Compact Cerebellar Model Articulation Controller for Ultrasonic Motors

This paper presents a compact cerebellar model articulation system for the position control of ultrasonic motors. Since the ultrasonic motors possess a time-variant and heavy nonlinearity and a variable dead-zone characteristic, it is difficult to create a precise model of ultrasonic motors. To overcome this problem, robust adaptive control technique is often utilized for the control of ultrasonic motors. However, it requires solving complicated mathematical equations to find the adaptive control law. Moreover, due to the variable dead-zone characteristic, on-line adaptive control technique needs more complicated algorithms to overcome the influence of dead-zone. As a result, the real-time control realization often needs a high-cost microcomputer to calculate the complicated mathematical equations or algorithms. Based on the proposed compact cerebellar model scheme, the appropriate control input can be generated only by calculating the sum of two activated memory cells. Therefore, the simplicity of mathematical equations reduces the computation loading of the controller and eliminates the need for the high-cost microcomputer. Simulation and experiment results are provided to verify the effectiveness of the proposed method.