Formulating situation awareness for multi-agent systems

Abstract

Improving autonomous agents' awareness has aroused a great deal of interests in agent research. The objective is to formulate an approach that provides advanced awareness capabilities in the agents' behaviour. In this paper, we explore the techniques proposed in the literature that have been adopted to enable agents' awareness which are Active Perception (AP), Situation Awareness (SA) and Context Awareness (CA). In addition, we illustrate the possible integration of the techniques' core phases with a multi-agent system (MAS). In doing so, we propose an approach to Situation Awareness for a decentralized MAS environment which is abbreviated as SAMAS. The approach assists in the agents' decision-making process through decomposing the context of an event's situations. Consequently, the SAMAS approach provides human-based assessments to the agents' performance. The assessments are fed back to the agents' to maintain the active perception of their actions. The paper contributes an independent SA approach considered from the perspective of MAS in a dynamic environment. © 2013 IEEE.