Wonchul Kim

About

Wonchul Kim is a robotics researcher whose work sits at the intersection of control theory, machine learning, and autonomous navigation. His primary research areas include model predictive control, deep reinforcement learning, and vision-based robotic control, with a particular focus on challenging off-road and manipulator platforms. Kim made significant contributions to the control of skid-steer vehicles, which are notoriously difficult to maneuver due to slippage on rough terrain. He pioneered the use of model predictive control enhanced with on-line sparse Gaussian processes to handle these nonlinear dynamics, a work that has garnered 26 citations. He also advanced the field of robot skill acquisition by combining Dynamic Movement Primitives with hierarchical deep reinforcement learning from demonstration, enabling robots to learn and generalize complex movements. His vision-based approaches, including guided policy search with field-of-view constraints for target tracking and deep RL for manipulator control from raw pixel data, demonstrate a consistent focus on end-to-end learning from perception to action. With over 60 total citations, Kim’s research provides practical frameworks for deploying autonomous robots in unstructured environments.

Research Focus

Key Achievements

Top Papers

1. 1
   Path Tracking for a Skid-steer Vehicle using Model Predictive Control with On-line Sparse Gaussian Process

   26 citations · 2017
2. 2
   Learning and Generalization of Dynamic Movement Primitives by Hierarchical Deep Reinforcement Learning from Demonstration

   16 citations · 2018
3. 3
   Vision-based Target Tracking for a Skid-steer Vehicle using Guided Policy Search with Field-of-view Constraint

   8 citations · 2018
4. 4
   Three-link planar arm control using reinforcement learning

   8 citations · 2017
5. 5
   Vision-based deep reinforcement learning to control a manipulator

   4 citations · 2017

Key Collaborators

Contact & Links