A self-organized fuzzy controller for wheeled mobile robot using an evolutionary algorithm

Designing the controller of a wheeled mobile robot is not as easy as might be supposed, on account of nonholonomic constraints. To overcome such difficulties and gain more accurate position and velocity control, a self-organized fuzzy controller is proposed. To find solutions of optimal fuzzy input and output membership functions and to determine a rule base, an evolutionary process is proposed. The procedure that derives this solution is composed of three steps, each step having its own unique evolutionary process. The elements of an output term set are increased first, and then the rule base is varied according to increase of the elements. The varied fuzzy system competes with a system that has no element increase. If the varied fuzzy system loses in competition, then the system naturally disappears. On the other hand, if the varied system survives, the fitness with each increased element of the output term set and rule base is tested and unnecessary parts are removed. After having finished regulation of output term set and rule base, searching for input membership functions is processed with constraints to reduce the unsuitability of the system. The searching constraints do not produce a 0 membership value for any input. After completing the search for the input membership function, fine tuning of output membership functions is processed