npowMIT

YAM GELLO Red Cap Manipulation Dataset (100 Demonstrations)

100 high-quality bimanual teleoperation demonstrations of tabletop object grasping and manipulation using the I2RT YAM robot arm teleoperated with a GELLO passive leader arm. Contains 25,874 timesteps (~22 minutes) of synchronized RGB video from 3 cameras and proprioceptive state data.

Downloads0
Episodes100
Hours0.37

Why This Matters for Physical AI

This dataset provides high-quality bimanual manipulation demonstrations with precise teleoperation feedback and multi-view observations, enabling research in imitation learning and robot control for complex grasping and manipulation tasks.

Technical Profile

Modalities
rgbproprioception
Robot Embodiments
I2RT YAM
Action Space
joint_positions
Environment
lab
Task Types
manipulationgraspingreaching
Episodes
100
Total Hours
0.37
Data Format
npz
License
MIT
Part of the YAM GELLO Red Cap Manipulation Dataset (100 Demonstrations) family

Access

Need custom rgb data?

Claru builds purpose-built datasets for lab applications with dense human annotations and quality assurance.

Request a Sample Pack

Related Datasets