Neural networks training for weapon selection in first-person shooter games

Abstract

First person shooters is probably the most well known genre of the whole gaming industry. Bots in those games must think and act fast in order to be competitive and fun to play with. Key part of the action in a first person shooter is the choice of the right weapon according to the situation. In this paper, a weapon selection technique is introduced in order to produce competent agents in the first person shooter game Unreal Tournament 2004 utilizing the Pogamut 2 GameBots library. The use of feedforward neural networks is proposed, trained with back-propagation for weapon selection, showing that that there is a significant increase at the performance of a bot. © 2010 Springer-Verlag Berlin Heidelberg.