A Motion Tracking Method for the Modeling and Simulation of Human Movement in 3D

Interest in computer models of the human musculoskeletal system is increasing amongst clinicians and movement scientists as tools to analyze muscle and joint function in ways that are both non-invasive and inexpensive. Quantifying internal loads resulting from muscles and external forces is necessary to understand and treat movement pathologies. Current simulation methods, however, fall short of accurately reproducing observed human performance when compared to observations made during movement experiments. It is a minimal requirement that models closely reproduce external observations before considering them to answer questions about severity of a disorder and possible treatments. Given this underlying requirement, a dynamical motion-tracking method was developed to leverage experimental observations directly to guide forward simulations and provide simulations that are more accurate. In the present study, we demonstrate how tracking can provide joint-moment information that produce accurate forward simulations in minutes using a detailed 23 degree-offreedom human model and 3D motion-capture data. Keywords-musculoskeletal modeling; dynamic optimization; jumping; forward dynamics; inverse dynamics