CHAMELEON: online learning for believable behaviors based on humans imitation in computer games

摘要

In some video games, humans and computer programs can play together, each one controlling a virtual humanoid. Thesencomputer programs usually aim at replacing missing human players; however, they partially miss their goal, as they can beneasily spotted by players as being artificial. Our objective is to find a method to create programs whose behaviors cannot bentold apart from players when observed playing the game. We call this kind of behavior a believable behavior. To achieventhis goal, we choose models using Markov chains to generate the behaviors by imitation. Such models use probabilityndistributions to find which decision to choose depending on the perceptions of the virtual humanoid. Then, actions arenchosen depending on the perceptions and the decision. We propose a new model, called CHAMELEON, to enhance expressivenessnand the associated imitation learning algorithm. We first organize the sensors and motors by semantic refinementnand add a focus mechanism in order to improve the believability. Then, we integrate an algorithm to learn the topology ofnthe environment that tries to best represent the use of the environment by the players. Finally, we propose an algorithm tonlearn parameters of the decision model. Copyright © 2013 John Wiley & Sons, Ltd.