Controlling CGF-Generated Entities Using a Fuzzy Logic-based System
(Room S320C)
29 Nov 17
11:30 AM
-
12:00 PM
Computer Generated Forces (CGF) is a key component in virtual and constructive simulations and offers a costeffective way to enhance realism by providing methods to control simulated entities. CGF is becoming an essential tool for tactical training, especially for mission rehearsal. CGF facilitates mission training by providing the means to design training scenarios. A training scenario consists of a set of predefined events that occurs during training, which involves setting a number of parameters of the computer-controlled simulation models. On one hand, CGF scenarios tend to be static. Once trainees have completed a scenario, they will likely know how it will behave during the next training session, thereby reducing or removing the reuse value of the scenario. On the other hand, a fundamental characteristic of CGF is decision-making based on artificial intelligence (AI). Current AI decision-making implementations are commonly simplistic, using a fixed set of “Rules of Engagement.” The nature of this behavior makes it easy for trainees to distinguish between computer-controlled and human-controlled entities in a simulated environment. These specific CGF characteristics can result in ineffective or negative training because trainees are able to quickly familiarize with the behaviors of the simulated entities and then easily defeat them, which would not occur with human-controlled opponents.
In this paper, we propose a method to control the behavior of constructive entities generated from a synthetic environment. This novel method makes essential use of a fuzzy logic-based system. We illustrate the proposed method with a simulation of an Air Defense Missile System (ADMS). The ADMS simulation computes the missile launch envelope and uses the simulation results to determine the “in-range” condition of hostile air targets. The result of the ADMS simulation demonstrated that the fuzzy logic-based system is suitable to emulate the decision-making process of a human ADMS