T-Norm adaptation in fuzzy logic systems using genetic algorithms

This paper investigates the performance of fuzzy inference systems having parameterized T-Norms in control of robotic manipulators. The adaptation of controller parameters is carried out by genetic algorithms. The error and the derivative of error are utilized in the decision process. The chromosomes, which include the adjustable parameters, are updated periodically by reproduction, crossover and mutation. Conventional reproduction and crossover methods and simulated-annealing type mutation are applied to find best-fit chromosomes. The efficiency of the proposed method is observed on a two degrees of freedom direct drive SCARA manipulator. It is seen that the proposed approach results in a distinguished performance comparatively to those using gradient based strategies.