THULab/Jeevesh2009_so101_gray_block_lowvar_test

SO101 Gray Block Lowvar Test TsFile

Apache TsFile edition of Jeevesh2009/so101_gray_block_lowvar_test, a LeRobot v2.1 SO101 dataset for picking up a gray block and placing it in a box. The numeric trajectories are stored in one table-model TsFile.

Source and attribution

• —Original author and repository owner: Madala Venkata Renu Jeevesh (Jeevesh2009).
• —License: Apache-2.0.
• —The source card does not provide a homepage, paper, or completed BibTeX citation.
• —Split: train; 100 episodes; 39,229 frame rows; one task; 30 FPS; 100 source Parquet shards.
• —Task 0: Pick the gray block and keep it in the box.

Data layout

The table is jeevesh2009_so101_gray_block_lowvar_test and contains 39,229 rows across 100 TAG devices. The source Parquet shards total 2,206,758 bytes; the TsFile is 759,135 bytes (34.4% of the source Parquet size).

timestamp is not retained as a separate FIELD because it is represented by Time / 1000 seconds. The vector column names preserve their source prefixes, with dots changed to underscores. No trajectory row, episode, task, action dimension, or state dimension is removed.

Videos and alignment

The 200 source AV1 videos are not included here. They remain under `videos/chunk-000` in the original repository:

• —`observation.images.side`: 100 files
• —`observation.images.top`: 100 files

The source template is videos/chunk-{episode_chunk:03d}/{video_key}/episode_{episode_index:06d}.mp4. Use episode_index and frame_index to align each numeric row with both 30 FPS video streams.

Read example

from tsfile import TsFileReader

reader = TsFileReader("data/jeevesh2009_so101_gray_block_lowvar_test.tsfile")
with reader.query_table(
    "jeevesh2009_so101_gray_block_lowvar_test",
    ["episode_index", "task_index", "frame_index", "sample_index", "action_0", "observation_state_0"],
    batch_size=1024,
) as result:
    batch = result.read_arrow_batch()
    print(batch.to_pandas().head())
reader.close()