Hyesung Lee

Hyesung Lee

Ph.D. student in the Kim Jaechul Graduate School of AI at KAIST & KIST.

About

I am Hyesung Lee, a Ph.D. student in the Kim Jaechul Graduate School of AI at KAIST & KIST, advised by Prof. Kimin Lee (KAIST) and Dr. Sungwook Yang (KIST). My research focuses on Learning-Based Dexterous Hand Manipulation, Dynamics-Aware Visuo-Tactile Representation Learning, and Task Adaptive Compliance Control for Multi-Fingered Robotic Hands.

Research Interests

  • Learning-Based Dexterous Hand Manipulation
  • Dynamic Aware Visuo-Tactile (also Kinesthetic) Representation Learning
  • Vision-Haptic-Language-Action (VHLA) Models
  • Sample Efficient Reinforcement Learning and Data Efficient Imitation Learning

News

  • March 2026: Started Ph.D. in the Kim Jaechul Graduate School of AI, KAIST & KIST.
  • January 2026: Work on “SLAM-Guided Real-time Closed-Loop Photodynamic Therapy” accepted to IEEE RA-L / ICRA 2026.
  • October 2025: Presented “DexRefine” at the IEEE Humanoids 2025 Dexterous Humanoid Manipulation Workshop.
  • May 2023: Published “EKF-Sensor Fusion based Endomicroscopic Image Mosaicking” work at IEEE EMBC 2023.

Education

KAIST & KIST
Ph.D. in Kim Jaechul Graduate School of AI, March 2026 - Present
Advisor: Sungwook Yang (KIST), Kimin Lee (KAIST)

Korea University & KIST
M.S. in Computer Science, February 2024 - February 2026
Advisor: Sungwook Yang (KIST), Jinkyu Kim (Korea University)
Thesis: Learning to Refine Human Motion into Physically Feasible Robotic Actions for Dexterous Manipulation

Kwangwoon University
B.S. in School of Robotics, March 2017 - August 2023
Advisor: Jin-Oh Kim

Experience

KIST (Korea Institute of Science and Technology), Seoul, Republic of Korea
Intern, March 2023 - August 2024

  • Researched Endomicroscopic Image SLAM and Sensor Fusion.
  • Performed Photo-Theranosis System Integration.

Kakao Mobility, Seoul, Republic of Korea
Contract Worker, December 2021 - May 2022

  • Built a Promotional Autonomous Driving Mobile Robot for Kakao NEMO 2022.
  • Conducted Stereo Camera Visual Odometry Research.

Stryx (Startup), Seoul, Republic of Korea
Intern, September 2021 - December 2021

  • Developed Road-Data Acquisition Software using 3D LiDAR, GNSS, and Machine Vision Cameras.
  • Built a Sensor Calibration Visualization Tool using ROS RViz.

Publication

Current Research

  1. DexRefine: Refine Human Motion to Physically Feasible Robotic Actions
    Authors: Hyesung Lee, Si-Hwan Heo, Sungwook Yang
    IEEE Humanoids Workshop, 2025
    Paper

Previous Research

  1. Real-Time SLAM-Guided Closed-Loop Photodynamic Therapy with Pixel-Accurate Light-Dose Control
    Authors: Hyesung Lee, Sungwook Yang
    IEEE RA-L / ICRA, 2026
    Paper
  2. Real-time endomicroscopic image mosaicking with an ekf-based sensor fusion approach
    Authors: Jin Kim, Hyesung Lee, Sang-Rok Oh, Sungwook Yang
    IEEE EMBC, 2023
    Paper

Awards

  • 2023.04 - Paper Excellence Award, KSME Bio Engineering Division Poster Presentation
  • 2019.10 - 1st Place, National Defense Robot Contest (Hanwha Defense)
  • 2019.08 - 2nd Place, Busan Robot Contest Turtlebot3 Autorace (Busan Bexco)
  • 2018.11 - 3rd Place, 2018 R-BIZ Challenge Turtlebot3 Autorace (Daegu EXCO)
  • 2018.10 - 5th Place, IRC 2018 Sumo Robot Competition (Ilsan KINTEX)
  • 2017.11 - Grand Prize, Introduction to Engineering Design Exhibition (Kwangwoon University)

Contact