Hugging Face's logo

Datasets:
THULab
/
soarm101-feeding-nuts

Tasks:
Robotics
Modalities:
Time-series
Size:
10K<n<100K
Tags:
LeRobot
robotics
so101
so-arm101
imitation-learning
manipulation
License:
Dataset card Files
xet
 Community
Episodes Preview SO-101 Visualizer
80 episodes · 30 fps

SO-ARM101 Feeding Nuts (TsFile)

Apache TsFile version of HCIS-Lab/soarm101-feeding-nuts.

Overview

Teleoperation demonstrations of a "Feed Nuts" manipulation task recorded on an SO-ARM101 follower arm. Each episode is one demonstration with per-frame 6-DOF joint position observations and actions, captured at 30 fps.

  • Robot: SO-ARM101 follower (6-DOF arm, robot_type = so101_follower).
  • Episodes: 80 demonstrations of one task ("Feed Nuts").
  • Frames: 74,355 timesteps total, recorded at 30 fps.
  • Source format: LeRobot v3.0.

Schema (TsFile structure)

All 80 episodes are stored in one TsFile table, using episode_index and task_index as TAG (device) dimensions.

  • Time (INT64, milliseconds) — per-frame timestamp, restarts at 0 for each episode (round(timestamp * 1000); ~33 ms apart at 30 fps).
  • episode_index (TAG) — demonstration id (0–79). Query a single episode with WHERE episode_index=0.
  • task_index (TAG) — task id (always 0; the dataset has one task, "Feed Nuts").
  • frame_index (FIELD, INT64) — frame number within the episode.
  • sample_index (FIELD, INT64) — the source global index column.
  • observation_state_0 .. observation_state_5 (FIELD, FLOAT) — 6-DOF observed joint positions (shoulder_pan, shoulder_lift, elbow_flex, wrist_flex, wrist_roll, gripper).
  • action_0 .. action_5 (FIELD, FLOAT) — 6-DOF commanded joint positions, same ordering as the state.

The source vector columns action[6] and observation.state[6] are flattened to single-precision FLOAT columns. The source timestamp column is dropped because it equals Time / 1000 seconds. All 74,355 rows are held in the single data/soarm101_feeding_nuts.tsfile.

Usage

Read the .tsfile files with the Apache TsFile Java or Python SDK.

# pip install tsfile
from tsfile import TsFileReader
reader = TsFileReader("data/soarm101_feeding_nuts.tsfile")
# query one episode: WHERE episode_index=0

Videos

The original dataset records two camera streams per episode (observation.images.wrist, observation.images.side). These MP4 videos are not included here — they are not time-series data. Retrieve them from the original dataset under HCIS-Lab/soarm101-feeding-nuts/videos.

Source & license

Downloads last month
90
Total file size:
1.42 MB