Authors: Pichit Promsutipong, Pakapon Kaewmark and Vishnu Kotrajaras
Enemy is one of the challenges that games have for a player. The player must figure out its characteristics to overcome the enemy and progresses through the game. Providing the player an enemy with emergent behavior can spice up the player's experience. To create emergent behavior, we need an algorithm that is able to explore and utilize the enemy characteristics space. Our ultimate goal is to build such algorithm. As a starting point, we have to shape the characteristics space. In this paper, we propose a game agent model for 2D avatar-based action games, along with a description language used to bridge the model to games. The model is created using data we analyze from five avatar-based action games and is validated by recreating enemies from other games. Our preliminary result shows an example of enemies described using our model, including enemies from a game with different mechanics from our data.
Keywords: enemy model; action game; description language