franka_multi_task_v1_vid (TsFile)

Apache TsFile version of Beegbrain/franka_multi_task_v1_vid.

Overview

A LeRobot v3.0 robot manipulation dataset covering 4 tasks. Each frame holds the commanded action and observed observation.state joint positions; camera views are stored as videos in the original dataset.

• Episodes: 200
• Frames: 75,843
• Sampling rate: 10 fps
• Tasks: 4
• Split: a single train split (0:200)
• Robot: not declared in the source metadata

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.
• observation_state_0..7 (FLOAT) — measured joint positions and gripper.
• action_0..7 (FLOAT) — commanded joint positions and gripper.

The vector columns are flattened per joint (joint_0..joint_6, gripper per the source meta/info.json feature names).

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/franka_multi_task_v1_vid.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/Beegbrain/franka_multi_task_v1_vid
• Author / publisher: Beegbrain
• License: apache-2.0
• Note: camera videos (wrist_camera, side_camera) are NOT included; see the original dataset for them.