Kyungjae Lee
Seoul National University, Chung Cheong University, Statistics Korea
Papers
13
Total Citations
152
H-Index
7
About
Kyungjae Lee is a leading researcher in robot learning and autonomous navigation, whose work bridges the gap between theoretical planning algorithms and real-world robotic systems. His core research spans Monte Carlo tree search (MCTS) in continuous spaces, reinforcement learning (RL) with entropy regularization, and nonparametric motion planning for dynamic environments. Lee’s major contributions include pioneering the use of Voronoi optimistic optimization for MCTS in continuous action spaces—a framework with applications from robotics to data-center management—and developing generalized Tsallis entropy RL, which extends maximum-entropy methods to improve exploration and robustness in soft mobile robots. His work on real-time nonparametric reactive navigation and Gaussian random paths has enabled mobile robots to navigate cluttered, dynamic settings with minimal computation. With over 148 citations across his top papers, Lee’s impact is evident in both theory and practice: his entropy-adaptive RL has been applied to tripod robots using soft vibration actuators, and his uncertainty-aware learning from demonstration methods have advanced human-robot interaction. Notable achievements include his 2020 papers on continuous-space MCTS and Tsallis entropy RL, which are foundational for next-generation autonomous systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4
- 5Gaussian random paths for real-time motion planning14 citations · 2016
- 6
- 7
- 8Hierarchical 6-DoF Grasping with Approaching Direction Selection7 citations · 2020
- 9
- 10