Customer Reviews

Artificial Intelligence for Games (The Morgan Kaufmann Series in Interactive 3D Technology)

5 star:		(4)
4 star:		(0)
3 star:		(1)
2 star:		(1)
1 star:		(0)

 

 

I have been fascinated with AI for a long time, so I was excited to see this book. I own 3 other AI books, and all of them are really good. This book explains things in a way that is easy to understand. The author doesn't use any C++ in the book every algorithm is done in pseudo-code to make it easy to implement using any language. It is a definitive guide to the basic and not so basic AI techniques. The aicore that the author provides on the CD is well documented and is very helpful.

The book covers:
Steering behaviors
Pathfinding
Decision Making
State Machines
Fuzzy Logic
Waypoints
Learning Behaviors
Communication
Teaching characters

And a break down of how a typical AI design is done in different types of games.

Just to give you a notion I am about 12 hours into this book. So I may add or change this as I get further along, but overall this is a must have book.

There are a few things that I personally don't like. One is no .exe are on the cd so everything has to be built. This is a new book, so maybe the author will build them and place them on his website. I would also like to see some solutions for Visual Studio on the cd. The author says several times he tries and makes the code as platform independent as possible, but It would be nice to have prebuilt .exe files at least so I can see the demo's in action.

Understand that the pseudo-code approach this book takes is what makes it such a standout from the rest of the crowd. The author is technically thorough and the syntax is straightforward enough to use in any language needed. Moreover, it frees the author to discuss AI in abstract terms which, in the end, proves to be much more valuable content. C++ source code puts the pseudo-code discussions into practice for those looking for real-world examples.

I would HIGHLY recommend this book as a follow up to Mat Buckland's "Programming Game AI by Example" (Nov., 2004)

This is a very solid book on AI for games.

The C++ source code provided with the book is excellent. While the examples are visually unexciting, they demonstrate the power of the book's principles without the clutter that a complete graphics game would require. I was able to compile and build all the examples on the CD in one evening. The code demonstrates many of the best practices of C++ programming and design patterns.

The author builds up a nice AI engine as you progress through the book. The C++ code from the CD (or web-site) is well commented and ties exactly into the pseudocode in the book.

Millington goes into considerable detail as he reveals the power of Artificial Intelligence for Games. He carefully explains each step including the math and physics required to carry out the execution. It is obvious that he has a great deal of experience in writing computer games. He shows you a clear solid way of doing things and then discussed the strengths and weaknesses by comparing it to other techniques and addressing possible optimizations.

To read and understand this book takes time and hard work. Artificial Intelligence is a large and complex topic in math and computer science programs. The author has brought many nuggets of wisdom from that branch of research and made them understandable and useful for game programmers. Not an easy job, but Millington is one of the best at explaining difficult concepts in a clear and straight forward way.

The other reviewer's that are knocking this book because of the code, don't know what they are talking about. The code is excellent and what makes this a 5-star pick.

This book is really good and is different from other ones in the field of Artificial Intelligence. Millington explains difficult stuff in an easy and readable way. I like the academic approach of the book, I used it during my last year in college and it turned out really useful. If you want implementation details you have the source code in c++. The use of pseudocode is the best idea when writing these sort of books.

I bought this book for an Artificial Intelligence for Games class at my University. I haven't read through all of the book, but I can already tell you that the pseudo-code in this book is very poor. It's obvious that the author didn't actually go through and run the code to make sure it worked. In the movement algorithms, the code would sometimes alter rotation (speed of orientation) when it should be altering the orientation itself. In the dynamic kinematics class, the code multiplies the velocity by the acceleration instead of just simply adding the acceleration to the velocity.

Even when you get the provided movement algorithms to work the way the author probably intended, there are still issues that aren't considered. I won't get into too much detail but an example is the "Arrive" behavior. It doesn't work properly because the bot never arrives at it's target. There is nothing in the algorithm that actually decelerates the bot or nullifies the velocity. So you basically get a bot that wiggles back and forth on stationary targets.

The explanations are pretty straight foward, and I admit I haven't read the full book yet. I just think it's pretty unacceptable to publish something with so many errors in the pseudo-code.

The author uses "pseudo-code" through out the book. The cd contains only a pc-executable program. There is no source code on the CD.

This book is a poor source of programming code where the author explains how ai works based on the pseudo-code.

If you're looking for source code (ie C++ source code) you'll not find it here.