RTS terrain analysis: An axial-based approach for improving chokepoint detection method

Abstract

In real-time strategy (RTS) games, to efficiently move autonomous agents, it is essential to understanding important characteristics of the terrain environment such as chokepoints that limit direction and speed of agents. We propose an axial-based algorithm that automatically detects chokepoints in more detail, especially for narrow areas. Results obtained are the colored-regions separated by chokepoints including the position and direction of significant information for region decomposition. Our terrain data is a map of tile-based RTS games represented as a passability map. The difficulty on finding the enclosed space and area chokepoints is overcome by applying the medial axis computation to capture the intrinsic geometric structure of the regions.