Modeling Distributed Interactive Agents

Conflict hinges on human behavior, but today's combat simulations only represent behavior in terms of heuristics. Yet these rule-based representations. fail to include human variability earning, and being intelligently interactive. They do not take advantage of an adversary's mistakes. An empirical modeling of behavior is difficult because behavior is intent and situation dependent. In contrast, a normative approach can be used with success, based on having the mission expressed in quantitative terms coupled with an evolutionary computation that discovers increasingly appropriate ways of allocating the available resources at each moment. Personality factors alter the way the mission is understood, as well as the performance to achieve that mission. The concerns here can be modeled across the levels of command and degrees of allegiance. This project will demonstrate the feasibility of non-rule-based intelligently interactive combat simulation that includes meaningful representation of human behavior. It will indicate how this can be incorporated into distributed interactive simulations using the High Level Architecture most suitable for training and analysis. BENEFITS: Including human behavior in combat simulation will improve both training and analysis. When incorporated into a non-rule-based intelligently interactive simulation, it provides a calibrated OFFOP, thereby reducing the need for and the cost of using qualified personnel while also increasing realism. This can be used for training at all levels. It can provide a basis for combat decision support during exercises/simulations and in actual combat. It can also improve the evaluation of new technology.