Monte-Carlo Tree Search Algorithm in Pac-Man Identification of commonalities in 2D video games for realisation in AI (Artificial Intelligence)

The research is dedicated to the game strategy, which uses the Monte-Carlo Tree Search algorithm for the Pac-Man agent. Two main strategies were heavily researched for Pac-Man’s behaviour (Next Level priority) and HS (Highest Score priority). The Pacman game best known as STPacman is a 2D maze game that will allow users to play the game using artificial intelligence and smart features such as, Panic buttons (where players can activate on or off when they want and when they do activate it Pacman will be controlled via Artificial intelligence). A Variety of experiments were provided to compare the results to determine the efficiency of every strategy. A lot of intensive research was also put into place to find a variety of 2D games (Chess, Checkers, Go, etc.) which have similar functionalities to the game of Pac-Man. The main idea behind the research was to see how effective 2D games will be if they were to be implemented in the program (Classes/Methods) and how well would the artificial intelligence used in the development of STPacman behave/perform in a variety of different 2D games. A lot of time was also dedicated to researching an ‘AI’ engine that will be able to develop any 2D game based on the users submitted requirements with the use of a spreadsheet functionality (chapter 3, topic 3.3.1 shows an example of the spreadsheet feature) which will contain near enough everything to do with 2D games such as the parameters (The API/Classes/Methods/Text descriptions and more). The spreadsheet feature will act as a tool that will scan/examine all of the users submitted requirements and will give a rough estimation(time) on how long it will take for the chosen 2D game to be developed. It will have a lot of smart functionality and if the game is not unique like chess/checkers it will automatically recognize it and alert the user of it.