Jin Woo Hwang

I am a second-year student in the MS in Mechanical Engineering (Research Track), working with Professor Inseung Kang at the MetaMobility Lab at Carnegie Mellon University. I am interested in understanding how people move and in using simulation to help enhance human mobility.

My recent work explores joint kinematics as a device-agnostic interface for estimating biological joint moments: estimators trained only on open-source biomechanics datasets that transfer across hip and knee exoskeletons and a wearable IMU suite without device-specific fine-tuning. Before CMU, at the Wearable Robotics Lab at Seoul National University, I worked on human-in-the-loop personalization of hip assistance for people with Parkinson’s disease, which increased stride length and improved whole-body gait. Earlier, I worked in computer vision, including multiview pedestrian detection (WACV 2024) and privacy-preserving representation learning.

I am currently building musculoskeletal simulations (Hyfydy/SCONE, MuJoCo Warp) to train reinforcement learning–based controllers through imitation learning and adversarial motion priors, with the goal of bridging the sim-to-real gap for assistive devices. Alongside this, I am learning more about neuromuscular control—how the nervous system coordinates muscles during locomotion—so that these models better reflect how people actually move.