Determining NPC Behavior in Maze Chase Game using Naïve Bayes Algorithm

Abstract

In this paper, we propose an implementation of Naïve Bayes algorithm in a chase game called Maze Chase. Maze Chase is a chase game where a player must avoid several chasings Non-Player Character (NPC). In our proposed implementation, the NPC will run automatically using artificial intelligence. There are four NPC in Maze Chase, each with its own characteristics. Because of the characteristic differences, the four NPC needs to communicate with each other. For communication we use a multi-agent system. Multi-agent system is a part of artificial intelligence which were used by NPC to communicate with each other using several defined parameters. We used several parameters, such as the number of coins in a zone, the amount of golden coins in a zone, and the centroid values. These parameters were used as variables for an implementation of Naïve Bayes algorithms. Our proposed implementation of Naïve Bayes was used to count the probabilities of NPC behavior, which will move closer towards the player according to several zones in the game map. From the testing results, Naïve Bayes algorithm could be used to decide the NPC movement according to its target zone on the Maze Chase game, with error rate 0.5%.