In recent years, the m edia has publicized accounts of paralyzed individuals “walking” , dancing, cycling, and performing coordinated m otor tasks through Functional Neuromuscular Stimulation (FNS) of their disused muscles. While these stories have been inspiring, they may have raised expectations regarding the capabilities of FNS beyond realistic levels. Some of these expectations focus on FNS as the means to restore the ability to walk;​ i.e. to support oneself and am bulate erectly using the legs as the prim ary drivers to the motion.

As experience with prosthetic devices has shown, the ability to walk with the appearance of normality is ultim ately desirable. Otherwise, a serviceable but conspicuous electrically-stimulated gait would probably find limited acceptance among the population of potential users. To date, the feasibility of utilizing FNS to elicit coordinated movements from paralyzed muscle in order to restore natural gait has never been formally demonstrated. This study examines the issue of engineering feasibility by sim ulating the movements of an artificially-controlled, dynamic, bipedal human walking model on a digital computer. The 3-D, 8 DOF musculoskeletal model includes the mechanical properties of the joints and ligaments, as well as many of the currently-known characteristics of artificially-stimulated musculotendon actuators. Specifically, the goal was to examine whether minimal sets of muscles could be used in order to generate approximately normal gait trajectories without requiring either high levels of force or unduly precise control of muscle activation. It is believed th at the process of sim ulating natural gait with this model will serve to highlight difficulties that will be encountered later should clinical trials come to pass.

The thrust of this thesis shows th at it may be possible to enable undisturbed, near-normal gait on level surfaces provided the artificially-stimulated ankle plantarflexors can be reconditioned to near-normal strengths. Presuming sufficient plantarflexion moments can be developed via orthoses and/or muscle actions, ten muscle groups were needed to sustain a simulated, unsupported step with normal appearance and speed. A novel approach employing dynamic programming was developed in order to initially coordinate the patterns of muscle activity. Thereafter, a trial-and-error process was utilized to more finely tune the artificial “stimuli” until the computer-animated movements exhibited near-normal characteristics.

However, the extreme difficulty of coordinating the limb motions to achieve stability in the stance leg and hip, adequate foot-floor clearance in the swing leg, and proper length and timing in the artificially-controlled step suggests that restoring natural gait is highly im probable once voluntary lower limb motor control has been lost. This conclusion is supported by calculations quantifying the highly-coupled nature of the linked segment model, which emphasize the need for extraordinarily fast and robust global controllers.