Performance measures in the genetic design of digital controllers for robotic manipulators

Abstract

Genetic algorithms are used to design digital multivariable PID controllers for robotic manipulators for typical trajectory-tracking tasks when various different performance measures are used. It is thus shown that, by using an appropriate performance measure, the set of controller parameters can be readily found that determines the optimal time-domain trajectory-tracking behaviour for such tasks. This use of genetic algorithms is illustrated by the design of digital trajectory-tracking PID controllers for a typical three-degree-of-freedom robotic manipulator.