Pattern Recognition Approach for Swarm Robots Reactive Control with Fuzzy-Kohonen Networks and Particle Swarm Optimization Algorithm

Abstract

A simple reactive control-based pattern recognition approach for swarm robots is presented. Fuzzy-Kohonen Networks is combined with Particle Swarm Optimization (FKNPSO) for avoiding static and dynamic obstacles, permitting flocking in the group, and efficiently planning the route to the target. A simple algorithm with less computational resources is developed that will produce good performance in terms of ability to avoid the obstacle in an unknown environment, maintain movement, and have a faster time to reach the target. Three simple robots is demonstrated in a series of practical tests in an unknown environment to see the effectiveness of the proposed algorithm. Results showed that the swarm robots successfully performed several tasks, were able to recognize the environment only use seven rules, and produced a small number of resources. The algorithm provides a much faster response to expected events compared to FKN and Fuzzy-PSO and allows the mobile robot to move to the target without collisions.