Papers
33
Total Citations
1,032
H-Index
16
About
Zhongyu Li is a robotics researcher whose work spans legged locomotion, safety-critical control, and human-robot interaction, with a particular focus on applying deep reinforcement learning to create agile, robust controllers for bipedal and quadrupedal robots. His most influential contributions lie in developing RL-based frameworks that enable bipedal robots—most notably the Cassie platform—to perform a remarkable range of dynamic behaviors, from versatile walking and running to precise jumping maneuvers, accumulating over 200 and 101 citations respectively for his foundational locomotion studies. Li has also pioneered adaptive control strategies, demonstrating how rapid motor adaptation techniques originally developed for quadrupeds can be successfully transferred to the far more challenging domain of bipedal systems. Beyond locomotion, his research addresses safety-critical robot operation through novel Control Barrier Function formulations, ensuring feasibility and guaranteed constraint satisfaction in real-world deployments. His creative applied work includes a leash-guided robotic guide dog for visually impaired users, quadrupedal soccer goalkeeping, and multi-robot cable-towed load transportation. With over 790 total citations across a focused body of work, Li has established himself as a leading voice in bridging theoretical reinforcement learning with practical, deployable robotic systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2Robotic Guide Dog: Leading a Human with Leash-Guided Hybrid Physical Interaction102 citations · 2021
- 3
- 4
- 5Adapting Rapid Motor Adaptation for Bipedal Robots60 citations · 2022
- 6
- 7Robust and Versatile Bipedal Jumping Control through Reinforcement Learning55 citations · 2023
- 8
- 9
- 10