SEMG-BASED NEURO-FUZZY CONTROLLER FOR A PARALLEL ANKLE EXOSKELETON WITH PROPRIOCEPTION

The harmonious mechanism design and effective motion control are the central problems of a lower extremity exoskeleton, especially for the ankle exoskeleton which has to fulfill the requirements of higher rigidity, lighter weight and three degrees of freedom (DOFs) motion in limited space. A novel ankle exoskeleton for assisting physical recovery in ankle joint is developed in this paper. An ankle exoskeleton with three revolute-prismatic-spherical (RPS) parallel mechanism is optimized by mimicking human ankle actuated by parallel muscles. A neuro-fuzzy controller integrating electromyographic (EMG) sensor and artificial proprioceptor, which imitates the closed-loop control system of human body, is developed to realize real-time control of the ankle exoskeleton. And the fuzzy neural network of the controller combines fuzzy rules established based on anatomical knowledge and the results of previously performed experiment with hybrid learning algorithm. It is built to decode the human motion in real time by the fusion of the fuzzy EMG signals reflecting human motion intention and the precise proprioception providing joint angular information feedback. Corresponding experimental results demonstrate that the parallel ankle exoskeleton meets the kinematical and dynamical requirements of ankle joint, and the neuro-fuzzy controller with proprioception is accurate and effective.