Research • Deep Learning/Robotics/Healthcare
About The Project
Researched user-independent, real-time bio-torque estimation for hip exoskeletons using Temporal Convolutional Networks trained on simulated IMU data. Created a Python toolbox for multimodal gait analysis (IMU/EMG/GRF) and clinical metrics (GDI, MGS), supporting studies with stroke survivors.
Achievements
Fine-tuned PyTorch TCN to reduce torque RMSE by 22.14% on stroke gait data
Generated synthetic IMU (simIMU) from motion capture for cross-device generalization
Built a clinical toolbox (IMU/EMG/GRF) over 20 stroke participants with GDI/MGS calculators
Ran human-in-the-loop pilots with second-skin sensor suits for real-time estimator refinement
Awarded GT President’s Undergraduate Research Award (PURA)
More Projects
Links
sunny.sunho.park@gmail.com
