Development of Non-Player Character for 3D Kart Racing Game Using Decision Tree

Abstract

Racing game is one of the genre that's still popular today. Unity is one of many game engines one can use to develop a racing game. At Unity Asset Store, there is a free template called Micro-Game Karting which can only be played alone. In order to play player versus enemy mode, an artificial intelligence (AI) is needed for directing non-player character (NPC) who acts as the opponent. In racing game, the AI requires the use of movement algorithm and decision making system. For this study, the movement algorithm will use pathfinding. The algorithm is used as a guiding path when NPC is moving and avoiding obstacles in the way. Pathfinding will use waypoint system and raycasting to accomplish it. The decision making technique that will be used is decision tree. It functions as decision maker for NPC so it can determine the correct action to be done at certain time. As for FPS (frame per second) test, performance suffers 0.2-0.3 FPS decrease for every addition of 2 NPCs. According to lap time test, the developed NPCs are faster than the machine learning NPC examples provided by the template and driving test showed favorable outcome.