Closed-loop EMG-informed model-based analysis of human musculoskeletal mechanics on rough terrains.

Abstract

This work aims at estimating the musculoskeletal forces acting in the human lower extremity during locomotion on rough terrains. We employ computational models of the human neuro-musculoskeletal system that are informed by multi-modal movement data including foot-ground reaction forces, 3D marker trajectories and lower extremity electromyograms (EMG). Data were recorded from one healthy subject locomoting on rough grounds realized using foam rubber blocks of different heights. Blocks arrangement was randomized across all locomotion trials to prevent adaptation to specific ground morphology. Data were used to generate subject-specific models that matched an individual's anthropometry and force-generating capacity. EMGs enabled capturing subject- and ground-specific muscle activation patterns employed for walking on the rough grounds. This allowed integrating realistic activation patterns in the forward dynamic simulations of the musculoskeletal system. The ability to accurately predict the joint mechanical forces necessary to walk on different terrains have implications for our understanding of human movement but also for developing intuitive human machine interfaces for wearable exoskeletons or prosthetic limbs that can seamlessly adapt to different mechanical demands matching biological limb performance.