Self adaptive neuro-fuzzy control of neural prostheses using reinforcement learning

Preliminary results are presented of a new functional electrical stimulation (FES) control methodology based on an adaptive fuzzy network using supervised and reinforcement machine learning techniques. The FES application example used to test these controllers used a computer model of swing phase assisted by a powered hybrid FES orthosis. The supervised learning controller was trained to a predetermined control strategy, and converged in approximately 15 trials. Using very simple reinforcement signals sent at the end of every swing phase, the reinforcement learning controller was able to develop a unique control strategy in approximately 150 trials. The reinforcement learning controller had the additional ability to continually re-adapt to changes in the system parameters which caused the other controller to fail.