Fuzzy identifier based inverse dynamics control for a 3-DOF articulated manipulator

Abstract

A method using fuzzy logic and neural network structures for the identification and control of robot dynamics is presented in this paper. Fuzzy identifiers are employed to match the effects in the robot dynamics model. Fuzzy logic systems used are represented as three-layer feedforward neural networks. Fuzzy system parameters are adjusted via backpropagation. An on-line tuning algorithm is employed. Simulation results obtained with a three degrees of freedom articulated arm are given.