NEW CONTROL STRATEGIES FOR NEUROPROSTHETIC SYSTEMS

The availability of techniques to artificially excite paralyzed muscle s opens enormous potential for restoring both upper and lower extremit y movements with neuroprostheses. Neuroprostheses must stimulate muscl e, and the artificial movements produced. accomplish tasks include the se feedforward (open-loop), feedback, and adaptive control. Feedforwar d control requires a great deal of information about the biomechanical behavior of the limb. For the upper extremity, an artificial motor pr ogram was developed to provide such movement program input to a neurop rosthesis. In lower extremity control, one group achieved their best r esults by attempting to meet naturally perceived gait objectives rathe r than to follow an exact joint angle trajectory. Adaptive feedforward control, as implemented in the cycle-to-cycle controller, gave good c ompensation for the gradual decrease in performance observed with open -loop control. A neural network controller was able to control its sys tem to customize stimulation parameters in order to generate a desired output trajectory in a given individual and to maintain tracking perf ormance in the presence of muscle fatigue, The authors believe that pr actical FNS control systems must exhibit many of these features of neu rophysiological systems.