Learning of membership functions of fuzzy behaviours for a mobile robot control system

This paper introduces a learning algorithm to automatically learn membership functions of fuzzy behaviours for a mobile robot reactive control system. Genetic algorithms are used to implement the learning processes. The learning process emphasises the functionalities of individual behaviours, various types of training environments and a simple-to-complex multistage learning course for the purpose of building a real world mobile robot. The effectiveness of the algorithm is presented