"Improving The Artificial Intelligence Component Of Warcraft II With Genetic Algorithms And Dynamic Scripting With Conditional Weights"
Zachary Patterson 2008
Abstract
Everyday a new game comes out with better graphics then the ones before it. They have huge interactive worlds and deep and involving plot lines. But they use the same artificial intelligence techniques that the games a decade before used. It is the goal of this project to apply the techniques of academic AI to the realm of game AI. This project contains two parts. The first part trains a genetic algorithm to optimize the Grunt Rush strategy. The second part attempts to improve the dynamic scripting algorithm. It keeps track of rules that were successful together and encourages them to be paired together again. Some of the genetic algorithms solutions were able to beat the default script consistently. Dynamic scripting was not able to consistently beat the default script even with the conditional weights improvement.