I am broadly interested in control and planning of dynamic systems, particularly in robotics. My research objective is to enable robots to perform complex behaviors and exhibit physical intelligence through learning. These interests span both manipulation and locomotion in real-world robotics.
I am an M.S. student in Mechanical Engineering at KAIST, advised by Prof. Jemin Hwangbo in the RAI Lab. Previously, I was an undergraduate research intern in the Robotics Lab at Seoul National University, working with Prof. Frank C. Park.
B.S. in Mechanical Engineering; B.S. in Electrical and Computer Engineering (Double Major)
Summa Cum Laude · Ranked 3rd in the department · GPA: 4.10/4.30
CoRL 2026 Spotlight
A generative policy's trajectory ensemble reveals feasible and constrained motion directions, enabling impedance control that stays stiff along feasible motion and compliant along constraints.
Research Intern · Prof. Jemin Hwangbo
Trained a recovery policy for the Raibo2 quadruped via deep RL in RaiSim. A 4-step reward curriculum addressed sparse rewards, while log-barrier rewards penalized violations of motor-aware hard constraints.
Early Crew Member
Developed an imitation learning pipeline for laparoscopic suturing using surgeon demonstrations, instrument sensing, digital twin feasibility checks, and trocar-adaptive diffusion policies.
AI & Robotics Engineer Intern
Built sim-to-real pipelines for deploying reinforcement learning policies trained in simulation on Unitree Go2 quadruped and G1 humanoid robots.
Research Intern · Prof. Frank C. Park
Conducted research on impedance control and bimanual manipulation, including a first-author CoRL 2026 paper.
Bimanual manipulation
Connects learned motion plans to screw-decomposed impedance control for bimanual articulated-object manipulation. An RL policy uses object screw axes and wrench feedback to adapt compliance and reduce opposing forces between the arms.
World models and planning
Predicts dynamics and vision-language embeddings in a recurrent world model for MPC with text or image goals. Delta-score rewards support sequential subtask planning without training a separate policy for each goal.
Computer vision · Aegis AI
Led development of Aegis AI's generative-image detection system for the K-Startup Challenge.
Sim-to-real reinforcement learning
A self-balancing suitcase that shifts its center of mass under PPO control, trained in Isaac Lab and deployed on hardware. My contribution covered the simulation environment, policy training, and sim-to-real bridge.