Advertisement

AI book?

Started by December 19, 2005 02:49 AM
9 comments, last by d000hg 18 years, 11 months ago
I'm going to get a book related to making games and I'm considering an AI book. I've looked at GD's reviews but I'm not sure if a game-AI book is best. I don't want to spend ages on search algorithms because I did that at uni; state machines and neural networks I am more interested in. So I'm not sure if I should just get a univeristy textbook on AI, but Brian Schwab's book sounds quite good for my purposes - racing and RTS games currently. I'm shying away from Mat Buckland's book because apparently it is biased towards FPS games. I'm quite happy for a book to be quite deep since I did some AI at uni.
Title: AI Game Engine Programming
Contributor: Schwab, Brian
Publisher: Charles River Media
Date: 2004
Dewey: 794.8/1526
LC Call Number: QA76.76.C672.S39 2004eb
ISBN: 1-58450-344-0
Subjects: Computer games -- Programming.,
Artificial intelligence.

This book is great for learning state machines (FSM, FuSM) and neural networks (haven't finished reading about neural networks yet though).
Advertisement
I've recently read AI Game Development by Champandard. I want to disrecommend this book to you; my university background left me virtually only reading stuff I already knew. Moreover it is slightly biased to FPS but with university level education you can see through. Perhaps the book is okay for people who have no clue whatsoever.

Illco
Well I've actually decided now after a bit of research (it's as a prize so I had to hurry up!).

I went for the highly reviewed Game Programming Wisdom.

One interesting thing I learned - the reviews on Amazon.co.uk and .com are different - so when the UK site has only a few reviews the US one may have many more. I don't rate the GD reviews much since most ahve only one or 2 reviews and only a few are by people I know/respect (ie the staff or moderators).

AI Game Engine Programming came in as my 2nd choice btw.
Quote: Original post by d000hg
I don't rate the GD reviews much since most ahve only one or 2 reviews and only a few are by people I know/respect (ie the staff or moderators).


You should also (always) consider any review such as these with the target audience of the review in mind. If you don't fit that demographic (perhaps because you're more highly educated, or more experienced in a given field) then the review is probably (nearly) worthless to you.

I find that the best way to find good quality books is to 1) ask people who work in the field; and/or, 2) go to a good quality bookstore, sit down and sift through what is on the shelf.
Exactly - if I dpon't know the GD member then I have no idea what level they're at. Moderators or staff I would expect to be generally of a good knowledge at least, demanding good in-depth knowledge rather than books which start off by covering Win32/D3D basics and just give you loads of code to paste into your engine!

If only I had a good bookstore around here...
Advertisement
Most people tend to be top heavy giving away the stars.

If you expand the range of 3 stars to 5 stars then you probably get a better idea. 3 is really bad, 4.5 is pretty good, 5 is excellent. For that person though.
I've bought 3 books: AI Game Engine Programming, AI Game Development and Programming Game AI by Example. I've read them all(well, now i'm on the last one, the AI Game Development book). Each one teaches you different stuff, so there is no on single recommandation. The first I've enumarated teaches you all kind of techniques, as an introduction and a little bit of implementetion. The second is more close to the academic stuff, teaching how to create a reactive reasoning synthetic creature. The last one is great for beginners in a practical approach. It doesn't cover the same amount of materials like the others, but it is great to see how the things are really made by providing a lot of code an examples. However, it doesn't feature any revolutionary techniques (that will be used in the future), only the basic ones. I recommand you to read as many book as you can, because this field is huge and there is really no point to begin. Some academic background(my case) it is great, because you will understand how to apply different techniques (ex: neural networks) in game development. happy reading:D
I can't really give an opinion on which book to pick, what I'd like to comment, is on meeshoo's comment about "revolutionary techniques". Most games I had to work on / read about didn't use any "revolutionary techniques"... only a few games used neural networks and obtained good results... most games still use very simple state machines and it does the job. I have nothing against "revolutionary techniques", but most of the time, when you have pressure to ship the game on time, you'll prefer good old techniques that you know how and when you'll achieve the desired results, instead of diving in a new field and *try* to have results on time, if any results at all.

I like Mat's book, but as you pointed out, it might not give you all the specific details to implement a racing game or a RTS, but it's still a very interesting book. As for Alex's book, I only read a few pages and didn't feel like I was his target audience... probably because I don't have a heavy academic background.

AI Wisdom / Game Programming Gems are always good books, as long as they contain articles on what you want to implement :)

Cheers

Eric
xEricx, you are right. Most of the games use only common techniques, like state machines. I've mentioned that books because d000hg asked something about academic books, and I thought he is interested. I'm working on a game right now, and we plan to use some advanced AI techniques. We are using the Nebula 2 engine, which has implemented some stuff, like navigation and scripted AI, but we want to build some advanced logic on that, but keeping the fun factor in place. That's why I'm reading those books :P

This topic is closed to new replies.

Advertisement