Donghyun1228
LIBERO π0.5 rollouts in shifted camera views
Zero-shot policy rollouts collected in shifted-camera LIBERO environments with inverse-dynamics adaptation windows. Includes both successful and failed episodes across multiple camera viewpoints for zero-shot transfer learning.
Downloads0
Episodes200 per suite and environment (199 for libero_10)
Why This Matters for Physical AI
This dataset enables research in zero-shot transfer and inverse-dynamics adaptation by providing policy rollouts across multiple camera viewpoints and environment configurations, essential for understanding how embodied agents generalize to new visual perspectives.
Technical Profile
- Modalities
- rgbproprioception
- Robot Embodiments
- Franka Panda
- Action Space
- joint_positions
- Environment
- simulation
- Task Types
- manipulationpick_and_place
- Episodes
- 200 per suite and environment (199 for libero_10)
- Data Format
- LeRobot
- Annotation Types
- action_labels
Access
Need custom rgb data?
Claru builds purpose-built datasets for simulation applications with dense human annotations and quality assurance.
Request a Sample Pack