Genetic-fuzzy approach in robot motion planning revisited: rigorous testing and towards an implementation

Abstract

This study endeavors to analyze the GA-Fuzzy approach and is an attempt to make it more practical, real-time and less computationally intensive. A new hybrid technique of using the GA-Fuzzy approach with a local search technique is presented and its efficacy viz a viz the GA-Fuzzy approach alone is demonstrated. It makes the existing technique more robust, efficient and computationally less expensive. The GA-Fuzzy approach converts an online problem of finding an obstacle-free, time optimal path for a mobile robot to an offline problem of optimizing a fuzzy rulebase using a genetic algorithm. Simulation results demonstrating the efficacy of the new design are presented.