Rule-based control for a flexible-link robot

Abstract

This paper presents a design and implementation case study that focuses on endpoint position control of a two degree-of-freedom robot with very flexible links. Linear and nonlinear conventional control techniques have been shown to be somewhat successful in achieving various control objectives for the laboratory test beds of this and many other studies; however, their reliance on an accurate mathematical model of the process often limits their chances of achieving good endpoint position control. Here the authors investigate an alternative to conventional approaches where they employ rule-based controllers to represent and implement two general forms of knowledge that they have about how to best control the mechanism: i) experience gained from the use of a mathematical model and conventional control; and ii) an intuitive understanding of the dynamics of the two-link flexible robot. The authors begin the case study by assuming that the controls for the two links of the robotic mechanism can be designed and implemented independently and investigate the performance of rule-based fuzzy controllers which only use simple intuitive knowledge about how to control two independent links. Next, the authors show that if the rule-base is augmented with knowledge about the coupling effects between the two links, the controller can achieve improved performance over the uncoupled case. The final portion of the case study, which represents the authors' primary contribution, investigates the use of a two-level hierarchical rule-based controller with a simple upper-level "expert controller" that captures knowledge about how to supervise the application of low-level fuzzy controllers during movements in the robot workspace. Overall, the rule-based supervisory control results have proven to be extremely effective for vibration suppression in the laboratory test bed of this study, comparing favorably (in terms of performance, design complexity, and implementation issues) to a variety of conventional techniques attempted to date (including linear robust designs, feedback linearization, input command shaping, and adaptive approaches).< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>