기계공학부 구성원들의 많은 관심과 참여 부탁드립니다.

▣ 주   제: Towards digital motor clones for gait assistive devices

▣ 연   사: 송성문 교수

▣ 소   속: Northeastern University

▣ 일   시: 2024. 7. 4.(목) 11:00

▣ 장   소: 제4공학관 D603호

▣ 초   청: 신동준 교수

▣ 초   록

My research goal is to develop digital motor clones that can accurately predict the movements of individuals in novel scenarios. We focus on neuromechanical simulations of human locomotion, where motor control models actuate musculoskeletal models within physics simulations to predict and explain human movement. This simulation framework has significant implications for studying human physiology and developing assistive devices. I will provide an overview of our earlier efforts involving a reflex-based control model that generates diverse human-like locomotion, responds to perturbations in a manner similar to humans, and can be artificially aged to simulate elderly-like gaits. I will then discuss our ongoing efforts to utilize deep reinforcement learning methods to broaden these capabilities to include atypical locomotion behaviors and to serve as a simulation testbed for gait-assistive exoskeletons. Additionally, we adapt the human simulation model to act as a virtual neuromuscular controller for robotic systems such as ankle exoskeletons. Lastly, I'll present our collaborative work on MyoSuite, an integrated open-source platform for neuromechanical simulations.