Neuromusculoskeletal Model-Based Controller for Voluntary and Continuous Assistance in a Broad Range of Locomotion Tasks

Neuromusculoskeletal modeling driven by electromyograms (EMG) has shown the ability to predict joint torque for a wide variety of movements. Taking advantage of this, we connected a real-time version of an EMG-driven model to a bilateral ankle exoskeleton to continuously assist during a wide repertory of locomotion tasks. The advantage is that the user controls the exoskeleton voluntarily, i.e. as a direct function of own muscle force. As a result, the developed framework can seamlessly assist in different tasks as well as transitions between tasks without having to change any parameters or using an additional algorithm to switch between states. Results on three participants show that an exoskeleton controlled via this framework reduces EMGs during locomotion tasks with various speeds and/or inclinations. This is the first step toward creating controller allowing to assist in tasks independent manner