DaoyuanZhuapache-2.0
wb_pick_pointed_bread
Whole-body teleoperation dataset of a Unitree G1 humanoid robot picking up bread from a table and handing it to a human, containing 98 episodes with dual RGB cameras and 43-D joint state/action vectors.
Downloads0
Episodes98
Hours0.5
Why This Matters for Physical AI
This dataset demonstrates whole-body humanoid control through teleoperation with hand pose annotations, providing essential training data for learning dexterous manipulation and coordinated locomotion in humanoid robots.
Technical Profile
- Modalities
- rgbproprioceptionpose_estimation
- Robot Embodiments
- Unitree G1 WholeBodyhumanoid
- Action Space
- joint_positions
- Environment
- lab
- Task Types
- manipulationpick_and_placewhole-body_controlteleoperation
- Episodes
- 98
- Total Hours
- 0.5
- Data Format
- LeRobot
- Annotation Types
- pose_estimationhand_landmarksbody_keypoints
- 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