Dynamic game difficulty balancing with user profiling

Abstract

Computer games have increasingly grown in terms of popularity due to the faster machines and new hardware being released. Two main factors which make up an enjoyable game are amazing graphics and challenging game-play employed within. Throughout the years, the graphics have been greatly enhanced, providing a well designed and realistic environment for the player by working hand-in-hand with the hardware. AI mechanisms are also being applied with the aim to improve the popularity of such games. The aim of this study is to explore several AI techniques to be able to filter the most suitable algorithms amongst the considered techniques, which can then be applied to create an adaptable environment for a specific game. Besides the implementation of these procedures, a user profiling system is also created to aid in the final response given by these techniques. An Artificial Neural Network together with a Genetic Algorithm were used to generate creep waves on the fly by considering the strategies currently adopted by the user. The purpose of the Artificial Neural Network is to provide an efficient result within a few milliseconds. The testing carried out in this project provided results which indicate that both these algorithms managed to reach the aim set in this project. It can therefore be concluded that the interaction of two or more algorithms can greatly enhance the difficulty settings by also considering the real time performance of the player, together with any other important variables enclosed in the respective profile.