Using Artificial Intelligence to Enable Dynamic Gameplay
By: Richard Porter
Supervisor: Michael McCoy
Master of Interactive Technology degree conferred May 16, 2009
Thesis / Project completed June 13, 2009

This thesis investigated the viability of using artificial intelligence to enable dynamic gameplay. Dynamic gameplay promotes exciting, challenging, non-repetitive player experiences. Artificial intelligence represents a powerful but complex tool for delivering these types of experiences. This project implemented an example of how to use this tool and evaluated its effectiveness.

Field review on the subject of artificial intelligence revealed several methods for dynamic decision-making. The project includes a custom UnrealScript AI Controller class that incorporates several of these methods.

The project features an original level, developed in Unreal 3, which uses the custom AI scripting to implement game encounters. These encounters provide opportunities for computer enemies to make decisions that affect gameplay. Additional scripting included new actions for the player, new enemy types and abilities and a custom weighted tree-based learning model for the enemy AI.

Twenty testers played the final product. The players experienced multiple nonscripted encounters with the same AI. The testers answered a survey of question regarding demographics, their knowledge of AI and their opinion of the opponents they faced in the level. Analysis of the results shows that the majority of the testers considered the opponent’s gameplay interactions to be a dynamic and enjoyable experience.

Download entire thesis (.pdf)