oxe-augecc-by-4.0
bridge_train_5000_10000_augmented
An augmented version of the BridgeData V2 dataset containing 5,000 episodes with robot-agnostic image augmentation across 9 different robot embodiments, enabling cross-robot learning and generalization.
Downloads156
Episodes5000
Why This Matters for Physical AI
This dataset enables cross-embodiment robot learning by providing augmented visual observations across multiple robot platforms, facilitating generalization and transfer learning in robotic manipulation tasks.
Technical Profile
- Modalities
- rgbproprioception
- Robot Embodiments
- Google RobotJacoKinova3KUKA IIWAFranka PandaSawyerUR5exArm7
- Action Space
- joint_positions
- Environment
- lab
- Task Types
- manipulation
- Episodes
- 5000
- Data Format
- parquet
- Annotation Types
- language_instructions
- License
- cc-by-4.0
Community Signals
Top 50% by downloads
Academic Citations12
- Beyond performance: Explaining generalisation failures of Robotic Foundation Models in industrial simulation2025 · Biomimetic Intelligence and Robotics
- Empowering natural human–robot collaboration through multimodal language models and spatial intelligence: Pathways and perspectives2025 · Robotics and Computer-Integrated Manufacturing
Access
Need custom rgb data?
Claru builds purpose-built datasets for lab applications with dense human annotations and quality assurance.
Request a Sample Pack