Legged Robotics · Learning-based Control
Ph.D. Student in Mechanical Engineering at KAIST
Learning-based control, motion generation, and perception for legged robots.
My research combines motion priors, reinforcement learning, and onboard perception for dynamic motion generation and perceptive locomotion. I am currently extending these methods from quadrupeds to humanoid locomotion, with whole-body manipulation as a future direction.
Dynamic Robot Control & Design Laboratory advised by Prof. Hae-Won Park
M.S. in Mechanical Engineering, KAIST · B.S. in Mechanical Engineering, Sungkyunkwan University
Selected work in learning-based control, motion generation, and perceptive locomotion.
July 2026 Cover Hover to preview Co-first author · * Equal contribution
A unified framework for agile quadrupedal locomotion that selects and transitions between reusable motor skills using onboard perception.
Hover to preview Co-first author · * Equal contribution
Dynamics-embedded flow matching that generates physically consistent motions and deployable actions directly from state-only demonstrations.
Platforms I have worked with in quadrupedal and humanoid robotics.
