svla_cap_pick_place (TsFile)

Apache TsFile version of cher2018/svla_cap_pick_place.

Overview

A LeRobot robot manipulation dataset. Each frame holds the commanded action and observed observation.state joint positions; camera views are stored as videos in the original dataset.

• Episodes: 35
• Frames: 26204
• Sampling rate: 30 fps
• Tasks: 1
• Split: a single train split
• Robot: so_follower
• Cameras (not uploaded): top, side

Schema (TsFile structure)

All episodes share one TsFile with episode_index and task_index as TAG columns; query a single episode with WHERE episode_index = N.

• Time (INT64, milliseconds) — round(timestamp * 1000); the source timestamp column is dropped (it equals Time / 1000).
• episode_index (TAG) — device dimension.
• task_index (TAG) — device dimension.
• frame_index (INT64) — measurement.
• sample_index (INT64) — measurement.
• action_0..N (FLOAT) — commanded joints.
• state_0..N (FLOAT) — measurement.

Vector columns are flattened element-wise with an index suffix.

Usage

Install the Apache TsFile Python SDK (pip install tsfile) and read a converted file:

from pathlib import Path
from tsfile import TsFileReader

path = Path("data/svla_cap_pick_place.tsfile")
with TsFileReader(str(path)) as reader:
    schemas = reader.get_all_table_schemas()
    print("tables:", list(schemas))
    table_name = next(iter(schemas))
    table = schemas[table_name]
    columns = [column.get_column_name() for column in table.get_columns()]
    print("columns:", columns)
    field_names = [
        column.get_column_name()
        for column in table.get_columns()
        if column.get_column_name() not in {"Time", "time"}
    ]
    if field_names:
        with reader.query_table(table_name, field_names[:3], batch_size=1024) as result:
            batch = result.read_arrow_batch()
            if batch is not None:
                print(batch.to_pandas().head())

Source & license

• Original dataset: https://huggingface.co/datasets/cher2018/svla_cap_pick_place
• Author / publisher: cher2018
• License: apache-2.0
• Note: camera videos are NOT included; see the original dataset for them.