Wanming Yu

Papers

Total Citations

H-Index

About

Wanming Yu is a leading researcher in robot motor learning, specializing in adaptive locomotion and dexterous manipulation. Her work addresses a core challenge in robotics: enabling machines to acquire versatile motor skills that generalize to new, unseen situations. Yu’s most influential contribution is the multi-expert learning architecture (MELA), which generates adaptive locomotion skills by learning from a group of representative expert behaviors. This work, with 50 citations, provides a foundational framework for skill transfer in robotics. She has also made critical advances in identifying the most important sensory feedback for learning robust locomotion, a study cited 22 times that helps demystify the black box of neural network-based control. Furthering data efficiency, Yu developed an accessibility-based clustering method for quadruped locomotion that automatically discovers optimal initial states, improving both robustness and sample efficiency. In manipulation, she pioneered a data-efficient imitation learning framework that leverages rich tactile sensing to achieve fine bimanual pinch-grasp skills from just a few real-world demonstrations. Her research consistently bridges the gap between simulation and real-world deployment, making her a key figure in the future of autonomous, adaptive robots.