knpt2026MIT
YAM bimanual — tool-rack placement
66 teleoperated episodes of a bimanual robot arm picking a screwdriver off a bench and inserting it into a tool rack, recorded at 40 Hz with three RGB cameras and converted to LeRobot v3.0 format.
Downloads316
Episodes66
Hours1.3147
Why This Matters for Physical AI
This dataset demonstrates contact-rich bimanual manipulation and insertion tasks with real robot data, enabling research into imitation learning and vision-language-action models for dexterous manipulation with DAgger-based interactive correction.
Technical Profile
- Modalities
- rgbproprioception
- Robot Embodiments
- YAM bimanual
- Action Space
- 14-DoF: left_yam_joint1, left_yam_joint2, left_yam_joint3, left_yam_joint4, left_yam_joint5, left_yam_joint6, left_yam_gripper, right_yam_joint1, right_yam_joint2, right_yam_joint3, right_yam_joint4, right_yam_joint5
- Environment
- lab
- Task Types
- manipulationgraspingtool_placementinsertion
- Episodes
- 66
- Total Hours
- 1.3147
- Data Format
- LeRobot
- Annotation Types
- language_instructionsprovenance_labels
- License
- MIT
Access
Need custom rgb data?
Claru builds purpose-built datasets for lab applications with dense human annotations and quality assurance.
Request a Sample Pack