THULab/put_drivers_in_bin_lerobot

Put Drivers In Bin (TsFile)

Apache TsFile version of IMsubin/put_drivers_in_bin.

Overview

A robot-arm manipulation dataset in the LeRobot v2.1 format, recorded on an omx_f arm. The single task is to put drivers (screwdrivers) into a bin. Each record is one control frame within a demonstration episode, holding the arm's joint state and the commanded joint action.

• Scale: 256 episodes, 244,580 frames total (≈ 955 frames per episode on average).
• Sampling rate: 30 fps.
• Tasks: 1 — "put drivers in bin".
• Robot: omx_f arm (5 joints + gripper).

Note: the source data/ directory contains 256 episodes, while the source meta/ files only describe the first 50 (info.json reports 50 episodes / 50,989 frames). All 256 episodes are included in this TsFile conversion.

Schema (TsFile structure)

• Time (INT64, milliseconds) — round(timestamp * 1000), restarting at 0 for each episode.
• episode_index (TAG) — source episode id (0–255).
• task_index (TAG) — source task id (always 0 here).
• frame_index (FIELD, INT64) — frame position within the episode.
• sample_index (FIELD, INT64) — the source global index column, renamed.
• observation_state_0 .. observation_state_5 (FIELD, FLOAT) — the 6-D joint state (joint1–5 + gripper_joint_1), flattened.
• action_0 .. action_5 (FIELD, FLOAT) — the 6-D joint action (joint1–5 + gripper_joint_1), flattened.

All 256 episodes share a single .tsfile; episode_index and task_index are the device (TAG) dimensions. Query one episode with WHERE episode_index=0.

Vector columns are flattened keeping the source name (observation.state → observation_state_0..5, action → action_0..5). The source timestamp column is dropped because it equals Time / 1000 seconds; frame_index is kept.

Usage

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

Source & license

• Original dataset: https://huggingface.co/datasets/IMsubin/put_drivers_in_bin
• Author / publisher: IMsubin (Hugging Face)
• Camera videos (observation.images.wrist_cam, observation.images.top_cam, 480×640 RGB @ 30 fps) are not included in this repository; see the original dataset's videos/ directory for them.
• License: not declared by the original dataset; please defer to the original.