The practical implementation of model reference robot control

Abstract

Model Reference Control (MRC) and adaptive MRC have been discussed at length by many authors. Despite the considerable volume of published work, no practical implementations of MRC by commercial robot manufacturers have been reported. The aim of the present paper is to examine the practical difficulties of implementing MRC on a robot and thus suggest reasons for the slow introduction of MRC. It is shown through numerical experiments that MRC provides effectively perfect control of a two degree of freedom manipulator provided that a correct dynamic model is assumed. However, it is clear that in physical systems there are a number of obstacles to practical implementation of MRC including motor voltage and current saturation, the requirement for frictional compensation and the possibility of errors in the dynamic models. Notwithstanding these difficulties, it is argued that MRC continues to hold potential for gross motion control in robot manipulators.