An Evolved Fuzzy Reactive Control System for Co-Operating Autonomous Robots

Abstract

Autonomous robots operating in unstructured environments clearly require some capacity for adaptive behaviour. This paper presents a method for achieving an adaptive response based on a multi layer control architecture. The primary level of control is composed of the Behaviour Synthesis Architecture [1] which is continuously modfied by a secondary layer, composed of a Fuzzy Logic controller. The Fuzzy rule base adjusts the relative weighting (utility) of each behaviour to allow the robot to “focus” its response on the current sensory stimuli. This allows the robot to balance conflicting behaviours and goals in a dynamic environment. The Fuzzy rule base is evolved by genetic algorithms in a simulation of the robot and environment. The advantages of the genetic search algorithm are therefore combined with a set of high level rules and behaviours to resolve the conflicting dynamics of two co-operating mobile robots.