(click my name for my CV)
I am a master's student at Korea University, Department of Electrical and Electronic Engineering,Control and Mechatronics Lab(CML), pursuing my degree under the guidance of Professor Myotaeg Lim. Concurrently, I am working as a research assistant at Korea Institute of Science and Technology (KIST), Advanced Robot Control Lab(ARC) under the supervision of Dr. Yisoo Lee .
My research interests lie in humanoid locomotion, sampling-based Model Predictive Control (MPC), and multi-contact dynamics. These are some of the topics I am currently exploring.
1️⃣ How can sampling-based MPC methods be scaled to handle high-dimensional humanoid systems in real-time?
2️⃣ How can we incorporate uncertainty (e.g., modeling errors, sensor noise) into MPC to improve robustness in complex, dynamic environments?
3️⃣ How can we model and predict the interactions between a humanoid robot and its environment during multi-contact scenarios? (and how are the robot going to react? ex. GRF)
I believe that for robots to closely interact with humans and provide convenience, their workspace needs to expand. In this regard, I think locomotion and balance control will remain crucial research areas. Ultimately, I aim to develop household robots or disaster-response robots that can assist humans effectively.
✔️ Major: Electrical and Computer Engineering (GPA: 4.5/4.5)
✔️ Concentration: Control, Robotics and Systems
✔️ Graduate researcher at Advanced Robot Control Lab (ARC) in Korea Institute of Science and Technology (KIST) 2024 - present
✔️ Graduate researcher at Control and Mechatronics Lab (CML) in Korea University 2024 - present
✔️ Major: Mechanical, Robotics and Energy Engineering (GPA: 3.69/4.5)
✔️ Concentration: Robotics
✔️ Undergraduate researcher at Advanced Robot Control Lab (ARC) at Korea Institute of Science and Technology (KIST) 2022-2023
✔️ Model : MAHRU's next version, KAPEX! (Awaiting public release)
✔️ MPPI-based whole-body optimization
- Target system: Validated on a 3-DoF leg system for real-time control
- Key feature: Efficient torque optimization by explicitly eliminating contact forces
- Selected for recommendation for submission to the Journal of ICROS!!
✔️ Model : KIST - MAHRU
✔️ Efficient MPPI-based CoM trajectory generation enabling real-time humanoid walking:
- Average computation time: 0.09 ms (2 kHz update frequency)
- Compatibility: Low-end hardware and real-time control
✔️ Model : G1, KAPEX(Awaiting public release)
✔️ Method : MJPC + MPPI
-Reference paper:
"Predictive Sampling: Real-time Behaviour Synthesis with MuJoCo"
"Real-Time Whole-Body Control of Legged Robots with Model-Predictive Path Integral Control"
"Full-Order Sampling-Based MPC for Torque-Level Locomotion Control via Diffusion-Style Annealing"
✔️ Model : MAHRU's new model, KAPEX!(Awaiting public release)
✔️ Method : Contact constraint projection dynamics + MPPI
-Reference paper:
"Contact-Implicit MPC: Controlling diverse quadruped motions without pre-planned contact modes or trajectories"
"Per-Contact Iteration Method for Solving Contact Dynamics"
"Contact Consistent Control Framework for Humanoid Robots"
✔️ Model : G1
✔️ Method : MPPI CoM generator + weighted whole-body controller + footstep planning
-Reference footstep planners:
"Extremely Fast Computation of CoM Trajectory Generation for Walking Leveraging MPPI Algorithm"
"BaselineFootstepPlanner"
Authored Report
✔️ Model : Linear Inverted Pendulum Model (LIPM)
✔️ Method : Adaptive dynamic programming - on-policy learning -Reference code
"Textbook : robust adaptive dynamic programming"
✔️ Model : KIST - MAHRU
✔️ Method : MPPI CoM generator + Inverse Kinematics (IK) controller
✔️ Model : 2-Link Manipulation
✔️ Method : Model Predictive Path Integral (MPPI)
-Reference papers :
"Model predictive path integral control: Theoretical foundations and applications to autonomous driving"
"Model Predictive Path Integral Control: From Theory to Parallel Computation"
✔️ Model : Linear Inverted Pendulum Model (LIPM)
✔️ Method : Model Predictive Control (MPC)
- Reference paper :
"Trajectory-free linear model predictive control for stable walking in the presence of strong perturbations"
To be Continued.....