Voxel512026cc-by-4.0
Hoi! Dataset (FiftyOne)
88 interaction episodes from the Hoi! CVPR 2026 dataset capturing humans and robotic grippers opening or closing articulated objects across 7 real-world scenes, with multimodal sensor streams including RGB, depth, force-torque, and tactile data spatially registered to a shared world frame.
Downloads24
Episodes88
Likes1
Why This Matters for Physical AI
This dataset enables training of force-aware manipulation policies and cross-embodiment learning by providing synchronized multimodal sensor data (vision, force, tactile) across different robot morphologies interacting with real articulated objects in naturalistic settings.
Technical Profile
- Modalities
- rgbdepthforce_torquetactileproprioceptionpoint_cloudhand_trackingeye_gaze
- Robot Embodiments
- Hoi! GripperUMI gripperhuman hand
- Environment
- labhomekitchenbedroombathroomofficefurniture_store
- Task Types
- articulated_object_manipulationopeningclosing
- Episodes
- 88
- Data Format
- MCAP
- Annotation Types
- articulation_annotationsforce_labelstactile_labelshand_trackingeye_gaze
- License
- cc-by-4.0
Access
Need custom rgb data?
Claru builds purpose-built datasets for lab applications with dense human annotations and quality assurance.
Request a Sample Pack