Static and Dynamic Multi-Robot Coverage with Grammatical Evolution Guided by Reinforcement and Semantic Rules

Abstract

In recent years there has been an increasing interest in the application of robot teams to solve some kind of problems. Although there are several environments and tasks where a team of robots can deliver better results than a single robot, one of the most active attention focus is concerned with solving coverage problems, either static or dynamic, mainly in unknown environments. The authors propose a method in this work to solve these problems in simulation by means of grammatical evolution of high-level controllers. Evolutionary algorithms have been successfully applied in many applications, but better results can be achieved when evolution and learning are combined in some way. This work uses one of this hybrid algorithms called Grammatical Evolution guided by Reinforcement but the authors enhance it by adding semantic rules in the grammatical production rules. This way, they can build automatic high-level controllers in fewer generations and the solutions found are more readable as well. Additionally, a study about the influence of the number of members implied in the evolutionary process is addressed.