Application of genetic algorithm to decentralized control of robot manipulators

This paper discusses the genetic algorithm to improve a decentralized adaptive control scheme to reduce tracking errors of robot manipulators. A simple PD conventional controller is used together with a cubic feedback and adaptive controller to ensure its global stability. A genetic algorithm is then proposed to determine the coefficients of the PD, nonlinear and adaptive controller such that the tracking errors of the robot, that is fitness function of genetic algorithm, are minimized. In order to show the performance of the proposed method, computer simulations are implemented on a simple two link robot manipulator.