Hyeonbeen (Edward) Lee

I am an incoming PhD student at Virginia Tech, where I'll work with Prof. Simon Stepputtis in the TEA Lab. I am interested in physics-aware machine learning and neurosymbolic AI for contact-rich robotics.

Before joining Virginia Tech, I received my B.Eng. and M.Eng. degrees in Mechanical Engineering at Kyung Hee University, where I worked with Prof. Jin-Gyun Kim in the Modeling & Simulation Lab. In 2023, I was fortunate to work with Prof. Joseph Lim in the CLVR Lab at KAIST AI.

I usually go by Edward, but you can also call me Hyeonbeen, which is my given name in South Korea.

edwardlee [at] vt.edu  /  CV  /  Scholar  /  LinkedIn  /  GitHub

Research

I'm interested in grounding AI models in physics to enable efficient, interpretable and generalizable control of robot-environment interactions. In Kyung Hee University, I used to work on physics-aware machine learning for modeling nonlinear dynamics of contact- and vibration-rich physical systems.

Frequency-aware Decomposition Learning for Sensorless Wrench Forecasting on a Vibration-rich Hydraulic Manipulator

Combines decomposition-based probabilistic modeling, frequency-awareness, and proprioception-to-wrench transfer learning to forecast contact- and vibration-rich wrench signals in the short-term future without a physical F/T sensor.

Buoyancy-integrated Hybrid Reaction Force Estimation Method with Real-time Haptic Feedback for Underwater Hydraulic Manipulation

cNN-DP: Composite Neural Network with Differential Propagation for Impulsive Nonlinear Dynamics

Sequentially predicts lower- to higher-order dynamic responses from time and time-invariant system parameters through physics-aware differential propagation, enabling accurate data-driven modeling of impulsive dynamic responses.

Multi-body Dynamics Model for Spent Nuclear Fuel Transportation System under Normal Transport Test Conditions

Simulates the dynamic response of a spent nuclear fuel transportation system (KORAD-21) under normal road and sea transport conditions using a multi-body dynamics model calibrated and validated against real-world transport data.

Design and source code from Jon Barron's website