How's my driving? Modeling player navigation in the Games United League of BZFlag

John Paolillo


3D game environments require players to adapt to the environment and user interfaces, imposing psycho-motor, cognitive and strategic challenges, which players may adapt to in different ways. Navigation in a 3D game is therefore complex, so that any specific instance of player movement may involve features of the virtual space, game-world physics, user-interface controls, the player's goals and intentions, and responses to other players' actions. Moreover, different players or different types of players may have different approaches to navigation. To fully understand the skills players acquire for successful game play, we need a statistical model for navigation. This paper presents one approach to modeling navigation in which players' actions govern transitions in a network of driving states. The structure of this model is explained, and exemplified from data drawn from 2421 half-hour long official matches from the Games United League of BZFlag, an online multiplayer tank battle game. Extensions of the model to obstacle or threat avoidance, strategic goal attainment and individual navigation styles are also discussed.