NN-compensator with fuzzy based switching for controlling 3-DOF underactuated manipulator

A novel control algorithm for controlling underactuated robot manipulators using fuzzy based switching with an NN-compensator is presented. One fundamental feature of this approach is to employ elemental controllers in order to fulfill the ultimate control objective, without performing rigorous linearizations or any other augmentation methods to the original nonlinear system. Parameters of both antecedent and consequent parts of a fuzzy indexer and the compensator are optimized off-line by using evolutionary computation. Such parameters are trained for different initial configurations of the manipulator and the common rules can be brought into the online operation of underactuated manipulators. The effectiveness of the proposed method is demonstrated using a computer simulation.