Control of multi-DOF ultrasonic motor using neural network based inverse model

A multi-DOF ultrasonic motor developed by the authors is capable of generating multi-DOF rotation of a spherical rotor using three orthogonal natural vibrations of a bar-shaped stator. The ultrasonic motor is suitable for making small, light weight and simple multi-DOF motion unit. In the present paper, a control methodology for the multi-DOF ultrasonic motor is proposed and a motion control test using the method is conducted. First, an inverse model of the multi-DOF ultrasonic motor is developed using our prior knowledge in ultrasonic motors and neural network technique in order to deal with the redundancy and non-linearity of driving characteristics. Second, a novel control method using the inverse model is proposed. Then, motion control tests are conducted to confirm the ability of the proposed control method. The results confirm that the rotor can be rotated around arbitrary rotational axis using the inverse model, and that the proposed control method does work successfully.