Intelligent Strategies for Super Mario

Abstract

In this project, an intelligent controller is implemented for the computer video game Super Mario. The main purpose of this controller is to help Mario avoid enemies, obstacles and gaps. A* search algorithm is used in the design of the controller as a path searching method together with competitive learning to choose the best move. Compared with rule-based methods, A* algorithm with competitive learning strategy is more sophisticated and more difficult to implement. With this method, Mario can find a better path to win a game in various environments and difficulty levels.