Automatic Level Difficulty Adjustment in Platform Games Using Genetic Algorithm Based Methodology

DOI: 10.5176/978-981-08-3190-5_482

Authors: Nirach Watcharasatharpornpong, Vishnu Kotrajaras

Abstract:

In platform games, enemy behavior is not complicated. Therefore, challenges in such games come from the right mixture between enemies and environments of each level. Platform games require manual testing for tuning the game balance for mass audience. This is very time consuming. In addition, the difficulty of each level obtained is not guaranteed to suit individuals. Very few researches tackle how balanced levels can be generated automatically for individuals. This paper proposes a new methodology for using artificial intelligence to adjust games difficulty to suit players by automatically generating levels in platform games. The method is inspired by genetic algorithm. It is much easier to implement compared to an existing reinforcement learning based method, while still maintains similar gameplay quality. The new methodology also consumes less memory.