Imperfect Situation Awareness: Representing the Role of Error and Uncertainty in Modeling, Simulation & Analysis

Abstract

Representation of Situation Awareness/Situation Understanding (SA/SU) is now an acknowledged need in computer simulations of military operations. An important aspect of such representation is decision-making with SA/SU that may be incomplete and/or erroneous. The goal of this research is to apply engineering methods to improve representation of SA/SU in such simulations, which at present rely too heavily on model “omniscience” in the data used to support decision-making and its effects.The approach employed uses agent-based modeling and simulation to represent military operations. Geo-spatially “lost” intelligent agents are simulated as trying to navigate in a virtual world. Each agent has a unique dynamic “mental map” - its idiosyncratic view of its geo-spatial environment. Its decisions are based on this idiosyncratic view, but behavior outcomes are based on ground truth. The rate and degree to which an agent’s expectations diverge from ground truth provide measures of how good its SA/SU is, and simulation outcomes provide measures of the operational effects of different levels of SA/SU. Being “lost” can be studied as a concrete geo-spatial condition. It can also serve as a metaphor for uncertainty and/or inaccuracy providing a foundation for exploring the more metaphorical aspects of decision-making under uncertainty.