Mission Plan Recognition: Developing Smart Automated Opposing Forces for Battlefield Simulations and Intelligence Analyses; Conference paper

摘要

A key challenge for battlefield simulation is the estimation of enemy courses of action (COAs). Current adversarial COA development is a manual time-consuming process prone to errors due to limited knowledge about the adversary and its ability to adapt. Development of decision aids that can predict adversary's intent and range of possible behaviors, as well as automation of such technologies within battlefield simulations, would greatly enhance the efficacy of training and mission rehearsal solutions. In this paper, we describe the development of OPFOR agents that can intelligently learn BLUEFOR's mission plan. This knowledge will allow OPFOR agent to reason about the intent of BLUE and counteract accordingly to prevent/influence the future BLUEFOR's operations by affecting current operations, challenging BLUE's resources, and preparing OPFOR for future battles.