Yiming Pang
Papers
3
Total Citations
28
H-Index
3
About
Yiming Pang is a leading researcher in legged robotics, specializing in reinforcement learning (RL) for agile locomotion and dynamic control. His work addresses the fundamental challenge of bridging the simulation-to-reality gap, enabling quadruped robots to perform complex, adaptive maneuvers in unstructured environments. Pang’s most cited paper, "Learning Locomotion for Quadruped Robots via Distributional Ensemble Actor-Critic" (2024, 14 citations), introduces a novel approach to handling aleatoric uncertainty from domain randomization, achieving remarkable sim-to-real transfer on real robots. He further advanced the field with "Learning Agility and Adaptive Legged Locomotion via Curricular Hindsight Reinforcement Learning" (2024, 8 citations), which proposes a curricular hindsight RL method that empowers robots with skills like fall recovery, high-speed turning, and sprinting in the wild. His work on "Dynamic Fall Recovery Control for Legged Robots via Reinforcement Learning" (2024, 6 citations) tackles the inevitable challenge of falls in real-world deployment, developing robust recovery motor skills. Collectively, Pang’s contributions are shaping the next generation of resilient, agile legged robots capable of operating in unpredictable terrains.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3