<![CDATA[Ms. Pac-Man is a game used by the researchers to make an artificial intelligence for playing the game automatically. Since Ms. Pac-Man is a non-deterministic game, there are many things in the game that are random, for example the ghosts movement, making Ms. Pac-Man quite hard. Researchers think that as a interesting challenge for developing computer science especially artificial intelligence. Even more, researchers made a competition for the best artificial intelligence in this game, using score as a ranking for the agent performance. One of the competitions is IEEE CIG 2011. For this problem, the author are using Hierarchical Path-Finding A* (HPA*) method, this method is an alternative from A* to make an agent controller in Ms. Pac-Man. The reason for choosing this method is because to test the HPA* method, whether it can achieve better result or worse. Agents using A* methods haven't reached the highest score in this game. In previous research, the maximum score achieved using A* method is 24640 in ten times testing. Compared to the newest research using decision tree method can reach a maximum score of 43.720. Roughly only half score that A* get in comparison to the newest research. To improve A* result, this research uses HPA* method. By dividing the game area into several areas using HPA*, the score will be expectedly better from the previous research. But in practice the results are not good, the final score is 17050. For this reason, this research implements the HPA* and analyze the performance of HPA* method.]]>
Copyrights © 2018
