A multi-agent structure system for the navigation of a mobile robot

A multi-agent structure system for the navigation of a mobile robot is presented. An agent is used for modelling an elementary behavior. A method derived from potential field is proposed. Every agent has a similar output data structure for an easy integration of the system and making agents independent each other. The interactions between agents are achieved by means of weights. Emergent behaviors have been observed by the interaction of behavioral agents. The multi-agent system can be setup according to the mission of the mobile robot or changed dynamically in response to the modification of the environment.