Can muscle models improve FES-assisted walking after spinal cord injury?

Abstract

Some persons with a spinal cord injury can use functional electrical stimulation (FES) to walk again, but many cannot, and for those that can the walking obtained is limited. This paper argues that muscle models can help improve FES systems, but only if these muscle models are enhanced. Part 1 reviews differences between muscle models for FES systems and those for "natural" movement; FES models emphasize limb angle, demand simplicity, exploit feedback, and grade force through recruitment rather than rate coding. Part 2 tells how FES systems have used muscle models. Those that do not use muscle models to control stimulation do not fare well, although two recent ones (rule-based control and neural-net control with feedback) may yet do so. Those that do use muscle models provide good control initially, but fare poorly as the muscle properties change. Part 3 lists important questions that muscle models must address: questions of goal, type of activation, spasticity, simulation, simplicity, and fatigue. If these features can be incorporated, models can improve both the design and control of FES systems.