mulligan2026apache-2.0
real-routing-d2-c06-dagger-mining-no-cf
A robotics dataset containing 50 episodes of real robot manipulation data collected for training routing policies, part of the Mulligan project. The dataset includes multi-camera observations and trajectory data with 19,121 frames recorded at 15 FPS.
Downloads18
Episodes50
Why This Matters for Physical AI
This dataset provides real robot demonstration data for training dagger-based policies on complex routing and manipulation tasks, contributing to embodied AI research through multi-camera observation perspectives and trajectory-level annotations.
Technical Profile
- Modalities
- rgb
- Environment
- lab
- Task Types
- manipulation
- Episodes
- 50
- Data Format
- parquet
- Annotation Types
- language_instructionsreward_labels
- 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