Enabling dynamic generation of levels for RTS serious games

Abstract This paper proposes a new family of Real Time Strategy (RTS) serious games that exploit a dynamic insertion of aerial/satellite images in games, in order to allow covering any target geographic area on user demand and automatically implementing several different gaming sessions. The approach intends to combine the possibility for the user of practicing decision strategies with the development of knowledge about specific geographical areas, which is important in particular to enhance the training of field operation personnel. This target requires that the system correctly interprets the terrain features – in particular roads – in order to build a semantic correspondence between the game’s logic and the background image. RealPath, a new solution for automatic definition of paths on aerial and satellite images, which reduces the cost for the creation of new game maps. The algorithm exploits the image’s pixel values and extracts information on the terrain in order to identify possible paths in the area. A parallel version of the algorithm has been implemented, exploiting multi-resolution pyramidal image processing, in order to allow better exploiting the power of current multicore processing architectures (CPUs and GPUs). The algorithm has been successfully tested on several real-world images, in real-time. As a proof of concept, we present a RealPath-based RTS serious game that we are developing for training military and civil protection personnel in field operations. Exploitation of the algorithm in 3D virtual environments is also shown.