Learning Control Applications to Mechatronics

Abstract

In this paper we discuss the use of function identification and adaptive control algorithms in learning controllers for mechatronic systems. The term learning is used because the development of the algorithms is motivated by the emulation of the human ability to improve motor skills through training. The learning algorithms are based on the representation of an unknown function as a linear integral transform with a known kernel and an unknown influence function. Identification of the unknown function proceeds indirectly by identifying the unknown influence function. The application of these learning algorithms to the design of repetitive controllers for disk file systems and learning controllers for robot manipulators is also presented.