An Agent-Based Model for Autonomous Planning in Distrubuted Critical Systems

Critical distributed systems are usually employed in situation where automation is appealing when dealing with dangerous situations. Anyway, the complexity of these systems is growing up faster and faster. Distributed autonomous elements like sensors networks, drones etc. are now widely used in critical system. In this systems, each autonomous element must implement procedures in order to reach their goals in a distributed, usually collaborative environments. New planning algorithms have to be coped with existing ones in order to fulfil performance and soundness requirements. In this paper we propose an autonomous planning methodology that is based on multi-agent models. The planning methodology exploits both classical and a new counter-example based approaches in order to build an effective multi-expert system able to face with increasing complexity of these systems. We also describe a framework we implemented in order to enact the methodology and we will present results from the study of a critical distributed system. © 2015 IEEE