README.md

---
license: cc-by-4.0
task_categories:
  - robotics
tags:
  - robotics
  - tactile
  - manipulation
  - multimodal
  - gelsight
  - realsense
  - motion-capture
  - dynamics
  - world-model
  - human-collected
pretty_name: React (Tactile-Visual Manipulation)
size_categories:
  - 100K<n<1M
configs:
  - config_name: episode_metadata
    data_files:
      - split: train
        path: metadata/episodes.parquet
  - config_name: motherboard
    data_files:
      - split: train
        path: processed/mode1_v1/motherboard/**/episode_*.pt
  - config_name: motherboard_segments
    data_files:
      - split: train
        path: processed/mode2_v1/motherboard/**/episode_*.segment_*.pt
  - config_name: all
    data_files:
      - split: train
        path: processed/mode1_v1/**/episode_*.pt
---

# React

Dense, contact-rich, synchronized multimodal interaction data collected from **human hands holding handheld GelSight tactile sensors — no robot arm involved**. Intended for **tactile-visual dynamics / world-model learning**, *not* a policy / demonstration dataset.

![Tactile intensity timeline](figures/contact_intensity_full.png)

> **106 min of robot-free human-hand multimodal interaction · 190,231 frames @ 30 Hz across 3 × RGB-D + 2 × GelSight + 3-body OptiTrack**

## What's different about this dataset

| | |
|---|---|
| **Robot-arm-free** | Recorded directly from a human operator holding two GelSight Mini sensors. No robot kinematics, no embodiment bias, no robot occluding the scene. |
| **Tactile + RGB-D + mocap, simultaneous** | Most manipulation datasets ship one of these. React ships all three, synchronized to a common 30 Hz clock. |
| **Contact-dense** | **64 % of post-trim frames** have confirmed tactile contact on at least one sensor — see [`figures/contact_intensity_full.png`](figures/contact_intensity_full.png). |
| **Long, continuous interaction** | Recordings are minutes long, not seconds. Median recording duration is 4 min; longest 19 min. Good for short-window sampling of dynamics, not for action-conditioned policy learning. |

![Comparison with other manipulation datasets](figures/dataset_figures/F7_comparison_table.png)

## At a glance

| | |
|---|---|
| Embodiment | **Human hands (no robot)** — handheld GelSight sensors with motion-capture rigid bodies |
| Intended use | Dynamics / world-model learning over short multimodal windows. Sample short trajectories (1 s – 10 s); recording-file boundaries are not action boundaries. |
| Total synchronized duration | **105.7 min** at 30 Hz (190,231 multimodal frames, post-trim) |
| Bimanual tactile-contact time | **64.3 % of post-trim frames** (3,302 contact events, median 0.73 s; see [`figures/dataset_figures/F2_contact_event_duration_histogram.png`](figures/dataset_figures/F2_contact_event_duration_histogram.png) and [`metadata/episodes.parquet`](metadata/episodes.parquet) for per-file numbers) |
| Cameras | 3× Intel RealSense D415 (color + depth), 480×640, 30 FPS |
| Tactile | 2× GelSight Mini (left, right), handheld |
| Motion capture | OptiTrack VRPN, 3 rigid bodies, ~120 Hz |
| Tasks | `motherboard` (more coming) |
| License | CC-BY-4.0 |

## Recording sessions

| Date | Kind | Active sensors | Notes |
|---|---|---|---|
| 2026-05-10 | session | left + right | First full bimanual session. |
| 2026-05-11 | session | left + right | Largest session. A handful of GelSight LED-flicker frames + one mocap teleport; see [`bad_frames.json`](bad_frames.json). |

See [`tasks.json`](tasks.json) for the machine-readable registry (per-date `active_sensors`, etc.).

**OT-uninitialized prefixes trimmed.** Three episodes had OptiTrack offline at the start of recording (1–11 min each); those prefixes have been cut from the published `.pt` files (`_contact_meta.trim_offset` per file). Future recordings use an OT watchdog that refuses to start an episode unless mocap is streaming. Full story: [`docs/caveats.md`](docs/caveats.md).

## Data quality

| Mode | Frames | % | Files | Cause |
|---|---:|---:|---:|---|
| GelSight LED flicker | 56 | 0.029 % | 5 | Single-frame LED dropout, recovers next frame |
| OptiTrack pose teleport | 56 | 0.029 % | 3 | Solver flip (translation > 5 m/s or angular > 15 rad/s) |
| OptiTrack track loss | 1,680 | 0.883 % | 6 | Marker briefly left mocap-volume / camera FOV mid-episode |
| **Total (union)** | **1,768** | **0.929 %** | **11** | |

Every flagged interval is in [`bad_frames.json`](bad_frames.json) keyed by `episode/episode_*` with TRIMMED-pt frame indices. A richer per-event view (with cross-modal motion + OT-gap + angular-velocity stats) lives in [`freeze_intervals.json`](freeze_intervals.json). Skip-list usage is shown below and in [`docs/quality.md`](docs/quality.md). Long start-of-episode OT-uninitialized prefixes (the dominant problem in the raw recordings) have already been trimmed from the published `.pt` files — see [`docs/caveats.md`](docs/caveats.md).

## Two schemas: mode1_v1 vs mode2_v1

The same recordings are shipped two ways depending on what your code wants to do:

- **`processed/mode1_v1/<task>/<date>/episode_*.pt`** — one file per recording. Includes bad intervals (LED flicker, pose teleport, OT track loss) inside; downstream code is expected to filter them out using `bad_frames.json`. This is what the example `ReactWindowDataset` consumes.
- **`processed/mode2_v1/<task>/<date>/episode_*.segment_*.pt`** — same recordings, but **pre-sliced into contiguous clean segments at every bad-frames boundary**. No `bad_frames.json` lookup needed; the data is clean by construction. Index lookup via [`segments.json`](segments.json). Each segment's `_contact_meta.source_h5_frame_range` maps it back to the original recording. The example `ReactSegmentDataset` ([`examples/react_segment_dataset.py`](examples/react_segment_dataset.py)) consumes these.

Both schemas have identical content (same source recordings, same frame data); only the file boundaries differ. `mode2_v1` produces 73 segments from 27 source episodes; **104.7 min total** (vs. mode1_v1's 105.7 min — the 1-min delta is the bad frames cut at segment boundaries).

## Quick start

```python
# Load by task with `datasets`
from datasets import load_dataset
ds = load_dataset("yxma/React", "motherboard", split="train")
```

Or grab a single recording file directly:

```python
import torch
from huggingface_hub import hf_hub_download

path = hf_hub_download(
    repo_id="yxma/React", repo_type="dataset",
    filename="processed/mode1_v1/motherboard/2026-05-11/episode_003.pt",
)
ep = torch.load(path, weights_only=False)
# ep["view"]                                 (T, 3, 128, 128) uint8 — overhead cam
# ep["tactile_left"], ep["tactile_right"]    (T, 3, 128, 128) uint8
# ep["sensor_left_pose"], ep["sensor_right_pose"]
#                                            (T, 7) float32 — xyz + quaternion
# ep["timestamps"]                           (T,)  float64
# Plus per-frame contact metrics:            tactile_{side}_{intensity, area, mixed}
```

Sampling short windows for dynamics learning — **drop windows that overlap any flagged interval**:

```python
import json

with open("bad_frames.json") as f:
    bad = json.load(f)["episodes"]   # frame indices are TRIMMED-pt coordinates

def is_clean_window(episode_key, t_start, t_end):
    """True iff [t_start, t_end] doesn't intersect any flagged span."""
    bf = bad[episode_key]
    intervals = (bf["intensity_spikes"]
                 + bf["pose_teleports_L"] + bf["pose_teleports_R"]
                 + bf["ot_loss_L"] + bf["ot_loss_R"])
    return all(not (s <= t_end and e >= t_start) for s, e in intervals)
```

Currently 1,768 / 190,231 frames (0.93 %) are flagged across 11 of 27 files — see [`docs/quality.md`](docs/quality.md) for the per-mode breakdown and more filtering recipes. The example dataloader below does this filtering for you when `skip_bad_frames=True`.

## Example dataloader — short contact-rich windows

A reference PyTorch `Dataset` is shipped under [`examples/react_window_dataset.py`](examples/react_window_dataset.py). It scans the processed `.pt` files, applies the contact filter, drops windows that overlap [`bad_frames.json`](bad_frames.json), and respects the per-date `active_sensors` field from [`tasks.json`](tasks.json).

```python
from examples.react_window_dataset import ReactWindowDataset
from torch.utils.data import DataLoader

ds = ReactWindowDataset(
    data_root="processed/mode1_v1/motherboard",
    bad_frames_path="bad_frames.json",
    tasks_json_path="tasks.json",
    window_length=16,            # frames per window
    stride=1,                    # within-window stride (1 = consecutive)
    window_step=16,              # step between window starts (overlap control)
    contact_metric="mixed",      # "intensity" | "area" | "mixed"
    tactile_threshold=0.4,
    min_contact_fraction=0.6,    # ≥ 60 % of window frames must have contact
    which_sensors="any",         # "any" | "both" | "left" | "right"
    skip_bad_frames=True,
    respect_active_sensors=True,
)
print(len(ds), "windows")
loader = DataLoader(ds, batch_size=8, shuffle=True, num_workers=2)
```

With the defaults shown above, the dataset assembles **~9.2 k contact-rich 16-frame windows** across the 27 recordings. Each sample is a dict of `(T, …)` tensors plus metadata (`episode`, `frame_start`, `active_sensors`, …).

### Example output

Four random windows, time runs left→right; each cell is `view | tactile_left | tactile_right` with sensor frame axes (X red, Y green, Z blue-ish) projected onto the view:

![dataloader sample grid](figures/dataloader_examples/sample_grid.png)

One window played frame-by-frame with the sensor-frame overlay:

![dataloader sample GIF](figures/dataloader_examples/sample_window.gif)

Full demo script: [`examples/demo_react_window.py`](examples/demo_react_window.py).

## Recording-file previews

Per-file previews live under [`figures/episode_previews/`](figures/episode_previews) as both `.gif` and `.mp4` (MP4s render inline on HF and are ~30× smaller). Each shows 60 frames evenly sampled across the episode in the recording-viewer layout: 3 RealSense cameras with projected GelSight axes, GelSight raw + diff thumbs, OptiTrack pose text panel. (The on-disk recording unit is called an "episode" purely for file naming — these boundaries don't carry semantic / action meaning for this dataset.)

## Repository layout

```
README.md                                        # this file
tasks.json                                       # task / session registry
bad_frames.json                                  # data-quality skip-list
processed/mode1_v1/<task>/<date>/episode_*.pt    # per-file tensors
figures/                                         # previews + analysis figures
docs/                                            # extended documentation
```

## More documentation

| File | Contents |
|---|---|
| [`docs/recording.md`](docs/recording.md) | Hardware setup, camera serials, sensor + mocap layout, robot-free collection method |
| [`docs/schema.md`](docs/schema.md) | Full `.pt` field reference and contact-metric definitions |
| [`docs/quality.md`](docs/quality.md) | Data-quality breakdown (per-mode), `bad_frames.json` schema, dataloader recipe, inspection figures |
| [`docs/figures.md`](docs/figures.md) | Dataset statistics + analysis gallery (F1–F8) |
| [`docs/caveats.md`](docs/caveats.md) | Known caveats and roadmap |

## License

Released under [Creative Commons Attribution 4.0](https://creativecommons.org/licenses/by/4.0/) (CC-BY-4.0).

## Citation

If you use this dataset, please cite (TODO: add bibtex).