Juhyeok Mun

Korea Advanced Institute of Science and Technology, Robotics Research (United States)

Papers

Total Citations

335

H-Index

About

Juhyeok Mun is a leading researcher in legged robotics, specializing in reinforcement learning for dynamic locomotion, state estimation, and autonomous navigation. His most impactful work introduces a groundbreaking framework for concurrent training of control policies and state estimators, enabling robots to achieve robust, high-speed locomotion without explicit state supervision—a paper that has garnered 179 citations and is reshaping how researchers approach policy learning. Mun further advanced the field by tackling the challenge of deformable terrain, developing simulation-based reinforcement learning methods that allow quadrupeds to maintain stability and agility on soft ground, a contribution cited 147 times and critical for real-world deployment. His recent work on semantic traversability, using egocentric video and automated annotation, pushes robots toward urban navigation with scene understanding. Notably, Mun led the RAIBO2 project, where a quadruped robot completed a full marathon on a single battery charge, demonstrating unprecedented energy efficiency and endurance. With a growing citation record and innovations bridging simulation to reality, Mun’s research is pivotal for creating legged robots that are both intelligent and practically deployable.