Pathfinding in Dynamically Changing Stealth Games with Distractions

Stealth games are genre of games that require the player to sneak past guards to reach a goal location. Designing a level for a stealth game is difficult, however, as the stealth paths in a level depend on a complex interplay of the physical space, the movements of enemies, and other factors. One such factor that makes the creation of good levels much more complicated, but also makes the levels much more interesting is distractions. They are player actions that change the movements of the enemies. By having dynamically changing guard movements, the realm of possible solutions increases greatly. The search space is thus also much larger, and hence finding good solutions becomes that much harder. Here, we create our own version of a probabilistic search algorithm to analyze a level and find stealth paths in the presence of distractions. This is done in a way that naturally allows variation in level design and enemy movements, while also allowing for a large variation in types and numbers of distractions that are present in the level for the player to use. We then introduce a series of optimizations that dramatically increase the success rate of the search while greatly decreasing its runtime, and create a compositional form of the search to solve larger, more complicated levels that are similar to real game levels. This design is integrated into the Unity 3D game development framework, which allows the creation and exploration of different levels, and exploration of how all of these factors including the placement and types of distractions affect the potential for stealth movement by players. We also model some levels from existing games to show the applicability of this design.