Jiashun Wang
Papers
6
Total Citations
226
H-Index
5
About
Jiashun Wang is a leading researcher in dexterous robotic manipulation, focusing on enabling multi-finger hands to grasp and interact with objects as naturally as humans do. His work bridges computer vision, generative modeling, and imitation learning to tackle one of robotics’ hardest challenges: generalizable, human-like manipulation. Wang’s most influential paper, “Hand-Object Contact Consistency Reasoning for Human Grasps Generation” (2021, 150 citations), pioneered a method to generate natural human grasps from 3D objects by reasoning about contact consistency, moving beyond traditional parallel-jaw gripper approaches. He further advanced the field with the Continuous Grasping Function (CGF), a generative model that produces smooth, continuous grasping motions from human demonstrations, and USEEK, an unsupervised SE(3)-equivariant keypoint method enabling robots to manipulate unseen objects in arbitrary poses from a single demonstration. His work on learning generalizable dexterous manipulation from human grasp affordance (2022) and hand-informed visual representations for reinforcement learning (H-InDex, 2023) continues to push boundaries. With over 200 total citations and a growing portfolio of top-conference publications, Wang is shaping the future of dexterous robotics, making human-like manipulation more robust, generalizable, and accessible.
Research Focus
Key Achievements
Top Papers
- 1Hand-Object Contact Consistency Reasoning for Human Grasps Generation150 citations · 2021
- 2
- 3
- 4Learning Generalizable Dexterous Manipulation from Human Grasp Affordance11 citations · 2022
- 5Hand-Object Contact Consistency Reasoning for Human Grasps Generation6 citations · 2021
- 6