Optimization of granulation for fuzzy controllers of autonomous mobile robots using the Firefly Algorithm

This paper describes a methodology based on optimal granularity allocation for fuzzy system design, and the main contribution is a method, based on the Firefly Algorithm, to generate and test information granules for fuzzy controllers of autonomous mobile robots. The Firefly Algorithm automatically generates and tests these granules which are defined by the parameter values of the membership functions, which are evaluated based on simulations of the robot plant and the final result is an ideal combination of information granules. The evaluation is made with a comparison of the actual trajectory generated by the fuzzy controller of the robot with respect to the desired path. To verify that the obtained results are significantly better, a statistical test is performed between the firefly and the genetic algorithms.