Datasets:

THULab
/

320-Quarter_Brain-lego_picking

A LeRobot teleoperation dataset recorded on an SO-101 follower arm performing a Lego pick-and-place task: "Grab the Lego and store it in red box." Each episode captures the synchronized robot joint state and commanded action at every control step while the arm picks up a Lego brick and places it in a red box.

Robot: so101_follower (6-DoF arm: shoulder pan/lift, elbow flex, wrist flex/roll, gripper).
Episodes: 46 (single train split).
Frames: 23,041 total.
Sampling rate: 30 fps.
Tasks: 1 — "Grab the Lego and store it in red box".

Schema (TsFile structure)

Time (INT64, milliseconds) — per-frame timestamp, restarts at 0 each episode (round(timestamp * 1000)).
episode_index (TAG) — episode id; query one episode with WHERE episode_index=0.
task_index (TAG) — task id (always 0 here).
frame_index (FIELD, INT64) — frame number within the episode.
sample_index (FIELD, INT64) — global sample index (the source index column).
action_0 … action_5 (FIELD, FLOAT) — commanded joint positions (shoulder_pan, shoulder_lift, elbow_flex, wrist_flex, wrist_roll, gripper).
observation_state_0 … observation_state_5 (FIELD, FLOAT) — measured joint positions, same joint order as action.

Fixed-size vector columns from the source (action, observation.state, each length 6) are flattened to single-precision FLOAT columns; . becomes _ and the element index is appended. The source timestamp column is not retained as a separate field because it equals Time / 1000 seconds.

Usage

Read the .tsfile files with the Apache TsFile Java or Python SDK.

Camera videos (observation.images.side, .top, .wrist) are NOT included in this repository. They remain in the original dataset under videos/.

Source & license

Original dataset: https://huggingface.co/datasets/LeRobot-worldwide-hackathon/320-Quarter_Brain-lego_picking
Author / publisher: LeRobot Worldwide Hackathon
License: apache-2.0

Downloads last month: 54

Number of rows:

Total file size:

505 kB