THULab/openarm_shirt_folding_new_converted

OpenArm Shirt Folding (TsFile)

Apache TsFile version of TobiBrtnr/openarm_shirt_folding_new_converted.

Overview

This LeRobot v2.1 dataset records a bimanual OpenArm robot performing the task "Fold the T-Shirt properly."

• Scale: 50 episodes, 135,766 frames, one task, and a single train split.
• Sampling rate: 30 Hz.
• Robot type: openarm_robot.
• Visual streams: one global camera and two wrist cameras at 480 x 640.

Schema (TsFile structure)

• Time (INT64, milliseconds) —round(timestamp * 1000), restarting for each episode.
• episode_index, task_index (TAG, INT64) —episode and task device dimensions; for example, query one trajectory with WHERE episode_index=0.
• frame_index, sample_index (FIELD, INT64) —frame position and global source sample index.
• observation_state_0..observation_state_47 (FIELD, FLOAT) —bimanual joint positions, velocities, and efforts.
• action_0..action_31 (FIELD, FLOAT) —bimanual joint position and angle commands.

The source timestamp field is represented by Time, and source index is renamed to sample_index. The three camera streams and their 150 MP4 files are not included; they remain available in the original videos directory.

Usage

Read data/openarm_shirt_folding_new_converted_train.tsfile with the Apache TsFile Java or Python SDK. All episodes share one table; use the TAG columns to select an episode or task.

Source & license

• Original dataset: https://huggingface.co/datasets/TobiBrtnr/openarm_shirt_folding_new_converted
• Author / publisher: TobiBrtnr
• Paper: not declared by the original dataset
• License: Apache License 2.0