Discrete Time Model Reference Adaptive Control of an Industrial Manipulator’s Waist Joint

Abstract

The current control systems adopted by most industrial manipulators are “open loop” designs based only upon kinematic consideration. Their servo gains are fixed and there is no flexibility of updating them. It is desirable, therefore, to have a robot control system that adapts to its payload and manipulator configuration in order to maintain optimal dynamic performance throughout the robot workspace, and regardless of the payload. This paper implements and evaluates the discrete time model reference adaptive control (MRAC) algorithm of robot control proposed by Horowitz and Tomizuka to a PUMA 560 industrial robot. Its original controller was overridden and the individual joint axes are controlled by a separate computer. This study shows the adaptive controller can maintain fast, consistent response regardless of the arm configuration.