Generating optimal behaviors of mobile robot using behavior network with planning capability

This paper presents an approach to the optimal method for behavior generation of mobile robot. Recently, a hybrid system that supplements gap between reactive and plan-based approaches by employing a reactive system for lower level control for fast reaction and a planner for higher level for generation of optimal sequence of behaviors has gained popularity. Behavior network structures may generate behaviors automatically through the internal links and external links with sensors and goals, and can be applied much into more complex problems. In this paper we propose an optimization method of behavior sequences for generating optimal behavior of mobile robot using behavior network with planning capabilities. Behavior network inspired by behavior selection of animals selects the next behavior that has the highest priority from the information acquired to sensors and goals. Globally optimal behavior selected in this behavior network is processed in advance and the next behavior is selected among behavior that approaches the goals. We could notice that robot reaches the goals faster in behavior network with planning capability rather than behavior network only through the experiment using Khepera mobile robot simulator.