oxe-augecc-by-4.0
berkeley_autolab_ur5_train_0_100
An augmented robotics dataset containing 100 episodes of manipulation tasks with robot-specific image augmentations for 9 different robot embodiments, providing cross-embodiment policy learning data with joint positions, end-effector poses, and natural language instructions.
Downloads178
Episodes100
Why This Matters for Physical AI
This augmented cross-embodiment dataset enables scalable policy learning across multiple robot platforms by providing robot-specific visual augmentations and normalized trajectory representations, advancing research in embodiment-agnostic robot learning.
Technical Profile
- Modalities
- rgbproprioceptionlanguage
- Robot Embodiments
- google_robotjacokinova3kuka_iiwapandasawyerwidowXxarm7
- Action Space
- joint_positions
- Environment
- lab
- Task Types
- manipulation
- Episodes
- 100
- Data Format
- parquet
- Annotation Types
- language_instructions
- License
- cc-by-4.0
Community Signals
Top 50% by downloads
Access
Need custom rgb data?
Claru builds purpose-built datasets for lab applications with dense human annotations and quality assurance.
Request a Sample Pack