Redundant Manipulator Control System Simulation with Adaptive Neural Network and Newton-Raphson Refinement Algorithm

Abstract

In the paper we consider modeling of a manipulator that is based on hybrid method of solving inverse kinematics (IK) problem with using adaptive neuro-fuzzy inference system (ANFIS) and an algorithm for iterative refinement by Newton-Raphson method. The process of control system synthesis for the multi-link redundant manipulator with the use of a programmable logic controller (PLC) is also demonstrated. The presented control system provides controlled accuracy of calculations in real-time systems. This method of solving IK problem could be applied for various constructions of manipulator with different parameters, this advantage makes the control system cross-platform. Simulation of the developed control system was performed in the Matlab environment. A mathematical equations of the constructing the manipulator workspace and an example of training neural networks of the control system are given. The results of the solution of IK problem are presented.