Shaowei Liu

UC San Diego Health System, University of California San Diego, University of Illinois Urbana-Champaign

Papers

Total Citations

280

H-Index

About

Shaowei Liu is a leading researcher in computer vision and robotics, specializing in the generation and understanding of dexterous human-object interactions. His work bridges the gap between human manipulation and robotic imitation, with a focus on generating realistic, multi-finger human grasps from 3D objects. His highly cited 2021 paper, *Hand-Object Contact Consistency Reasoning for Human Grasps Generation* (150 citations), tackles the challenging problem of producing natural hand poses that maintain consistent contact with objects, moving beyond traditional parallel-jaw gripper approaches. Liu further advanced the field with *DexMV: Imitation Learning for Dexterous Manipulation from Human Videos* (117 citations), which demonstrates how human demonstration videos can be leveraged to teach robots complex manipulation skills. His more recent work, *Building Rearticulable Models for Arbitrary 3D Objects from 4D Point Clouds* (2023), introduces a novel method for decomposing everyday objects into articulated parts from dynamic point cloud videos, enabling robots to understand and interact with objects that have moving components. With a strong citation impact and a focus on practical, real-world applications, Liu’s research is shaping the future of dexterous robotics and human-robot collaboration.