THULab/BobBobbson_so100_First_Gripper_Test

SO100 First Gripper Test TsFile

This is a TsFile edition of BobBobbson/so100_First_Gripper_Test, a LeRobot v2.1 SO100 robot dataset containing numeric trajectories, timing, and episode/task tags. Camera videos remain in the original Hugging Face repository.

Source and Attribution

• Original dataset: BobBobbson/so100_First_Gripper_Test
• Creator, repository owner, uploader, and commit author: BobBobbson
• License: Apache-2.0
• Task: "Pick up cube and place in grey container, use gripper visualisation feed."
• Robot: so100; LeRobot version: v2.1
• Split and frequency: train, 30 fps
• Scale: 30 episodes, 28,889 frames, 1 task, 30 Parquet shards
• Source layout: data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet
• The source card provides no paper, homepage, or completed citation.

TsFile Data

• Path: data/bobbobbson_so100_first_gripper_test.tsfile
• Table: bobbobbson_so100_first_gripper_test
• Rows: 28,889; episodes/devices: 30; tasks: 1
• Time precision: milliseconds

Schema and Mapping

Time = round(timestamp * 1000) milliseconds. Time starts at zero and is strictly increasing in every episode. The source timestamp is omitted because it is recoverable as Time / 1000 seconds.

TsFile column	Role	Type	Source mapping
Time	TIME	TIMESTAMP	round(timestamp * 1000) ms
episode_index	TAG	STRING	Original INT64 episode index
task_index	TAG	STRING	Original INT64 task index
frame_index	FIELD	INT64	Preserved
sample_index	FIELD	INT64	Renamed from index
action_0 ... action_5	FIELD	FLOAT	Flattened from action[6]
observation_state_0 ... observation_state_5	FIELD	FLOAT	Flattened from observation.state[6]

The vector order is main_shoulder_pan, main_shoulder_lift, main_elbow_flex, main_wrist_flex, main_wrist_roll, and main_gripper. No numeric row, episode, task, action dimension, or state dimension is dropped.

Videos

Videos are not included here. The 90 original AV1 MP4 files are 640x360 at 30 fps with no audio, with 30 episode files in each stream:

• observation.images.OBS_Gripper_Viz
• observation.images.overhead
• observation.images.wrist

The source layout is videos/chunk-{episode_chunk:03d}/{video_key}/episode_{episode_index:06d}.mp4. Use episode_index and frame_index to align numeric rows with video frames.

Minimal Read Example

from tsfile import TsFileReader

reader = TsFileReader("data/bobbobbson_so100_first_gripper_test.tsfile")
with reader.query_table(
    "bobbobbson_so100_first_gripper_test",
    ["episode_index", "task_index", "frame_index", "sample_index",
     "action_0", "observation_state_0"],
    batch_size=65536,
) as result:
    batch = result.read_arrow_batch()
    print(batch.to_pandas().head())
reader.close()

Citation

The source card supplies no paper or completed citation. Cite the original Hugging Face dataset and BobBobbson when using this edition.