Fuzzy Neural Network Impedance Control of Soft Lower Limb RehabilitationExoskeleton Robot

To solve the motor dysfunction due to the stroke or the traffic accidents, a wearable soft knee exoskeleton robot for rehabilitation training is designed. Based on the emphatic introduction of the Hill-muscle-model-based tendon-sheath artificial muscle drive system design and real-time control platform, the derivation process of the fuzzy neural network impedance control algorithm is analyzed. Finally, the rehabilitation training experiment is carried out in the human-machine cooperative training mode, in the conditions of the fixed and the variable impedance parameter control strategies. The influence of the rehabilitation exoskeleton system on the muscle activity of subjects is also compared and analyzed. The experimental results show that the flexor/extensor torques are increased by 9.70% and 9.06% during the constant frequency and amplitude training, and the flexor/extensor torques are increased by 88.34% and 57.68% during variable frequency and amplitude training. Therefore, the stability and the safety of the lower limb rehabilitation robot can be enhanced by selecting the suitable impedance model that conforms to the physiological muscle stiffness characteristics, as well as the human-robot interactive compliance and coordination can be improved.