Applying Behavior characteristics to decision-making process to create believable game AI

Abstract

With the development of artificial intelligence in computer games the problem of creating characters with believable and diverse behavior to inhabit in-game worlds becomes more and more actual. A big number of required characters and high standards of a modern game artificial intelligence makes the problem even more complex. In this paper we propose a utility-based decision-making model which gives the possibility to generate characters with believable behavior. The believability of such characters comes from their decision-making process that takes into account not only assessment of game environment, but also their personal characteristics and social status. Designed model was used to generate AI driven characters for a development of player’s opponents with personality traits for a computer card strategy. The model was used to simulate human flows on main railroad hub of 2014 Winter Olympics in order to reveal areas where pedestrian flow should be controlled.