Integrating a Potential Field Based Pilot into a Multiagent Navigation Architecture for Autonomous Robots

In this paper we present a new Pilot for a Multi-Agent based control architecture for Autonomous Robots. This Pilot is based on the use of virtual potential fields and is easy to implement yet effective. The Pilot functions as an autonomous agent in a complex Multi-Agent Architecture for the control and navigation of an autonomous robot. In this architecture, various agents are responsible for different tasks, and they might have to compete and cooperate for the successful completion of a particular navigation mission. The interaction between different agents is achieved through the use of a bidding mechanism. In this respect, we also present a way to define the bid for the new pilot, so that the pilot can be integrated easily into the Multi-Agent Architecture. The pilot has been tested both on simulations and navigation experiments involving a real robot and it has given successful and encouraging results. We also deal with some problems of this pilot and ways to get around them.