FUZZY LEARNING CONTROL FOR A FLEXIBLE-LINK ROBOT

Fuzzy control has emerged as a practical alternative to several conven tional control schemes since it has shown success in some application areas; however, there are several drawbacks to this approach: i) the d esign of fuzzy controllers is usually performed in an ad hoc manner wh ere it is often difficult to choose some of the controller parameters (e.g., the membership functions), and ii) the fuzzy controller constru cted for the nominal plant may later perform inadequately if significa nt and unpredictable plant parameter variations occur. In this paper w e illustrate these two problems on a two-link flexible robot testbed b y i) developing, implementing, and evaluating a fuzzy controller for t he robotic mechanism, and ii) illustrating that payload variations can have negative effects on the performance of a well designed fuzzy con trol system. Next we show how to develop and implement a ''fuzzy model reference learning controller'' (FMRLC) [1]-[5] for the flexible robo t and illustrate that it can: i) automatically synthesize a rule-base for a fuzzy controller that will achieve comparable performance to the case where it was manually constructed, and ii) automatically tune th e fuzzy controller so that it can adapt to variations in the payload s o that it can perform better than the manually constructed fuzzy contr oller.