Multiple mobile robots - Fuzzy behavior based architecture and behavior evolution

Abstract

Fuzzy logic and genetic algorithms (GAs) have been successfully applied to various problems. In this thesis, interdisciplinary research works involving the fuzzy logic control (FLC), robotic system and genetic algorithms are presented. A comprehensive fuzzy behavior based architecture is proposed to control multiple robots in robot soccer system. Fuzzy logic is employed to realize all the behaviors and the behavior coordination. To further improve the system, an adaptive tuning methodology for this architecture is proposed. The tuning methods focus on the self-adjustment of fuzzy membership functions. With the help of a robot soccer simulator, genetic algorithm is used to evolve the fuzzy behaviors at different levels of the fuzzy behavior based architecture. The evolution results in performance improvements observed both in the simulation and real-world environments. Associated with the work on evolutionary fuzzy system, the DNA like coding methods for genetic algorithms are also developed and explored.