← Sinan Lab › All datasets › Open-TeleVision teleoperation datasets
Dataset · OpenTeleVision (Xuxin Cheng et al.)

Open-TeleVision teleoperation datasets

Landing page →

Real-robot teleop recordings; per-task demo counts in paper: Can Sorting 10, Can Insertion/Folding/Unloading 20 each; no total episode count or dataset license stated.

Facts

Organisation
OpenTeleVision (Xuxin Cheng et al.)
Form tags
humanoid · dual arm
Embodiments (as stated)
Unitree H1 with 6 DoF Inspire hands · Fourier GR-1 with jaw gripper
Episodes
unverified
Hours
unverified
Frames
unverified
Tasks
4
Scenes
unverified
Modalities
rgb · state
Format
custom (raw recordings placed in data/recordings/ and processed by repo scripts)
Licence
Apache-2.0 (code repository); dataset license not separately stated
Commercial use
unverified
Released
2024-07
Size
unverified
Sources · 6 sources
embodiments https://arxiv.org/html/2407.01512 Official page · Fetched 2026-09-18 · No snapshot stored yet 'Unitree H1' with '6 DoF Inspire' hands; 'Fourier GR-1' with 'jaw gripper'
tasks https://robot-tv.github.io/ Official page · Fetched 2026-09-18 · No snapshot stored yet 'Can Sorting, Can Insertion, Folding, and Unloading'
modalities https://arxiv.org/html/2407.01512 Official page · Fetched 2026-09-18 · No snapshot stored yet 'A ZED Mini stereo camera ... stereo RGB streaming' at 480x640 per eye, 60 Hz; hand/head/wrist SE(3) poses and joint positions recorded
format https://github.com/OpenTeleVision/TeleVision Official page · Fetched 2026-09-18 · No snapshot stored yet Dataset on Google Drive (drive.google.com/drive/folders/11WO96mUMjmxRo9Hpvm4ADz7THuuGNEMY); 'Place the downloaded dataset in data/recordings/'
license https://raw.githubusercontent.com/OpenTeleVision/TeleVision/main/LICENSE Official page · Fetched 2026-09-18 · No snapshot stored yet Apache License, Version 2.0 (code)
release_date https://arxiv.org/abs/2407.01512 Official page · Fetched 2026-09-18 · No snapshot stored yet Submitted 1 Jul 2024; CoRL 2024

Related bodies in the index

Matched by stated robot names or form; a match means the form is covered, not that a model was trained on it.

SinanLab · Sinan LabClear direction, grounded judgement.