LeWAM2026apache-2.0
LeWAM Eval PushT B601
Real-robot evaluation rollouts of the LeWAM self-supervised policy on a Rebot B601 7-DoF arm performing non-prehensile PushT manipulation tasks. Contains 30 policy rollouts with human-annotated success labels and full sensor data from synchronized camera streams and joint state.
Downloads143
Episodes30
Hours0.3057
Why This Matters for Physical AI
Provides real-world evaluation data of a self-supervised vision-based manipulation policy on physical hardware, demonstrating practical performance of end-to-end learned controllers on non-prehensile pushing tasks with explicit success annotations.
Technical Profile
- Modalities
- rgbproprioception
- Robot Embodiments
- Rebot B601
- Action Space
- 7-DoF: shoulder_pan.pos, shoulder_lift.pos, elbow_flex.pos, wrist_flex.pos, wrist_yaw.pos, wrist_roll.pos, gripper.pos
- Environment
- lab
- Task Types
- manipulationnon-prehensile_pushing
- Episodes
- 30
- Total Hours
- 0.3057
- Data Format
- LeRobot
- Annotation Types
- reward_labels
- 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