Development of man-machine interface utilizing myoelectricity

This paper aims to develop a man-machine interface utilizing myoelectricity as input signals. The accurate relationship between an operator's intention and myoelectricity arising as the result of the intension is required to realize such an interface. In our study, paying attention to the wrist motion, an NARX model from myoelectricity as input to the wrist angle as output is derived, and is utilized to estimate the wrist angle from myoelectricity measured by EMG sensors. The Lebesgue-type identification method for NARX models is used to obtain the model. Control of unstable systems is a difficult task in general, hence, the concept of safe manual control is combined with our interface. The effectiveness of the proposed man-machine interface is verified by experiments in which the interface is used for control of a Furuta pendulum.