THULab/aloha_static_tape

ALOHA Static Tape (TsFile)

Apache TsFile version of lerobot/aloha_static_tape.

Overview

This dataset was created using LeRobot. It contains ALOHA robot demonstrations represented as synchronized robot state, action, frame, task, and episode time-series data.

• Scale: 50 episodes, 35,000 frames, one train split, one task, and 50 fps.
• Source layout: LeRobot v3.0 frame Parquet data with four associated camera video streams.

Schema (TsFile structure)

• Time (INT64, milliseconds) — round(timestamp * 1000), restarting for each episode.
• episode_index, task_index (TAG) — original LeRobot device dimensions; query an episode with WHERE episode_index=0.
• frame_index, sample_index, next_done (FIELD, INT64) — frame number, source row index, and terminal-frame flag (0 or 1).
• observation_state_0..observation_state_13 (FIELD, FLOAT) — 14-dimensional robot observation state.
• action_0..action_13 (FIELD, FLOAT) — 14-dimensional robot action.

Vector fields are flattened into scalar FLOAT measurements. The source index and next.done fields are renamed to sample_index and next_done.

Usage

Read data/aloha_static_tape.tsfile with the Apache TsFile Java or Python SDK.

Source & license

• Original dataset: https://huggingface.co/datasets/lerobot/aloha_static_tape
• Author / publisher: LeRobot (Hugging Face)
• Paper: not declared by the original dataset
• License: Apache-2.0

The source timestamp column is not retained because it is exactly recoverable as Time / 1000 seconds. Camera videos are not included in this repository; they remain available in the original dataset's videos/ directory. Aside from the redundant timestamp column and video streams, no numeric time-series rows or columns are intentionally dropped.