Control of bipedal locomotion assisted with functional electrical stimulation

Functional electrical stimulation (FES) can be used for restoring of the standing and walking of humans with spinal cord injury. Existing control algorithms typically use a dynamic model of the human body and methods of analytic mechanics. Available models do not take into account the characteristics of the segmented spinal column, upper body, neck and the head, arm support over crutches or walker, and substantial changes of the neuro-musculo-skeletal characteristics caused by the injury; hence, control does not work well enough. The alternative method for controlling bipedal walking is cloning of a walking pattern of an able-bodied subject. Walking with a constant speed, over even terrain, with no or small perturbations is a cyclic activity, which can be considered as a finite automaton. The automaton includes sensory inputs and motor (muscle) outputs. The method of cloning of this automaton includes two phases: 1) simulation of a customized biomechanical model of a potential user walking with a pattern of an able-bodied subject; 2) determining the mapping by a radial basis function neural network between the input (trajectory) and outputs (simulated muscle activation).