MODELING AND CONTROL OF 5DOF ROBOTIC ARM USING NEURO-FUZZY CONTROLLER

Abstract

Robotic Arms are considered essential components of any automation system. They present considerably complicated electromechanical systems with mutual interactions of robot mechanics and drives, at design of which the mechatronic approach should be taken into consideration. The modeling problem is necessary before applying control techniques to guarantee the execution of any task according to a desired input with minimum error. The physical modeling in the SimMechanics / SIMULINK environment facilitates simulation efforts of such complex systems by seamless interfacing of ordinary Simulink block diagrams. This is not only more intuitive, it also saves the time and effort in deriving the equations of motion. Problem of Inverse Kinematics (IK) is solved using a machine learning technique i.e. Adaptive Neuro-Fuzzy Inference System (ANFIS) in contrast with the analytical solution. MATLAB, Simulink, SimMechanics and SolidWorks are used as simulation platform. This research will undertake the following five developmental stages; firstly, the complete computer-aided design (CAD) model of a 5 DOF robotic arm is developed in SolidWorks. In the second stage, the CAD model is to be converted into physical modeling by using SimMechanics Link. Then, the ANFIS networks are trained to compute the inverse kinematics of the robot arm. In the fourth stage, the research intends to perform the simulation in which, the trajectory tracking of robot manipulator’s end effector is considered as a test scenario. In last stage, the performance parameters of implemented technique are studied i.e., residual plots, convergence plots, comparison between the predicted results and analytical solution, analysis of trajectory and dynamics of robotic arm, joint torque computation through SimMechanics.