TWO DOF ROBOT CONTROL WITH FUZZY BASED NEURAL NETWORKS

摘要

In this study, trajectory control of robotic arm which has two degrees of freedom (DOF) is conducted by using the control methods of Proportional-Derivative (PD), Adaptive Neuro Fuzzy System (Anfis), hybrid PD-Anfis and its performance analysis is carried out. In the design of the robot, forward kinematics, inverse kinematics and dynamic equations are used. Firstly, the PD controller is executed, and then the PD controller and Anfis controller are compared applying to a different controller approach with the Anfis of Matlab/Simulink software. The positive and negative sides of the Anfis controller are compared and hybrid PD-Anfis controller method is conducted as a different approach to eliminate the negative sides. While the system constants K p and K v are kept constant by the classical PD control method, the output of PD controller is trained with Anfis in the new method and the output value is adjusted according to the error and the change rate of the error. By this way, outputs which have less error rate and which are able to follow the reference better are obtained.