Human Universal Grasping
Introduces HUG, a flow-matching model that predicts diverse human grasps from a single RGB-D image.
Evidence notes
- Trains on 1M-HUGs, an egocentric human grasp dataset spanning more than one million frames and thousands of objects.
- Includes HUG-Bench and real-world retargeting results across robot hand embodiments.
Company context
Research university with robotics groups publishing open-source dexterous hand hardware and robot-learning systems.