F-1232 Learning of Multi Objective Behavior Coordination of A Mobile Robot
Hiroyuki Masuta, Naoyuki Kubota, Kazuhiko Taniguchi · The proceedings of the JSME annual meeting · 2001
Recently, intelligent robots have been used in various fields. Therefore, a robot should perform flexible perception and decision making. Based on the concept of the structured intelligence, we have proposed a sensory network for the perception and multi-objective behavior coordination (MOBC) for the decision making of the robot. However, the multi-objective behavior takes much time comparing to subsumption architecture, since the multi-objective behavior coordination makes outputs based on the inference results of multiple behavior rules. From the viewpoint of real-time control, the robot should quickly take feasible actions, not optimal actions. Therefore, this paper proposes a control method of the behavioral dimension in multi-objective behavior coordination of the mobile robot. We discuss the effectiveness of the propose method through several experiments of the mobile robot.