Electromyography (EMG)-signal based fuzzy-neuro control of a 3 degrees of freedom (3DOF) exoskeleton robot for human upper-limb motion assist

An electromyography (EMG) signal based fuzzy-neuro control method is proposed in this paper for a human upper-limb motion assist exoskeleton robot. The upper-limb exoskeleton robot (named W-EXOS) assists the motions of human forearm pronation/supination, wrist flexion/extension and ulnar/radial deviation. The paper presents the EMG signal based fuzzy-neuro control method with multiple fuzzy-neuro controllers and the adaptation method of the controllers. The skin surface EMG signals of muscles in the forearm of the exoskeleton user and the hand force/forearm torque measured from the sensors of the exoskeleton robot are used as input information for the controllers. Fuzzy-neuro control method, which is a combination of flexible fuzzy control and adaptive neural network control, has been applied to realize the natural and flexible motion assist. In the control method, multiple fuzzy-neuro controllers are applied, since the muscle activation levels change in accordance with the angles of motions. The control method is able to adapt in accordance with the changing EMG signal levels of different users. Experiments have been performed to evaluate the proposed control method.