rtrtwrwapache-2.0
granular-material-pour -- buckwheat, single target
290 teleoperated pours of buckwheat onto a scale using a single YAM arm, recorded over 3.41 hours at 50 Hz with 3 cameras and a 1 kg load cell. Policies trained on this session perform open-loop pouring using visual feedback only.
Downloads38
Episodes290
Hours3.41
Why This Matters for Physical AI
This dataset addresses the challenge of learning precise manipulation skills with granular materials and demonstrates open-loop control strategies where policies must rely on visual feedback rather than force or weight information.
Technical Profile
- Modalities
- rgbproprioceptionforce_torque
- Robot Embodiments
- YAM arm
- Action Space
- joint_positions
- Environment
- lab
- Task Types
- pouringmanipulation
- Episodes
- 290
- Total Hours
- 3.41
- Data Format
- LeRobot
- Annotation Types
- language_instructions
- License
- apache-2.0
Access
Need custom rgb data?
Claru builds purpose-built datasets for lab applications with dense human annotations and quality assurance.
Request a Sample Pack