Creating efficient walls using potential fields in real-time strategy games

Abstract

Real-time strategy games (RTS) has been recently the focus of many works in the area of artificial intelligence due to their similarity with real world problems. In this context, only a few papers have addressed the problem to construct walls of buildings that enable to prevent the advancement of the opponents on the map of these games. In this paper, we propose a new model able to efficiently construct walls of buildings using StarCraft Brood War game as a case study. We evaluated our model by comparing it with one of the few solutions proposed by other researchers, as well as the pattern proposed by professional competitors. The results obtained by our model in a competition map are close to those adopted by professionals.