taewonkooapache-2.0
Multi Pick and Place Mixed Suboptimal Seed1000 70pct 40ep
A suboptimal quality robotics dataset containing 40 episodes of pick and place tasks collected with a mobile manipulator, featuring mixed-quality demonstrations at 30fps with multi-view RGB observations and joint position actions.
Downloads0
Episodes40
Why This Matters for Physical AI
This dataset provides suboptimal demonstration data for training robust robotic manipulation policies that can learn from mixed-quality human teleoperation, addressing the challenge of learning from realistic imperfect data.
Technical Profile
- Modalities
- rgbproprioception
- Robot Embodiments
- so_follower
- Action Space
- joint_positions
- Environment
- lab
- Task Types
- pick_and_place
- Episodes
- 40
- Data Format
- parquet
- 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