Design and Implementation of a Fuzzy-Based Gain Scheduling Obstacle Avoidance Algorithm

Abstract

This article presents a novel obstacle avoidance algorithm. Using a combination of fuzzy logic and gain scheduling theories, a new methodology that reduces computational costs compared to conventional fuzzy methodologies, specially when the variables to be controlled are too many. For comparison purposes, a potential field algorithm was implemented. Both algorithms are tested in a series of experiments to determine if the new algorithm is at least as good as the potential field algorithm. The metrics defined for these experiments are: the number of times that the agent collides (collisions), the time spent to finish a traced course (time spent) and the remaining stamina of an agent at the end of an experiment (stamina consumption). The results show that the proposed algorithm achieve a low level of collisions. Also, the proposed algorithm shows a considerable improvement in the time spent for the completion of the proposed tasks. Last but not least, the results demonstrate a considerable reduction in the stamina consumption using the proposed algorithm over the potential field algorithm.