Hanwen Jiang

UC San Diego Health System, University of California San Diego, The University of Texas at Austin

Papers

Total Citations

279

H-Index

About

Hanwen Jiang is a rising researcher at the forefront of robotic dexterous manipulation and computer vision, whose work bridges the critical gap between human-like grasping and autonomous robot control. His primary research areas include human grasp generation, imitation learning, and dense visual correspondence. Jiang’s major contribution is the development of novel frameworks that enable robots to learn natural, multi-finger grasping strategies directly from human demonstrations. His influential paper on "Hand-Object Contact Consistency Reasoning for Human Grasps Generation" (150 citations) introduced a reasoning framework that ensures physically plausible hand-object interactions, moving beyond traditional parallel-jaw gripper approaches. He further advanced the field with "DexMV: Imitation Learning for Dexterous Manipulation from Human Videos" (117 citations), a seminal work that demonstrated how robots can acquire complex manipulation skills by observing human hand movements in video. More recently, his work "Doduo" tackles the fundamental challenge of establishing dense visual correspondence across dynamic scenes, a capability vital for robust robotic perception. Through these contributions, Jiang is shaping how robots understand and interact with the physical world, making dexterous manipulation more intuitive and human-like.