A Within-Stride Switching Controller for Walking with Virtual Constraints: Application to a Hybrid Neuroprosthesis

In this paper, we investigated if a switching controller and optimal virtual constraint design for a class of walking exoskeleton with redundant actuation can achieve a stable walking limit cycle. The motivation was to address control allocation problem in a hybrid walking neuroprosthesis that can obtain joint torques either from electric motor or functional electrical stimulation. A generalized switching controller is designed to switch between the motor and the muscle stimulation based on the fatigue state of the stimulated muscle. The controller is used to meet the virtual constraints designed to achieve a stable walking cycle and is proven to achieve exponential stability despite arbitrary switching. The virtual constraints were optimized based on the results of combination of genetic-particle swarm optimization algorithm and SQP method. A simple model of walking (3-link) that has active actuation at the hip joints was simulated to depict the controller feasibility.