Monitoring and Diagnosis of Multi-Agent Plan: Centralized Approach

Recently, a number of model-based approaches to monitoring and diagnosis of a Multi-Agent Plan (MAP) have been proposed; this fact signifies that the Artificial Intelligence (AI) community is interested in systems where activities are executed simultaneously by various agents. The current paper introduces a centralized approach for plan execution monitoring, where parallel operations are executed by a team of cooperating agents in a partially observable context. Each agent is monitored on-line by seeking all the possible evolutions of the agent both under nominal and faulty behavior and the system estimates the belief state at each time moment.