Design of self-learning controllers using expert system techniques

Abstract

The use of real-time expert system techniques to control systems, including robot manipulator control systems, is discussed. A novel type of intelligent controller structure, the expert learning controller prototype called ELEC (expert learning controller), is developed for the control series optimization of trajectory tracking problems in repeat operations. The ELEC, acting as an intelligent real-time controller in a closed-loop system, can modify the control series in a human-expert-like way using the experience of previous operations in order to force the system output to converge to the previously given desired trajectory. With the self-learning functions, the ELEC does not require the knowledge of system models; thus, it can be used in a wide range of control problems, especially in robot control. Numerical examples and simulation results of nonlinear, time-varying and multiple-variable robot systems are given to show the satisfactory performance of ELEC.<>