Customer Reviews

AI for Computer Games and Animation: A Cognitive Modeling Approach

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

 

 

When I bought this book I wasn't quite sure what to expect since the title, "AI for Games and Animation: ..." is so broad sounding. What I did expect was that it would be written in an easy to read form and hopefully provide enough in-depth coverage of the topics so that a reasonable game programmer could implement them based on the information provided, without further research. This criteria seems to be adhered to in the technical articles in a publication such as Game Developer.

Apparently, no-one bothered to tell Mr. Funge about this. Although the book does provide some case studies (examples) the actual details of their implementations are often glossed over or shrouded in unnecessary mathematical formalism that is out of place in a game programmer's book. To give an example, complex topics such as inverse kinematics and coupled spring systems are given several short paragraphs. In describing coupled spring systems for deforming a mesh, Funge uses "x dot" notation that most game programmers probably are not familiar with.

The only saving grace is the reference section which can point the reader to more specific literature that may actually be helpful in constructing implementations of some of the techniques described in the book. This increases the rating from the worst (1 star) to 2 stars since it is actually pretty comphrensive.

The title of this book is its biggest flaw because it appears to cater to game programmers but the techniques described within it are not that useful. It appears that the book is marketed to the game programming community due to its immense size. Anybody interested it this book would be well advised to purchase the siggraph proceedings that it comes from because not only will you learn what this book is about, you will also have a large resource of other techniques that can be used for graphics and animation.

Yes, John's technique is interesting and different ... however he does not have a gaming background, has never worked on games, and it is obvious that this book is simply an extension of a research paper. You can find better resources for both AI and game programming elsewhere.

This book is a fascinating mixture of perfectly reasonable AI techniques which are seldom useful for games, combined with useful but obvious techniques which are described better in a standard AI textbook.

Overall, this book suffers greatly from a lack of relevance. There is no pseudocode, and the few examples in the book are poorly chosen and generally not representative of the real problems encountered in game AI programming. Similarly, many of the valuable techniques often used in game AI are not even mentioned.

The book gives a brief overview of breadth-first, best-first, and depth-first search, but it entirely omits mention of the A* algorithm, some form of which is used in nearly every game that involves pathfinding.

The book also covers a number of topics that are entirely out of place in a book about game AI. Chapter 2 is entirely devoted to game physics. Chapter 6 covers machine learning, a technique which is seldom used for games. And Chapter 11 covers texture mapping and deformable models, which, although important for games, have nothing to do with AI.

The book does have some high points, such as the and the introduction to interval arithmetic. But even then, it wastes the reader's time with several pages of unnecessary mathematical proofs.

Overall, I recommend avoiding this book. Not only will it not inform the reader, it presents a misguided notion of what game AI is all about.

The field of "artificial intelligence" is so broad, that even confining it to the domain of computer games doesn't help matters much. This book is not a cookbook on "how to write my monster code", but instead covers more of the general theory of behaviors and intelligence.

I don't have a background in AI coding, so I'm not familiar with the standard AI texts another reviewer points to. Where this book excelled for me was its strong overview of AI and its applications. If you're looking for a book to hand you code for your next big game, I would look elsewhere. For a solid understanding of many of the issues of AI, this book worked quite well for me.

I only gave it three stars because of its terseness and the lack of direct relevance (read: immediately usable code) for many people. For some, like myself, the book is valuable just for briefly surveying the field, allowing me to explore specific areas in more depth.

In this title, the author strives to take Artificial Intelligence to the next level. The cognitive approach is one where a character is given planning and reasoning skills with a behavioral outline and the character uses his own logic to determine its next actions. The book is designed for intermediate to advanced AI programmers, so the format of the book is quite technical (read as 'mathematics~ based'). The book guides you through such topics as, "Reactive Behavior Rules", "Hierarchical Finite-state Machines", "Precondition Axioms", "Sensing", "Embedding Goal-directed Behavior", Priority-based Control Systems", "Subgoals" and many other related issues.

The book is roughly about 200 pages, and has 3 different examples for you to follow: a maze, a dinosaur world, and an undersea world. The maze is simple enough, its when you see the use of the cognitive approach in the dinosaurs and undersea creatures, that you realize the true potential of this method. In particular, I found the undersea world quite interesting. It consists of predatory sharks and merpeople. A shark will forage around systematically, going back to any spots it last saw a merperson, and will chase and try to eat any merperson it sees. Sounds simple enough. The merpeople have no weapons, but use their superior reasoning capacities to elude the 'jaws of death'. This is when things get cool. Using techniques described in the book, the author demonstrates how a merperson who has been given 'brave' characteristics will try and help a fellow merperson crying for help, as it tries to evade becoming the shark's next meal. The 'brave' merperson even risks his own cyber-life by hiding between rocks and getting dangerously close to the shark, trying to get its attention. A close call or two and the 'brave' merperson is finally able to distract the shark long enough for the other merperson to escape. The 'brave' one then leads the other back to the safety of the rocks and caves. Overall, pretty interesting reactions.

When you start thinking like the author does, the whole realm of game character AI takes on a whole new meaning. Speaking of meanings, according to the Merriam Webster's Dictionary, cognitive is described as = 1 : relating or involving the act or process of knowing; including both awareness and judgment. ( from Latin -cognitio, from cognoscere, to become acquainted with, know, to come to know ) 2 : based on or capable of being reduced to empirical factual knowledge. This truly is the theory being presented in this book. It is definitely not a beginner's guide to AI, nor by any means a final authority on the subject. AI for Games and Animation is a book that helps you, the animator/ game designer, progress the way you think about your characters and how they will function for you in your next project. If the subject of AI programming pertains to you, you would be well off adding this new release to your collection.

John's book is definitely not for the faint of heart. I am a professional AI designer, I have a degree in AI, I understand the basic concepts at work and I found it very involving at times. A strong math background would be useful.

One reviewer makes the point that the book does not cover any "game AI". You should be aware that "game AI" is a marketing term that, unfortunately, the industry itself has bought into wholesale. It is not relevant to this book which covers AI. In other words "game AI" and AI are not the same things. Some developers may be surprised at this statement but I would say look at my credentials first. AI is the future and game developers would do well to get a heads up on some useful aspects of AI that they can apply to their products, cognitive modeling for characters being one of them.

This is not a great book, it is unfocused and simply a review of John's SIGGRAPH presentation and work at University. It describes a great framework for creating intelligent characters and the subject deserved a better presentation. There will be better books on AI for Interactive Entertainment but this is probably the best so far.

Real AI will be the difference between products that sell and those that end up in the retailers bargain bin, the time to start learning is now.

The idea of creating an autonomous virtual creature with behaviors that dynamically change in response to alterations in the virtual environment is intriguing. With a set of many different creatures available, one could add as many of each type as necessary and have a game or simulation with an enormous number of possible outcomes. This idea is consistent with the object-oriented approach to programming, where complex blocks of code are called components. In the ideal scenario, to build a program from components, you would select the ones that you need and then specify how many of each type and how they are to interact.
Funge puts forward the principles of a high-level language that can be used to abstractly define the behaviors of virtual characters. Unfortunately, he does not describe enough of the language to be completely convincing that his approach has been effectively implemented. Even insects demonstrate behavior that appears complex, and his examples are those of a Tyrannosaurus Rex trying to herd a small set of Velociraptors and a set of merpeople trying to avoid a predatory shark. Both examples include fixed obstacles and to be interesting, the creatures must exhibit behaviors that surprise human observers as well as their virtual opponents.
The treatment of artificial intelligence (AI) is also too light to completely prepare the way for the case studies. For example, the title of chapter 6 is "Learning", where the topic is ways in which a machine (virtual creature) can learn. However, it is only fourteen pages long, which is hardly enough paper and ink to even scratch the surface of this complex topic.
This is not to say that there is a lack of good ideas in the book, in fact there are many. My criticism is that they are not expanded out to the point where someone who reads only this book will be able to act on them. Intelligence is a complex topic, and making it artificially has proven to be very hard. Therefore, none of the ideas put forward in this book are adequately covered.

When you are reading this book you fell like a man telling you things that you already know, nothing new. If you dont know nothing about AI maybe will help you.There is not algorithms on it, only obvious info. If you want to learn real AI buy "Artificial Inteligence" by Rich

This book covers the fundamentals of AI techniques used in building intelligent characters. Not a step by step cook book, but gives a solid overview of various AI methods for behavior modeling and proposes a novel and totally compelling approach (cognitive modeling) to creating a new breed of characters - characters that are able to think for themselves! Definitly a good addition to your collection!

Computer Generated Imagery (CGI) has transformed the art of animation. Computer-generated characters are now an everyday part of contemporary culture from advertising to video games, from movies to the Internet. Artificial intelligences gives computer-generated characters the ability to react to their environment in logical ways. In most cases, a character is programmed to present a specific set of behaviors that allow it to interact with other characters or game players. But this approach also results in severe limitations on responses to a characters world and its behavior becomes too easily predictable.

In AI For Games And Animation: A Cognitive Modeling Approach, John Funge introduces a new approach for creating autonomous characters. Cognitive modeling provides computer-animated characters with logic, reasoning, and planning skills. Using this technique, an animator or programmer need only specify a behavior outline and, through reasoning, the character will automatically work out a detailed sequence of actions. AI For Games And Animation is an invaluable contribution to personal and professional CGI animation reference libraries.