筋電位の電気力学的遅延を用いた零位相追従型マンマシンインターフェースの開発

Abstract

This study aims to develop a man-machine interface which realizes the perfect tracking to human motions. With such an interface, a user may operate machines with improved maneuvering feel. The perfect tracking can be realized by utilizing Zero Phase Error Tracking Control (ZPETC) to eliminate or reduce phase lags including mechanical and communication delays in systems. In general, ZPETC however requires future information as reference signals. We propose to utilize a time difference from arising electromyogram (EMG) signals in muscles until appearing the corresponding actual human motion. The time difference is called Electro Mechanical Delay (EMD). Hence, if human motions can be estimated from electromyogram signals, we can obtain the future information on the human motions by EMD. Our system focuses on wrist motion as commands toward a system. Utilizing EMD, the wrist angle is estimated from EMG signals by a nonlinear delayed ARX (NDARX) model. A discrete-time Model Following (MF) controller is used for letting the system track the wrist motion. Consequently, this demonstrates the possibility that the controlled system can move tracking the wrist motion with no or less lag with respect to the practical human motion. Friction compensation plays an important role from the implementation point of view, because ZPETC is a feedforward controller for discrete-time linear systems but the real system is affected by friction. The effectiveness is verified through numerical simulations and experiments comparing tracking performance using only a PID controller with the proposed control system in terms of cross-correlation analysis.