Decision Engine using Neural Network in group intelligence games

With the rapid improvement of CPU and the widely application of GPU in games, hardware is no longer the bottleneck of games. Thus the focus of game development is changing from graphics to AI gradually. How to endow an entity with credible intelligence, making Non Player Character (NPC) reflect a more realistic human behavior, studying independently, and making real-time decision according to the changing environment, which has become the hotspot and difficult point of game development. In this paper, we implemented a Decision Engine (DE) using Neural Network, and developed an intelligence Robot Soccer game based on this DE. The NPC has showed a triangular meet, help defense, the role of placeholder, infiltration and other humanlike advanced intelligence. Meanwhile, this DE can be widely used in risk assessment, information classification, and tactics prognosis of games with good effect. Copyright © 2010 Binary Information Press May, 2010.