Review of "Synthetic Creatures with Learning and Reactive Behaviors by Alex J. Champandard", New Riders Publishing, 2004, 768 pp

Georgios N. Yannakakis
2005 Computers in Entertainment  
____________________________________________________________ This book is the first attempt to bridge the current gap between artificial intelligence (AI) research in academia and computer-game development in industry. The book bridges the gap successfully. The author uses FEAR (Flexible Embodied Animat 'Rchitecture), which is an open-source project integrated with a commercial first-person shooter (FPS) game, to analyze, implement, test, and evaluate various AI approaches to various behavioral
more » ... tasks for nonplayer characters (NPCs). The tasks start with low-level obstacle-avoidance and progress to high-level character emotions; each task with its own part in the book.] Each part is structured into chapters and follows a standard schema: the problem is analyzed, potential mechanisms (including AI) are discussed informally, details of the selected approach are outlined formally, an implementation is provided, the approach is evaluated (primarily visually) through simulation, and finally the advantages and disadvantages of the proposed approach are discussed. The book's content and structure make it primarily suitable for game development, but it can also be used for didactic purposes, owing more to its practicality than to the theoretical background it covers. Most of the chapters end with demonstrations of the proposed prototypes, which are available from the book's web site. Source code implementation and demonstrations may very well provide suggestions for lab assignments. The quantity of mechanisms, the detailed description of open-source material, and the steps through the game development process are all helpful for game developers at all levels. However, researchers focusing on applications of AI techniques in computer games may find this book merely an extended overview of the area, since the quantity of AI methodologies described and implemented are at the expense of theoretical depth and extended literature review expected by researchers. In addition, this book is not suitable as a complete course text because there are too few chapters where the theoretical aspect of the AI approaches is detailed to the level required by lecturers. Overall, the book's strong points are its careful consideration of problems that may arise in any phase of AI game development and the suggested solutions, which are outlined. Moreover, the game environment, the problem, and the case study assumptions in each part are analyzed in detail. On the other hand, weaknesses can be found in the endeavor to generalize AI methods on the basis of the platform of a single genre of computer games (i.e., first-person shooter). Case studies in this book cover a large set of
doi:10.1145/1057270.1057289 fatcat:txmfzeic6jg4zgymmsx6vr6m34