Han-Byeol Kim
Papers
2
Total Citations
6
H-Index
2
About
Han-Byeol Kim is a researcher at the forefront of robotic manipulation, specializing in the integration of vision-based deep reinforcement learning and large language models (LLMs) for autonomous control. Their work addresses a critical challenge in robotics: enabling manipulators to perform complex, real-world tasks with minimal human intervention. Kim’s foundational paper, “Vision-based deep reinforcement learning to control a manipulator” (2017, 4 citations), pioneered the use of camera images and RL algorithms to guide robotic actions through reward-maximizing policies, laying the groundwork for data-driven control. Building on this, their recent breakthrough, “TARG: Tree of Action-reward Generation With Large Language Model for Cabinet Opening Using Manipulator” (2025, 2 citations), introduces a novel framework that leverages LLMs to generate hierarchical action-reward trees, enabling robots to reason and adapt during tasks like cabinet opening. This work bridges symbolic reasoning and physical interaction, showcasing Kim’s ability to fuse cutting-edge AI with practical robotics. With a growing citation impact, Kim’s contributions are shaping the future of autonomous manipulation, offering scalable solutions for industrial and service robotics. Their research is a must-read for students and engineers exploring the synergy between reinforcement learning, language models, and robotic control.
Research Focus
Key Achievements
Top Papers
- 1Vision-based deep reinforcement learning to control a manipulator4 citations · 2017
- 2