The use of neural networks for determining tank routes
Eldridge, Dwayne L.
Bui, Tung X.
MetadataShow full item record
The U.S. Army uses a combat simulator, Janus (A), to simulate high-tech ground battle exercises. The algorithms used to represent battlefield behavior and to generate battle scenarios must be calibrated by well-trained, combat-experienced technicians. The calibration is time-consuming and subject to human errors. A Single Exercise Analysis System (SEAS) is under development for automating and improving the battlefield scenario generation process for Janus (A). A neural network based model has been proposed to support the route determination process within SEAS. The purpose of this thesis is to (1) determine the best neural network architecture for determining tank routes and (2) develop a prototype for generating these routes.