mulligan2026MIT
real-square-d2-c05-dagger-mining-no-cf
A robotics dataset part of the Mulligan project containing real robot manipulation data for the square-d2 task with 50 episodes collected using dagger mining without counterfactual augmentation.
Downloads78
Episodes50
Why This Matters for Physical AI
This dataset contributes to physical AI research by providing real robot manipulation trajectories with multi-camera observations that can be used for training and evaluating embodied AI policies through imitation learning.
Technical Profile
- Modalities
- rgb
- Environment
- lab
- Task Types
- manipulation
- Episodes
- 50
- Data Format
- parquet
- Annotation Types
- language_instructionsreward_labels
- License
- MIT
Access
Need custom rgb data?
Claru builds purpose-built datasets for lab applications with dense human annotations and quality assurance.
Request a Sample Pack