THULab/DanqingZ_pick_yellow_3

Pick Yellow 3 TsFile

This Apache TsFile edition comes from DanqingZ/pick_yellow_3, a LeRobot v2.1 SO100 robot-manipulation dataset. It contains numeric robot trajectories, frame timing, and episode/task tags. Camera videos remain in the original repository.

Source Dataset and Attribution

• Original dataset: DanqingZ/pick_yellow_3
• Original author/uploader: DanqingZ
• Authorship evidence: the repository is owned by DanqingZ. The source card gives no separate author list.
• License: Apache-2.0
• Task: "Grasp the yellow cuboid and put it in the bin."
• Robot: so100; LeRobot version: v2.1
• Split: train; sampling rate: 30 fps
• Scale: 50 episodes, 29,715 frames, 1 task
• Source shards: 50 Parquet files in one chunk
• Source frame layout: data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet

TsFile Data

• Path: data/danqingz_pick_yellow_3.tsfile (344,938 bytes)
• Table: danqingz_pick_yellow_3
• Rows: 29,715; episodes/devices: 50; tasks: 1
• Time precision: milliseconds
• Source metadata remains under meta/; meta/info.json points to the TsFile.

Schema

Time = round(timestamp * 1000) milliseconds. Time restarts at zero in every episode. Source timestamp is omitted afterward because it equals Time / 1000 seconds.

TsFile column	Source	Type	Role	Notes
Time	timestamp	TIMESTAMP	TIME	Milliseconds, zero-based per episode
episode_index	same	STRING	TAG	Source episode identity
task_index	same	STRING	TAG	Source task identity
frame_index	same	INT64	FIELD	Frame number within the episode
sample_index	index	INT64	FIELD	Global sample number
action_0 ... action_5	action[6]	FLOAT	FIELD	Flattened action vector
observation_state_0 ... observation_state_5	observation.state[6]	FLOAT	FIELD	Flattened state vector

The six vector dimensions are main_shoulder_pan, main_shoulder_lift, main_elbow_flex, main_wrist_flex, main_wrist_roll, and main_gripper. Dots in source vector names become underscores. No numeric row, episode, task, or vector dimension is dropped.

Videos

Videos are not included in this TsFile repository. The source has 150 frame-aligned AV1 MP4 files (785,221,304 bytes, about 748.8 MiB), 640x480, 30 fps, no audio, in three streams:

• observation.images.on_robot
• observation.images.phone
• observation.images.side_view

Each stream uses videos/chunk-000/{video_key}/episode_{episode_index:06d}.mp4. episode_index, frame_index, and meta/episodes.jsonl preserve alignment.

Usage

from tsfile import TsFileReader

reader = TsFileReader("data/danqingz_pick_yellow_3.tsfile")
with reader.query_table(
    "danqingz_pick_yellow_3",
    ["episode_index", "task_index", "frame_index", "sample_index",
     "action_0", "observation_state_0"],
    batch_size=65536,
) as result:
    print(result.read_arrow_batch().to_pandas().head())
reader.close()

Source Notes

The source card states that the dataset was created with LeRobot. It does not provide a paper, homepage, or formal citation.