<![CDATA[In a previous study, the Monte Carlo Tree Search (MCTS) algorithm proved to have been successfullyapplied to turn-based GO games. In the application, MCTS has produced the highest score comparedwith previous scores. The application of MCTS is also performed on Ms. Pacman realtime games,where the resulting score is satisfactory compared to the previous highest score. Seeing the success ofthe application of MCTS, in this research applied MCTS to the enemy agent in the game MazeTreasure. Testing is done to find out how to validate the behavior and performance of agents in thegame. For behavior validation is done by looking at the level of completeness. The completeness levelis tested by comparing the scores obtained by the agent and the scores available. The result of thebehavior validation test shows that the completeness level of 25 simulation maps is 100%, where thecompleteness value of each map is true. For performance testing is done by comparing the frames persecond (FPS) in each simulation map. The results show that the best average performance is on the16x12 grid with value 261.78 FPS. An increament in the size of the labyrinth will cause a decrease inperformance. The use of MCTS on the size of the above 52x39 grid map will cause the game not wellto be played, which is the minimum FPS for games that are worth for playing is 30 FPS.]]>
Copyrights © 2017
