Identificaion of Fuzzy Control Rules Utilizing Genetic Algorithms and Its Application to Mobile Robot

Abstract In this paper, an approach to identify fuzzy control rules is presented. The decision of the optimal number of fuzzy control rules is made by means of soft c-means clustering which is a clssification oriented approach based on quantitative information of a system and produces fuzzy clusters that are optimal in a generalized least squared error sense. The identification of the parameters of fuzzy implications is carried out by use of genetic algorithms which display an excellent robustness in complex optimization problems. For the efficient and fast parameter identification, the reduction technique of search areas of genetic algorithms is proposed. The feasibility of the proposed approach is evaluated through the identification of the fuzzy model to describe an input-output relation of Gas Furnace. Utilizing the navigation data of a mobile robot obtained from the expert's control actions to follow the center of a corridor, the fuzzy control rules are identified and applied to the navigation control.