Rutav
MimicDroid: In-Context Learning for Humanoid Robot Manipulation from Human Play Videos
A dataset of 8 hours of human play videos and 30 objects across 8 kitchen environments for training humanoid robots to solve manipulation tasks through in-context learning. Built on the RoboCasa simulation framework with evaluation spanning three difficulty levels.
Downloads119
Technical Profile
- Modalities
- rgb
- Robot Embodiments
- humanoid
- Environment
- simulationkitchen
- Task Types
- manipulationpick_and_placegrasping
Community Signals
HuggingFace Discussions1
Access
Need custom rgb data?
Claru builds purpose-built datasets for simulation applications with dense human annotations and quality assurance.
Request a Sample Pack