A Decentralized Architecture Using the Null-Space-Based Behavioral Control For Multi-Agent Systems with Application to Border Patrolling, Search and Retrieval

Abstract

This report proposes a new method for searching and retrieving a dynamically changing border using robotic multi-agent systems as an extension to the patrolling task - the problem of coordinating autonomous agents patrolling a border. The proposed method is a decentralized and self-organizing system built in the framework of the Null-Space-Based behavioral control and is controlled by a Finite State supervisor. The system encourages cooperation among the agents without the use of explicit and direct communication and from the simple interactions between agents the system hopes to emerge a global behavior in order to help fulfill the patrolling task.