Welcome to Exercises in Game AI Programming! Over the next few months, this tutorial series is going to explore pratical issues involved in building AI for so-called “deathmatch bots.”
One major premise of my book is that most AI techniques are relatively narrow — focusing on solving one kind of problem very well. The process of creating an advanced system is thereby a matter of picking the most effective combination of techniques and integrating them together.
This tutorial series takes the opposite direction and embraces it to the extreme. Each NPC is developed as a monolothic architecture using only one AI technique. The purpose of this is twofold:
Push each technique to their maximum and reveal their strengths and weak points.
Take a very pragmatic approach in applying each technique to an assortment of problems.
Each tutorial does not cover any AI theory explicitly, although anyone with a technical background should be able to follow relatively easily. For a more in-depth look at the principles behind each technique, don't hesitate to check out my book. Naturally, there are plenty of online resources too, so I'll provide links whenever possible.
The source code is available with each demo so everyone is free to experiment for themselves. It is included as part of the FEAR project, an open source game AI initiative. The SDK is available from this download page.
