han-xudong/fingernet-img-100k

FingerNet-Img-40k

This dataset contains about 40,000 samples of image and force data collected for soft robotic fingers interacting with various objects both on land and underwater.

Dataset Schema

The dataset is organized as follows:

fingernet-img-40k/
├── onland/
│   ├── data_0.parquet
│   ├── data_1.parquet
│   ├── data_2.parquet
│   └── ...
└── underwater/
    ├── data_0.parquet
    ├── data_1.parquet
    ├── data_2.parquet
    └── ...

Each record in contains:

Field	Type	Description
image	Image	Image containing finger structure, 640x480 pixels
Fx, Fy, Fz, Mx, My, Mz	float32	6D force-torque data in N and Nmm
object	string	Contacted object name
env	string	Environment (onland / underwater)

Usage

To read the dataset, you can use the following code:

from datasets import load_dataset

# Load the onland dataset
ds = load_dataset("asRobotics/magiclaw-touch-dataset", "onland")
print(ds[0])

# Load the underwater dataset
ds = load_dataset("asRobotics/magiclaw-touch-dataset", "underwater")
print(ds[0])

Citation

If you use this dataset in your research, please cite the following paper:

@article{guo2024autoencoding,
  title={Autoencoding a Soft Touch to Learn Grasping from On-Land to Underwater},
  author={Guo, Ning and Han, Xudong and Liu, Xiaobo and Zhong, Shuqiao and Zhou, Zhiyuan and Lin, Jian and Dai, Jiansheng and Wan, Fang and Song, Chaoyang},
  journal={Advanced Intelligent Systems},
  volume={6},
  number={1},
  pages={2300382},
  year={2024},
  publisher={Wiley Online Library},
  doi = {10.1002/aisy.202300382}
}