Implementing game artificial intelligence to decision making of agents in emergency egress

Evacuation is the movement of people away from threat in order to minimize loss. In deriving a proper evacuation plan, traditional approach or computer simulation may be used. There are wide selections of evacuation model currently made available, which can be categorized into three major approaches. Of all three modelling approach, Multi Agent System (MAS) model is considered as the best approach, in which it provides a realistic insight into evacuation process by including the human factor. Still, it leaves much room for improvement. This paper proposes an algorithm composed of two popular techniques from the field of Game Artificial Intelligence (Game AI), namely Behaviour Tree and Utility-Based AI, which will be applied in the decision making process of an agent. This algorithm is develop to simulate a realistic decision making process that can reflect a human thinking process, in order to overcome the issue of intention persistence of an agent in MAS model.