, 8.5], [105.0, 26.5], [104.0, 94.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg2_task_0104_user_0007_scene_0007_cfg_0002_104122061850#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'drag the plate back after holding it down', and now you are performing subtask 'drag the plate back after holding it down.'\nYour initial coordinate is [0.53, 0.29]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.29], [0.53, 0.55], [0.54, 0.5], [0.52, 0.52], [0.57, 0.38], [0.57, 0.44], [0.58, 0.39], [0.57, 0.42], [0.6, 0.38], [0.56, 0.39]]}" + } + ], + "images": "RH20T_cfg2_task_0104_user_0007_scene_0007_cfg_0002_104122061850#0", + "gt": "[[340, 106], [337, 199], [345, 181], [333, 188], [365, 138], [362, 157], [370, 139], [366, 150], [385, 137], [357, 140]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "44675_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the maize cob to the left', and now you are performing subtask 'move the maize cob to the left.'\nStart at [0.81, 0.0]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.81, 0.0], [0.75, 0.26], [0.62, 0.47], [0.62, 0.59], [0.58, 0.64], [0.48, 0.6], [0.46, 0.69], [0.49, 0.77], [0.49, 0.57], [0.44, 0.58]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/44675_exterior_image_2_left#0.jpg", + "gt": "[[260, 0], [239, 47], [197, 84], [198, 107], [185, 115], [155, 108], [148, 124], [157, 138], [157, 103], [142, 104]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0057_user_0014_scene_0008_cfg_0006_104122063633#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press the buttons on a row of four power strips to turn on the power strips', and now you are performing subtask 'press the third button on the power strip, press the fourth button on the power strip.'\nStart at [0.31, 0.45]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.31, 0.45], [0.3, 0.45], [0.3, 0.47], [0.3, 0.46], [0.3, 0.43], [0.33, 0.4], [0.35, 0.4], [0.36, 0.41], [0.36, 0.41], [0.36, 0.43]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0057_user_0014_scene_0008_cfg_0006_104122063633#220.jpg", + "gt": "[[197, 161], [191, 163], [189, 168], [192, 166], [195, 155], [210, 144], [222, 143], [233, 146], [233, 149], [228, 154]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg1_task_0051_user_0015_scene_0002_cfg_0001_038522063145#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the dish off the dish rack', and now you are performing subtask 'take the dish off the dish rack.'\nYou start from [0.69, 0.27]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.27], [0.79, 0.3], [0.75, 0.34], [0.76, 0.29], [0.74, 0.29], [0.73, 0.18], [0.66, 0.1], [0.6, 0.07], [0.56, 0.13], [0.64, 0.19]]}" + } + ], + "images": "RH20T_cfg1_task_0051_user_0015_scene_0002_cfg_0001_038522063145#0", + "gt": "[[443, 98], [503, 108], [480, 122], [486, 105], [473, 103], [468, 66], [421, 35], [381, 26], [357, 46], [411, 69]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "6777_exterior_image_2_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'the contents in the flask', and now you are performing subtask 'throw away the brush.'\nBegin at [0.29, 0.24] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.29, 0.24], [0.29, 0.17], [0.35, 0.36], [0.35, 0.57], [0.3, 0.17], [0.34, 0.31], [0.44, 0.42], [0.47, 0.64], [0.49, 0.65], [0.48, 0.53]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/6777_exterior_image_2_left#179.jpg", + "gt": "[[94, 44], [93, 30], [113, 65], [113, 103], [95, 30], [110, 56], [140, 76], [149, 116], [156, 117], [153, 95]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "85991_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'push down the switch of the toaster', and now you are performing subtask 'push down the switch of the toaster.'\nYou begin at [0.79, 0.25]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.79, 0.25], [0.7, 0.23], [0.6, 0.14], [0.55, 0.12], [0.48, 0.17], [0.44, 0.24], [0.44, 0.34], [0.46, 0.28], [0.51, 0.23], [0.61, 0.23]]}" + } + ], + "images": "85991_exterior_image_2_left#0", + "gt": "[[253, 45], [225, 42], [193, 25], [175, 22], [155, 30], [141, 43], [142, 61], [148, 51], [164, 42], [195, 42]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0034_user_0014_scene_0001_cfg_0007_104122064161#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the squares into a pyramid shape', and now you are performing subtask 'pick up the square, place the square on the table, pick up the square.'\nYou begin at [0.57, 0.57]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.57], [0.58, 0.59], [0.58, 0.64], [0.57, 0.66], [0.57, 0.65], [0.56, 0.62], [0.56, 0.62], [0.55, 0.57], [0.58, 0.6], [0.58, 0.59]]}" + } + ], + "images": "RH20T_cfg7_task_0034_user_0014_scene_0001_cfg_0007_104122064161#189", + "gt": "[[364, 204], [371, 213], [369, 232], [365, 238], [365, 235], [358, 223], [359, 222], [354, 206], [370, 215], [373, 213]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "30069_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the cloth and the shirt off the back of the chair and put them on the seat', and now you are performing subtask 'throw the cloth on the seat of the chair.'\nBegin at [0.69, 0.13] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.13], [0.64, 0.01], [0.66, 0.03], [0.57, 0.13], [0.55, 0.17], [0.56, 0.19], [0.52, 0.08], [0.47, 0.06], [0.45, 0.1], [0.45, 0.17]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/30069_exterior_image_1_left#0.jpg", + "gt": "[[221, 24], [205, 1], [210, 6], [183, 23], [177, 30], [179, 35], [166, 15], [151, 10], [143, 18], [143, 31]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0096_user_0015_scene_0005_cfg_0001_038522063145#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the block out of the drawer and put it on the table', and now you are performing subtask 'pull the drawer handle to open the drawer.'\nWith start location [0.69, 0.27], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.27], [0.75, 0.29], [0.68, 0.18], [0.65, 0.17], [0.62, 0.22], [0.63, 0.21], [0.65, 0.19], [0.66, 0.19], [0.7, 0.14], [0.7, 0.13]]}" + } + ], + "images": "RH20T_cfg1_task_0096_user_0015_scene_0005_cfg_0001_038522063145#0", + "gt": "[[441, 98], [479, 105], [433, 64], [415, 61], [395, 78], [404, 77], [413, 69], [420, 70], [448, 50], [445, 48]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0061_user_0020_scene_0001_cfg_0007_104122061018#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the lid of the gift box on the table, then take everything out of the gift box and place it all on the table', and now you are performing subtask 'place the object on the tabletop, pick up the block from the gift box, transfer the block from the gift box to the tabletop, place the block on the tabletop.'\nStart at [0.45, 0.27]. According to the current scene, compute 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.27], [0.45, 0.24], [0.49, 0.35], [0.52, 0.47], [0.49, 0.31], [0.45, 0.2], [0.41, 0.13], [0.39, 0.21], [0.42, 0.14], [0.49, 0.07]]}" + } + ], + "images": "RH20T_cfg7_task_0061_user_0020_scene_0001_cfg_0007_104122061018#477", + "gt": "[[288, 98], [289, 86], [316, 126], [334, 169], [316, 113], [291, 73], [261, 48], [250, 76], [267, 50], [316, 25]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg5_task_0058_user_0007_scene_0002_cfg_0005_104122063678#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press a row of two power strip buttons to turn off the power strips', and now you are performing subtask 'press the second button on the power socket.'\nYou start from [0.52, 0.62]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.52, 0.62], [0.51, 0.62], [0.51, 0.63], [0.51, 0.64], [0.51, 0.64], [0.51, 0.65], [0.51, 0.65], [0.51, 0.64], [0.51, 0.62], [0.51, 0.62]]}" + } + ], + "images": "RH20T_cfg5_task_0058_user_0007_scene_0002_cfg_0005_104122063678#276", + "gt": "[[333, 224], [327, 224], [327, 226], [325, 229], [326, 232], [327, 233], [328, 233], [326, 230], [326, 224], [325, 223]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg6_task_0003_user_0014_scene_0010_cfg_0006_104122060811#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'press three buttons from left to right in sequence', and now you are performing subtask 'press the green button in the middle.'\nYou begin at [0.61, 0.3]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.61, 0.3], [0.6, 0.3], [0.6, 0.32], [0.61, 0.33], [0.61, 0.33], [0.61, 0.34], [0.61, 0.34], [0.61, 0.34], [0.61, 0.34], [0.61, 0.32]]}" + } + ], + "images": "RH20T_cfg6_task_0003_user_0014_scene_0010_cfg_0006_104122060811#116", + "gt": "[[390, 108], [387, 109], [386, 115], [389, 118], [389, 120], [392, 121], [392, 123], [390, 124], [391, 123], [388, 117]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg3_task_0008_user_0016_scene_0009_cfg_0003_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'grab the block and place it at the designated location', and now you are performing subtask 'grab the block and place it at the designated location.'\nStarting at [0.38, 0.0], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.18, 0.0], [0.41, 0.0], [0.34, 0.0], [0.36, 0.12], [0.33, 0.05], [0.34, 0.0], [0.36, 0.12], [0.44, 0.22], [0.43, 0.09]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg3_task_0008_user_0016_scene_0009_cfg_0003_104122062823#0.jpg", + "gt": "[[242, 0], [118, 0], [264, 0], [220, 0], [231, 43], [214, 18], [218, 1], [233, 43], [281, 79], [276, 31]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "865_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the wooden spoon to stir the contents of the pot', and now you are performing subtask 'use the wooden spoon to stir the contents of the pot.'\nBegin at [0.41, 0.33] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.41, 0.33], [0.5, 0.31], [0.56, 0.41], [0.6, 0.32], [0.53, 0.32], [0.6, 0.32], [0.53, 0.31], [0.61, 0.34], [0.55, 0.37], [0.54, 0.32]]}" + } + ], + "images": "865_exterior_image_1_left#0", + "gt": "[[131.0, 59.5], [160.0, 56.5], [178.0, 73.5], [191.0, 57.5], [171.0, 57.5], [193.0, 58.5], [171.0, 55.5], [194.0, 60.5], [175.0, 66.5], [174.0, 57.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_038522063145#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'build with small Lego blocks for the nth time with n blocks', and now you are performing subtask 'pick up the white small Lego block from the desktop, transfer the white small Lego block from the desktop to the black small Lego block, place the white small Lego block on the black small Lego block.'\nStarting at [0.73, 0.37], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.73, 0.37], [0.75, 0.39], [0.74, 0.38], [0.74, 0.39], [0.73, 0.4], [0.73, 0.39], [0.73, 0.4], [0.73, 0.38], [0.7, 0.45], [0.7, 0.45]]}" + } + ], + "images": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_038522063145#840", + "gt": "[[465, 133], [482, 141], [472, 136], [472, 142], [470, 144], [469, 140], [466, 145], [466, 137], [447, 161], [447, 161]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "488_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the marker from the cup and place it on the table', and now you are performing subtask 'pick up the marker from the cup and place it on the table.'\nWith start location [0.99, 0.03], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.99, 0.03], [0.86, 0.21], [0.76, 0.21], [0.72, 0.31], [0.74, 0.38], [0.7, 0.39], [0.7, 0.31], [0.61, 0.24], [0.62, 0.46], [0.7, 0.51]]}" + } + ], + "images": "488_exterior_image_1_left#0", + "gt": "[[318.0, 5.0], [274.0, 37.0], [242.0, 38.0], [230.0, 56.0], [236.0, 69.0], [224.0, 70.0], [224.0, 55.0], [195.0, 44.0], [200.0, 82.0], [225.0, 91.0]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "784_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pour the contents in the jug into the red bowl', and now you are performing subtask 'pour the contents in the jug into the red bowl.'\nYour initial coordinate is [0.02, 0.48]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.02, 0.48], [0.17, 0.62], [0.22, 0.71], [0.2, 0.72], [0.21, 0.81], [0.27, 0.67], [0.41, 0.71], [0.38, 0.6], [0.19, 0.74], [0.22, 0.79]]}" + } + ], + "images": "784_exterior_image_1_left#0", + "gt": "[[5, 86], [55, 112], [69, 128], [63, 129], [68, 145], [87, 120], [132, 127], [121, 108], [61, 133], [72, 142]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "69180_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open the first drawer from the right, remove the glue stick from it and put the glue stick on the counter', and now you are performing subtask 'pick up the glue stick from the first drawer on the right.'\nBegin at [0.48, 0.23] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.22], [0.43, 0.26], [0.39, 0.35], [0.43, 0.5], [0.48, 0.45], [0.42, 0.54], [0.4, 0.52], [0.39, 0.47], [0.4, 0.42], [0.42, 0.46]]}" + } + ], + "images": "69180_exterior_image_1_left#0", + "gt": "[[152.5, 40.5], [138.5, 47.5], [123.5, 62.5], [137.5, 90.5], [153.5, 81.5], [134.5, 97.5], [127.5, 93.5], [124.5, 84.5], [127.5, 75.5], [135.5, 83.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "76060_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'use the spatula to pick up the slice of bread', and now you are performing subtask 'use the spatula to pick up the slice of bread.'\nYour starting position is [0.9, 0.57]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.9, 0.57], [0.83, 0.45], [0.78, 0.47], [0.8, 0.56], [0.77, 0.58], [0.86, 0.43], [0.82, 0.33], [0.69, 0.37], [0.68, 0.52], [0.73, 0.4]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/76060_exterior_image_1_left#0.jpg", + "gt": "[[287.0, 102.5], [266.0, 81.5], [250.0, 85.5], [256.0, 101.5], [245.0, 104.5], [275.0, 76.5], [264.0, 59.5], [222.0, 66.5], [216.0, 93.5], [235.0, 71.5]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "76007_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up a bottle from the table and place it on the tray', and now you are performing subtask 'place the bottle on the tray.'\nStarting at [0.98, 0.0], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.98, 0.0], [0.93, 0.1], [0.94, 0.11], [0.93, 0.15], [0.88, 0.19], [0.78, 0.39], [0.73, 0.15], [0.74, 0.18], [0.66, 0.18], [0.59, 0.26]]}" + } + ], + "images": "76007_exterior_image_1_left#0", + "gt": "[[314, 0], [296, 18], [302, 20], [297, 27], [280, 35], [249, 70], [232, 27], [237, 33], [210, 32], [189, 47]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "83661_exterior_image_2_left#TrajectoryPrediction_Task#4", + "task": "TrajectoryPrediction_Task", + "p_index": 4, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'fold the cloth on the table and put the corn cob in the pot', and now you are performing subtask 'fold the cloth on the table again.'\nBegin at [0.62, 0.52] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.62, 0.52], [0.67, 0.58], [0.71, 0.58], [0.7, 0.54], [0.68, 0.54], [0.67, 0.58], [0.68, 0.43], [0.66, 0.48], [0.68, 0.35], [0.68, 0.35]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/83661_exterior_image_2_left#222.jpg", + "gt": "[[198, 93], [214, 104], [227, 105], [225, 97], [218, 98], [214, 105], [216, 78], [211, 86], [218, 63], [218, 63]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "67326_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'unfold the T-shirt on the table', and now you are performing subtask 'unfold the T-shirt on the table.'\nYour starting position is [0.7, 0.57]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.57], [0.71, 0.58], [0.68, 0.61], [0.65, 0.68], [0.62, 0.69], [0.63, 0.72], [0.6, 0.75], [0.59, 0.64], [0.52, 0.56], [0.49, 0.54]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/67326_exterior_image_2_left#0.jpg", + "gt": "[[224, 103], [228, 104], [216, 109], [208, 123], [198, 125], [201, 129], [193, 135], [188, 116], [167, 101], [157, 98]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "78246_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the lid from the pot and place it on the table', and now you are performing subtask 'remove the lid from the pot and place it on the table.'\nYour initial coordinate is [0.35, 0.1]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.35, 0.1], [0.36, 0.25], [0.3, 0.41], [0.3, 0.45], [0.3, 0.39], [0.32, 0.36], [0.4, 0.35], [0.44, 0.37], [0.49, 0.46], [0.5, 0.57]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/78246_exterior_image_1_left#0.jpg", + "gt": "[[112, 18], [114, 45], [95, 73], [95, 81], [96, 70], [103, 64], [127, 63], [142, 67], [157, 82], [159, 103]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0002_user_0010_scene_0001_cfg_0005_104422070044#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pull out a napkin', and now you are performing subtask 'pull out a napkin.'\nBegin at [0.58, 0.46] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.58, 0.46], [0.58, 0.48], [0.59, 0.48], [0.59, 0.42], [0.61, 0.39], [0.6, 0.38], [0.63, 0.34], [0.67, 0.36], [0.66, 0.39], [0.67, 0.42]]}" + } + ], + "images": "RH20T_cfg5_task_0002_user_0010_scene_0001_cfg_0005_104422070044#0", + "gt": "[[373, 166], [373, 172], [378, 174], [377, 152], [388, 139], [387, 135], [404, 124], [429, 129], [425, 141], [431, 150]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "17317_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the cup to the right', and now you are performing subtask 'move the cup to the right.'\nStarting at [0.48, 0.48], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.48, 0.48], [0.52, 0.66], [0.51, 0.69], [0.53, 0.71], [0.55, 0.77], [0.55, 0.79], [0.53, 0.78], [0.55, 0.74], [0.6, 0.77], [0.57, 0.68]]}" + } + ], + "images": "17317_exterior_image_1_left#0", + "gt": "[[154, 86], [165, 118], [162, 124], [170, 127], [175, 138], [175, 142], [171, 141], [175, 133], [193, 139], [182, 123]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "66082_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the cup forward and to the right', and now you are performing subtask 'move the cup forward and to the right.'\nStarting at [0.73, 0.03], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.73, 0.03], [0.63, 0.08], [0.56, 0.16], [0.55, 0.27], [0.51, 0.36], [0.55, 0.26], [0.54, 0.21], [0.51, 0.25], [0.51, 0.19], [0.51, 0.19]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/66082_exterior_image_1_left#0.jpg", + "gt": "[[235, 6], [203, 15], [180, 29], [177, 49], [164, 65], [176, 46], [172, 38], [163, 45], [164, 34], [164, 34]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "21793_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'turn the caps on the bed', and now you are performing subtask 'pick up the black hat on the bed.'\nWith start location [0.17, 0.84], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.17, 0.84], [0.21, 0.83], [0.29, 0.74], [0.32, 0.73], [0.32, 0.64], [0.32, 0.64], [0.33, 0.64], [0.32, 0.64], [0.32, 0.63], [0.33, 0.62]]}" + } + ], + "images": "21793_exterior_image_1_left#122", + "gt": "[[54, 152], [68, 149], [93, 134], [103, 132], [104, 116], [102, 115], [106, 115], [102, 116], [101, 114], [106, 112]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "32876_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pour the contents from the left bowl into the right bowl', and now you are performing subtask 'place the spoon on the table.'\nYour starting position is [0.18, 0.4]. Given the current scene, predict 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.18, 0.4], [0.19, 0.46], [0.25, 0.51], [0.27, 0.56], [0.28, 0.49], [0.24, 0.45], [0.24, 0.49], [0.25, 0.48], [0.26, 0.52], [0.26, 0.48]]}" + } + ], + "images": "32876_exterior_image_2_left#0", + "gt": "[[59, 72], [60, 83], [80, 91], [85, 100], [88, 88], [78, 81], [77, 88], [81, 86], [82, 93], [82, 86]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "69345_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'take the cloth from the clear screen and put it on the table', and now you are performing subtask 'take the cloth from the clear screen and put it on the table.'\nYou start from [0.67, 0.23]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.67, 0.23], [0.64, 0.26], [0.6, 0.26], [0.61, 0.26], [0.6, 0.28], [0.52, 0.33], [0.52, 0.37], [0.58, 0.41], [0.58, 0.44], [0.59, 0.42]]}" + } + ], + "images": "69345_exterior_image_1_left#0", + "gt": "[[215, 42], [206, 47], [193, 46], [195, 47], [193, 51], [166, 60], [166, 66], [187, 73], [186, 80], [189, 76]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg4_task_0094_user_0007_scene_0009_cfg_0004_104122062823#TrajectoryPrediction_Task#6", + "task": "TrajectoryPrediction_Task", + "p_index": 6, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'open a box', and now you are performing subtask 'push the box to open the box, push the box to open the box.'\nYou start from [0.42, 0.11]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.42, 0.11], [0.45, 0.0], [0.47, 0.0], [0.45, 0.0], [0.45, 0.07], [0.47, 0.0], [0.52, 0.03], [0.52, 0.0], [0.5, 0.0], [0.53, 0.0]]}" + } + ], + "images": "RH20T_cfg4_task_0094_user_0007_scene_0009_cfg_0004_104122062823#253", + "gt": "[[267, 41], [287, 0], [298, 0], [286, 0], [290, 25], [298, 0], [331, 9], [335, 0], [321, 0], [342, 0]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "27910_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'put the orange cube on the green cube', and now you are performing subtask 'put the orange cube on the green cube.'\nYour initial coordinate is [0.07, 0.46]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.07, 0.46], [0.2, 0.48], [0.25, 0.53], [0.3, 0.65], [0.34, 0.6], [0.36, 0.6], [0.37, 0.61], [0.36, 0.66], [0.36, 0.66], [0.35, 0.64]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/27910_exterior_image_2_left#0.jpg", + "gt": "[[22, 82], [65, 87], [80, 95], [96, 117], [109, 108], [116, 108], [118, 110], [116, 119], [114, 119], [112, 116]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg5_task_0013_user_0007_scene_0001_cfg_0005_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'place the block on the scale', and now you are performing subtask 'place the block on the scale.'\nFrom start point [0.45, 0.3], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.45, 0.3], [0.36, 0.3], [0.28, 0.36], [0.28, 0.43], [0.29, 0.4], [0.32, 0.38], [0.34, 0.4], [0.44, 0.41], [0.45, 0.44], [0.43, 0.31]]}" + } + ], + "images": "RH20T_cfg5_task_0013_user_0007_scene_0001_cfg_0005_036422060215#0", + "gt": "[[286, 107], [232, 109], [177, 131], [178, 155], [183, 143], [205, 136], [218, 143], [279, 147], [288, 160], [276, 113]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg7_task_0031_user_0020_scene_0007_cfg_0007_104122061602#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'cover the box', and now you are performing subtask 'cover the box.'\nStarting at [0.7, 0.32], infer 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.7, 0.32], [0.72, 0.44], [0.77, 0.51], [0.78, 0.6], [0.79, 0.52], [0.78, 0.42], [0.7, 0.31], [0.67, 0.33], [0.71, 0.29], [0.7, 0.26]]}" + } + ], + "images": "RH20T_cfg7_task_0031_user_0020_scene_0007_cfg_0007_104122061602#0", + "gt": "[[448, 116], [462, 158], [496, 185], [497, 217], [507, 187], [499, 151], [445, 111], [431, 120], [455, 103], [449, 95]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "86272_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the white bowl and place the white bowl on the right side of the table', and now you are performing subtask 'pick up the white bowl and place the white bowl on the right side of the table.'\nFrom start point [0.76, 0.0], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.76, 0.0], [0.66, 0.24], [0.64, 0.39], [0.7, 0.28], [0.8, 0.27], [0.89, 0.34], [0.88, 0.44], [0.82, 0.53], [0.86, 0.38], [0.8, 0.1]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/86272_exterior_image_1_left#0.jpg", + "gt": "[[242, 0], [211, 44], [205, 70], [224, 51], [257, 49], [284, 62], [283, 79], [262, 96], [274, 69], [255, 18]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "18908_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'remove the marker from the mug and place it on the table', and now you are performing subtask 'remove the marker from the mug and place it on the table.'\nFrom start point [0.88, 0.0], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.88, 0.0], [0.87, 0.27], [0.82, 0.35], [0.82, 0.61], [0.86, 0.29], [0.79, 0.28], [0.72, 0.43], [0.76, 0.3], [0.77, 0.12], [0.77, 0.12]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/18908_exterior_image_1_left#0.jpg", + "gt": "[[281, 0], [277, 49], [262, 63], [261, 109], [274, 53], [252, 51], [230, 77], [244, 54], [246, 22], [246, 22]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "54044_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up and put down the green block, then put the salt cellar into the pot and the green block into the sink', and now you are performing subtask 'place the green block on the countertop.'\nThe trajectory starts at [0.34, 0.0]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.34, 0.0], [0.42, 0.13], [0.44, 0.2], [0.44, 0.36], [0.41, 0.26], [0.44, 0.26], [0.45, 0.29], [0.42, 0.37], [0.46, 0.39], [0.44, 0.34]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/54044_exterior_image_2_left#0.jpg", + "gt": "[[108, 0], [135, 24], [141, 36], [141, 64], [130, 47], [140, 47], [143, 53], [134, 66], [146, 70], [142, 62]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg6_task_0120_user_0020_scene_0005_cfg_0006_104122064161#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'installing the light bulb by rotating it', and now you are performing subtask 'installing the light bulb by rotating it.'\nFrom start point [0.53, 0.31], estimate 10 trajectory points based on the current scene to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.31], [0.63, 0.47], [0.63, 0.54], [0.64, 0.46], [0.62, 0.41], [0.63, 0.35], [0.65, 0.36], [0.65, 0.4], [0.66, 0.38], [0.61, 0.29]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg6_task_0120_user_0020_scene_0005_cfg_0006_104122064161#0.jpg", + "gt": "[[338, 111], [404, 168], [404, 195], [410, 165], [398, 146], [403, 125], [415, 130], [414, 143], [422, 136], [393, 106]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg2_task_0128_user_0007_scene_0003_cfg_0002_036422060215#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'plug in the USB drive to the port', and now you are performing subtask 'plug in the USB drive to the port.'\nThe trajectory starts at [0.46, 0.45]. Considering the current scene, return 10 trajectory points to achieve the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.46, 0.45], [0.38, 0.38], [0.36, 0.38], [0.35, 0.43], [0.35, 0.37], [0.38, 0.34], [0.43, 0.39], [0.42, 0.42], [0.44, 0.39], [0.44, 0.44]]}" + } + ], + "images": "RH20T_cfg2_task_0128_user_0007_scene_0003_cfg_0002_036422060215#0", + "gt": "[[294, 162], [242, 135], [232, 135], [224, 156], [223, 133], [241, 124], [273, 140], [268, 152], [279, 139], [283, 158]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "20353_exterior_image_1_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'close the lid on the clear container', and now you are performing subtask 'close the lid on the clear container.'\nBegin at [0.77, 0.28] (trajectory start). From the scene, output 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.77, 0.28], [0.65, 0.52], [0.61, 0.65], [0.64, 0.66], [0.6, 0.75], [0.59, 0.7], [0.7, 0.57], [0.74, 0.63], [0.68, 0.5], [0.68, 0.5]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/20353_exterior_image_1_left#0.jpg", + "gt": "[[247, 51], [208, 94], [195, 117], [205, 118], [191, 135], [189, 126], [225, 102], [238, 113], [219, 90], [219, 90]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg7_task_0037_user_0020_scene_0006_cfg_0007_104122061018#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'stack the blocks in a vertical line of five', and now you are performing subtask 'pick up the first block from the table, transfer the first block from the table to above the second block, pick up the third block from the table.'\nYou begin at [0.57, 0.09]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.57, 0.09], [0.56, 0.16], [0.57, 0.26], [0.54, 0.45], [0.52, 0.36], [0.5, 0.34], [0.44, 0.36], [0.45, 0.29], [0.43, 0.24], [0.42, 0.25]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg7_task_0037_user_0020_scene_0006_cfg_0007_104122061018#0.jpg", + "gt": "[[363, 31], [359, 57], [365, 92], [346, 162], [333, 130], [321, 121], [280, 130], [285, 103], [278, 88], [267, 91]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "RH20T_cfg4_task_0045_user_0014_scene_0007_cfg_0004_104122062823#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the brush from the table and place it on the pen rack', and now you are performing subtask 'pick up the brush from the table and place it on the pen rack.'\nYour initial coordinate is [0.38, 0.0]. Using the scene context, generate 10 trajectory points to finish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.38, 0.0], [0.36, 0.0], [0.4, 0.13], [0.38, 0.19], [0.36, 0.21], [0.4, 0.24], [0.49, 0.0], [0.67, 0.0], [0.57, 0.0], [0.57, 0.0]]}" + } + ], + "images": "RH20T_cfg4_task_0045_user_0014_scene_0007_cfg_0004_104122062823#0", + "gt": "[[243, 0], [229, 1], [253, 47], [245, 67], [233, 74], [258, 87], [316, 0], [430, 0], [367, 0], [367, 0]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "59697_exterior_image_1_left#TrajectoryPrediction_Task#2", + "task": "TrajectoryPrediction_Task", + "p_index": 2, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the objects from the orange bowl to the black pot, then cover the black pot with the glass lid', and now you are performing subtask 'place the item in the black pot.'\nYou begin at [0.53, 0.29]. Given the scene as observed, supply 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.53, 0.29], [0.55, 0.26], [0.58, 0.27], [0.58, 0.43], [0.6, 0.37], [0.6, 0.31], [0.57, 0.26], [0.52, 0.28], [0.52, 0.37], [0.53, 0.27]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/59697_exterior_image_1_left#209.jpg", + "gt": "[[170, 52], [177, 46], [186, 49], [187, 77], [192, 66], [191, 55], [183, 47], [167, 51], [168, 66], [170, 49]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "50112_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'move the cup slightly down on the table', and now you are performing subtask 'move the cup slightly down on the table.'\nYou start from [0.39, 0.04]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.39, 0.04], [0.5, 0.04], [0.58, 0.08], [0.57, 0.12], [0.61, 0.24], [0.57, 0.34], [0.6, 0.46], [0.57, 0.45], [0.54, 0.48], [0.55, 0.37]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/50112_exterior_image_2_left#0.jpg", + "gt": "[[126, 7], [161, 8], [185, 15], [184, 22], [196, 43], [184, 61], [192, 83], [182, 81], [174, 86], [177, 66]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + }, + { + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_038522063145#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'putting a portion of the Lego blocks on the Lego baseplate', and now you are performing subtask 'pick up the red Lego brick from the table, transfer the red Lego brick from the table to the Lego base, insert the red Lego brick into the Lego base.'\nWith start location [0.69, 0.27], and given the scene, produce 10 trajectory points to complete the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.69, 0.27], [0.84, 0.42], [0.86, 0.5], [0.86, 0.47], [0.84, 0.5], [0.86, 0.48], [0.86, 0.44], [0.78, 0.41], [0.75, 0.47], [0.79, 0.4]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_038522063145#0.jpg", + "gt": "[[441, 97], [536, 150], [551, 179], [553, 169], [540, 181], [550, 172], [550, 158], [499, 148], [482, 170], [506, 144]]", + "h": 360, + "w": 640, + "new_h": 364, + "new_w": 644 + }, + { + "id": "14360_exterior_image_2_left#TrajectoryPrediction_Task#0", + "task": "TrajectoryPrediction_Task", + "p_index": 0, + "conversations": [ + { + "from": "human", + "value": "\nThe robot task is 'pick up the object from the sink and put it on the counter', and now you are performing subtask 'pick up the object from the sink and put it on the counter.'\nYou start from [0.67, 0.51]. Using the present scene, provide 10 trajectory points to accomplish the task.\nReturn in JSON format: The trajectory is a list of tuples, i.e. [[x1, y1], [x2, y2], w...], where each tuple contains the x and y coordinates of a 2d point." + }, + { + "from": "gpt", + "value": "{\"future_traj\": [[0.67, 0.51], [0.7, 0.73], [0.74, 0.81], [0.71, 0.84], [0.69, 0.72], [0.67, 0.71], [0.62, 0.75], [0.59, 0.84], [0.6, 0.78], [0.66, 0.76]]}" + } + ], + "images": "Generation/image/val/traj_init_point_qa/14360_exterior_image_2_left#0.jpg", + "gt": "[[213, 91], [225, 132], [236, 145], [228, 152], [220, 130], [213, 128], [199, 135], [188, 152], [191, 141], [211, 137]]", + "h": 180, + "w": 320, + "new_h": 168, + "new_w": 308 + } +] \ No newline at end of file