Online learning neural network controller for pneumatic robot position control

Abstract

This paper presents the implementation of online learning neural network controller in the pneumatic robot position servo control. The advantages of this design include: the ability to compensate for nonlinearities, and it is insensitive to system parameter time-varying. The traditional PID controller is replaced by neural network controller trained online to learn the inverse model of the pneumatic manipulator by backpropagation of the performance error. The simulation studies and experimental results on the PID controller, online learning neural network controller and off-line training neural network controller, are presented and discussed.