Human-like Behaviour in Real-Time Strategy Games : An Experiment With Genetic Algorithms

Abstract: If a computer game company wants to stay competitive they must offer something extra. For many years, this extra has often been synonymous with better graphics. Lately, and thanks to the Internet, the focus has shifted in favour of more multi-player support. This also means that the requirements of one-player games increases. Our proposal, to meet these new requirements, is that future game AI is made more human-like. One way to achieve this is believed to be the use of learning AI techniques, such as genetic algorithms and neural networks. In this thesis we will present the results from an experiment aiming at testing strategy game AI. Test persons played against traditional strategy game AI, a genetic algorithm AI, and other humans to see if they experienced any differences in the behaviour of the opponents.