Robust robot control - a frequency domain approach

This paper is concerned with independent joint control of industrial robots. Despite the fact that a robot is a highly non-linear, time-varying multivariable system, this simple control structure is widely and successfully used in practice. Various alternative approaches have been proposed and investigated, adaptive (self-tuning) control (e.g. Lelic and Wellstead (1)) and non-linear compensation (e.g. Freund (2)) being the main directions of research. As such more sophisticated schemes are of course more demanding in terms of computing power and because of not yet resolved robustness problems, linear time-invariant (lti) independent joint controllers are preferable as long as they can provide the closed-loop performance which is required. The performance of the robot controllers which are commercially available however is not only limited by the fact that independent lti controllers are used but also by the fixed controller structure, usually cascaded controllers of PI-/PD-type. The pu rpose of this paper therefore is two-fold: 1) to provide a framework for the analysis of the potential of lti independent joint controllers; 2) to discuss control structures of lti independent joint controllers based upon the previous analysis. As an illustration of our approach, the control of one joint of the IR Kuka 160 is discussed in detail. (IITB)