Statistical analysis of GA based PID controller optimization for robotic manipulator

The robotic manipulators have extensively been used in many industry like manufacturing, assembly etc., space and medical applications for the purpose of pick and place, positioning and path following etc. This requires accurate control of an end-effector of robotic manipulators as these are Multi-Input Multi-Output (MIMO), nonlinear and coupled systems. Being complex in nature tuning of the controllers is a critical issue and have been an area of research in the control community. In this paper, Genetic Algorithm (GA) is used for optimization of gains of the two Proportional-Integral-Derivative (PID) controllers applied to each link of a two-link planar rigid robotic manipulator. Performance indices namely Integral Square of Error (ISE) and Integral Square of Change in Controller Output (ISCCO) with equal weightage for both the links are considered for optimization. As GA is a global search technique, normally there are several solutions available in a given search space for same performance indices. The PID controllers used in the said two-link planar rigid robotic manipulator were optimized 2000 times and a statistical analysis is carried out on the obtained optimal gains and fitness function values to estimate the diversity of optimized solutions. The motivation behind this simulation study has been to provide an overall knowledge to the user for choosing the appropriate controller gains yielding suitable controller signals which should be preferred for system design specifications and cost effective hardware implementation of robotic manipulators while maintaining the desired tracking performance.