Methodology for human decision making using fuzzy ARTMAP neural networks

One way to improve how simulations more realistically model human behavior is through use of intelligent computer generated objects (ICGOs). This paper implements a methodology using fuzzy ARTMAP neural networks. The various modeled scenarios involve an ICGO representing military combat vehicles that select an appropriate weapon system and munition based on a set of parameters queried from the simulation state space. Additionally, a more complex scenario is modeled that involves a platoon leaders decision to attack, defend or withdraw. The result is an ICGO that obeys previously existing computer generated object rules, but also exhibits learned behavior by making decisions based on the outcomes of past decisions made under similar conditions. We also discuss an application of the methodology in a quite different direction, to develop decision rules to detect and identify Airborne targets (helicopters) and ground based targets such as tanks and personnel carriers using power spectral density or 2D-wavelet plots generated from the acoustic sensor data in a battlefield environment.