INVESTIGATION OF TRAJECTORY TRACKING RECURRENT NEURAL CONTROLLER FOR ROBOT MANIPULATOR EMPLOYING HIL SIMULATION TECHNIQUE

Abstract

In this work, recurrent neural network based controller is implemented for controlling Robot Manipulator using Hardware in the Loop (HIL) simulation technique. The gains of the PID based recurrent neural network controller scheme are initialized with the cuckoo search algorithm (CSA) optimization method rather than assuming randomly. The Least Mean Square (LMS) adaptive algorithm is then investigated for the online adaptation of the gains of the controller. The performance of the designed controller is tested against the plant parameters uncertainties and external disturbances for all the links of a three link rigid robotic manipulator with variable payload. The stability analysis of the presented control system is investigated out using Lyapunov's approach. The Simulink model of the robotic manipulator has been developed using Matlab-Simulink software and the performance of the controller implemented using the HIL technique in C2000 real time controller was analyzed. From the HIL simulation for trajectory tracking results it is evident that the dominance and effectiveness of the CSA optimized recurrent neural network PID controller (RNPID) over the optimized neural network PID (ONPID), optimized PID (OPID), and PID controllers for variable payload and disturbance rejection.