Multi-Agent System for Threat Assessment and Action Selection under Uncertainty and Ambiguity

This paper describes an approach to designing a multi-agent decision support system, which monitors uncertain and ambiguous dynamic environments to produce probable explanations of the current situation and identify asymmetric threat as well as selects actions to prevent catastrophic consequences or mitigate their impact while acting under resource and time constraints. The proposed system incorporates decision makers' attitude toward risk and uncertainty in situations involving regular and high consequence rare events. An example scenario showing potentials of the described approach is also presented.