Liang-ZX2025MIT
DexHandDiff Manipulate Block
Expert demonstration trajectories for in-hand dexterous manipulation with a simulated Shadow Hand performing block reorientation tasks, collected using TQC+HER for training diffusion-based planners.
Downloads0
Episodes5000
Why This Matters for Physical AI
Provides large-scale expert demonstrations for contact-rich dexterous manipulation that enable training of diffusion-based planning methods for goal-adaptive in-hand object reorientation.
Technical Profile
- Modalities
- proprioceptionstate
- Robot Embodiments
- Shadow Hand
- Action Space
- joint_positions
- Environment
- simulation
- Task Types
- manipulationdexterous_manipulationobject_reorientation
- Episodes
- 5000
- Data Format
- HDF5
- Annotation Types
- reward_labelsaction_labelsgoal_states
- License
- MIT
Access
Need custom proprioception data?
Claru builds purpose-built datasets for simulation applications with dense human annotations and quality assurance.
Request a Sample Pack