mulligan2026apache-2.0
real-marker-d2-r04-eval-b02
Evaluation dataset of 150 episodes from the real-marker-d2 task, part of the Mulligan robotics benchmark collection. Contains multi-camera observations from robot manipulation tasks with 68,250 frames recorded at 15 FPS.
Downloads67
Episodes150
Why This Matters for Physical AI
This dataset contributes to physical AI research by providing real robot manipulation demonstrations with multi-view camera observations for training and evaluating visuomotor policies in the Mulligan benchmark framework.
Technical Profile
- Modalities
- rgb
- Environment
- lab
- Task Types
- manipulation
- Episodes
- 150
- Data Format
- parquet
- Annotation Types
- reward_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