Application of a neural network based classifier system to AGV obstacle avoidance

This paper describes the application of a neural network based classifier system to the control of a simulated autonomous guided vehicle (AGV) in a simple obstacle avoidance task. A mechanism for concurrent exploration and exploitation of a maze environment based on Kohonen feature maps is proposed and its implementation in a hybrid learning system is described. Two variations of the underlying connectionist path structure are presented and their performance analyzed in a simple simulated maze environment.