A stochastic agent approach (SAA) for mission effectiveness

Wargaming;Stochastic;Agent-based model;Agent based model;Artificial intelligence;Decision tree;Mission simulation;Mission space exploration;Surrogate model;Lanchester equations;Lanchester law;Engagement analysis;Engagement simulation

Gordon, Seth Edward ; Mavris, Dimitri Aerospace Engineering Schrage, Daniel Borowitz, Mariel Domercant, J. Charles Steffens, Michael ; Mavris, Dimitri

University:Georgia Institute of Technology

Department:Aerospace Engineering

关键词: Wargaming;    Stochastic;    Agent-based model;    Agent based model;    Artificial intelligence;    Decision tree;    Mission simulation;    Mission space exploration;    Surrogate model;    Lanchester equations;    Lanchester law;    Engagement analysis;    Engagement simulation;   

Others  :  https://smartech.gatech.edu/bitstream/1853/60220/1/GORDON-DISSERTATION-2018.pdf

美国|英语

来源: SMARTech Repository

【 摘 要 】

A process for using stochastic and probabilistic agent decision-making was proposed. This process is employed to conduct a Mission Space Exploration to evaluate a wide assortment of possible matériel and non-matériel solutions to an identified capability gap. The Stochastic Agent Approach (SAA) simplifies the previously intractable Mission Design Space using Agent-Based Variables (ABVs). In this method, rather than a detailed discretization of the actions (or decision-making logic) undertaken by every simulated asset at every time-step, the variable space becomes a series of decision-affecting coefficients. The method was developed through a series of experiments to identify the stages necessary to achieve this capability. Case studies involving an aircraft strike mission and a contested reconnaissance scenario were used to explore the approach. However, this methodology is applicable to a variety of scenarios and situations. The procedure for generating and leveraging Agent-Based Variables is simulation method- and tool-agnostic, and adaptable to most potential scenarios. In summary, the SAA illustrates how to craft Agent-Based Variables to achieve a degree of Mission Space Exploration and capability approximation not currently feasible. This enables a superior ability to understand, adapt to, and conduct trades within the Mission Design Space through this inclusion of decision-making affecting parameters.