Artificial Intelligence for Computer Games

Transcription

1 Artificial Intelligence for Computer Games

2

3 Pedro Antonio González-Calero Marco Antonio Gómez-Martín Editors Artificial Intelligence for Computer Games ABC

4 Editors Pedro Antonio González-Calero Universidad Complutense de Madrid Dpto. de Ingeniería del Software e Inteligencia Artificial 28040, Madrid Spain pedro@fdi.ucm.es Marco Antonio Gómez-Martín Universidad Complutense de Madrid Dpto. de Ingeniería del Software e Inteligencia Artificial 28040, Madrid Spain marcoa@fdi.ucm.es ISBN e-isbn DOI / Springer New York Dordrecht Heidelberg London Library of Congress Control Number: c Springer Science+Business Media, LLC 2011 All rights reserved. This work may not be translated or copied in whole or in part without the written permission of the publisher (Springer Science+Business Media, LLC, 233 Spring Street, New York, NY 10013, USA), except for brief excerpts in connection with reviews or scholarly analysis. Use in connection with any form of information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed is forbidden. The use in this publication of trade names, trademarks, service marks, and similar terms, even if they are not identified as such, is not to be taken as an expression of opinion as to whether or not they are subject to proprietary rights. Printed on acid-free paper Springer is part of Springer Science+Business Media (

5 Here stoourwivesandgirlfriends...may they never meet! Groucho Marx

6

7 Preface Games has been an interesting domain for artificial intelligence (AI) research since the origins of the discipline back in the 1950s, although mainly board games such as chess, checkers, or backgammon. Nevertheless, it is between 1999 and 2000 when a number of researches on AI identify interactive video games as an interesting research domain for AI (see, for example, John E. Laird, Michael van Lent: Human-Level AI s Killer Application: Interactive Computer Games. AAAI/IAAI 2000: ). The long-term dream of AI research of building intelligent robots with human-like abilities for interacting in the real world is still far from being fulfilled due, among other causes, to the complexities of sensing and acting in the real world. Video games provide synthetic worlds where complex behavior can be shown but where perception and actuation are perfectly under control and, therefore, became a perfect platform for experimenting with software robots (i.e., agents). Techniques used for AI in commercial video games are still far from state-of-the art in academia, but with graphics in video games coming close to photo realistic quality, and multi-processor architectures getting common in console and PC game platforms, sophisticated artificial intelligence is getting into the focus of the video game industry as the next big thing for enhancing the player experience, while profiting from the number of spare CPU cycles available in modern hardware. For that reason, industry is getting interested in academic research in AI to provide rich, robust, and scalable techniques for controlling non-player characters and provide richer narrative schemes in games. This book collects some of the most relevant results from academia in the area of artificial intelligence for games. The selection of contributions has been biased toward rigorous and theoretically grounded work that is also supported with developed prototypes, which should pave the way for the integration of academic AI techniques into state-of-the-art electronic entertainment games. The chapters in the book cover different areas relevant to AI in commercial games: pathfinding, decision making, learning, authoring, and storytelling. Regarding pathfinding, the book describes recent real-time heuristic search algorithms that alleviate the scalability problem of A* techniques used in commercial games, which at the same time exhibit a visually appealing behavior. Techniques are also presented that, based on the semantic annotation of 3D virtual worlds, vii

8 viii Preface can learn pathfinding behavior by analyzing traces from actual players. Tools and techniques are described for incorporating semantic information into the game design and development process, thus improving the embedded information contained in immersive game worlds, and leading to new possibilities for NPC constructions such as meaningful in-game learning and agent portability. Regarding decision making, the book describes new techniques for authoring tools that facilitate the construction by game designers (typically non-programmers) of behavior controlling software, by reusing patterns or actual cases of past behavior, represented as behavior trees. Using domination games as a test bed, the book describes different approaches for building cooperative agents, with different requirements for knowledge engineering, from purely hand-coded to inductive approaches. Techniques for automatically or semi-automatically learning complex behavior from recorded traces of human players using different combinations of reinforcement learning and case-based reasoning are also described. Much research on artificial intelligence in games has been devoted to creating opponents that play competently against human players, while an alternative goal is to try to deliver the best possible experience within the context of the game. This novel goal is much more attainable by approaching AI reasoning for games as storytelling reasoning. Several technological approaches are presented in the context of such a perspective, including the use of planning techniques for camera placement and sequencing of plot points in a game, and constraint optimization for automatically adapting lighting qualities of a scene to the player preferences. Key results from applied research on AI within the last 10 years have been collected here to provide a reference work for both academia and industry that will help to close the gap between both worlds. Madrid October 2010 Pedro Antonio González-Calero Marco Antonio Gómez-Martín

9 Contents Real-Time Heuristic Search for Pathfinding in Video Games... 1 Vadim Bulitko, Yngvi Björnsson, Nathan R. Sturtevant, and Ramon Lawrence Embedding Information into Game Worlds to Improve Interactive Intelligence G. Michael Youngblood, Frederick W.P. Heckel, D. Hunter Hale, and Priyesh N. Dixit Empowering Designers with Libraries of Self-Validated Query-Enabled Behaviour Trees Gonzalo Flórez-Puga, David Llansó, Marco Antonio Gómez-Martín, Pedro P. Gómez-Martín, Belén Díaz-Agudo, and Pedro Antonio González-Calero Game AI for Domination Games Chad Hogg, Stephen Lee-Urban, Héctor Muñoz-Avila, Bryan Auslander, and Megan Smith Case-Based Reasoning and User-Generated Artificial Intelligence for Real-Time Strategy Games Santiago Ontañón and Ashwin Ram Game AI as Storytelling Mark Riedl, David Thue, and Vadim Bulitko Intelligent Machinima Generation for Visual Storytelling Arnav Jhala and R. Michael Young Intelligent Adaptive Lighting Magy Seif El-Nasr, Joseph Zupko, and Chinmay Rao Index ix

10

11 Contributors Bryan Auslander Department of Computer Science & Engineering, Lehigh University, Bethlehem, PA 18015, USA, Yngvi Björnsson School of Computer Science, Reykjavik University, IS-101 Menntavegur 1, Reykjavik, Iceland, Vadim Bulitko Department of Computing Science, University of Alberta, Edmonton, AB T6G 2E8, Canada, Belén Díaz-Agudo Complutense University of Madrid, Madrid, Spain, Priyesh N. Dixit The University of North Carolina at Charlotte, 9201 University City Blvd., Charlotte, NC 28223, USA, Magy Seif El-Nasr Simon Fraser University, Burnaby, BC, Canada V5A 1S6, Gonzalo Flórez-Puga Complutense University of Madrid, Madrid, Spain, Marco Antonio Gómez-Martín Universidad Complutense de Madrid, Dpto. de Ingeniería del Software e Inteligencia Artificial, 28040, Madrid, Spain, marcoa@fdi.ucm.es Pedro P. Gómez-Martín Complutense University of Madrid, Madrid, Spain, pedrop@fdi.ucm.es Pedro Antonio González-Calero Universidad Complutense de Madrid, Dpto. de Ingeniería del Software e Inteligencia Artificial, 28040, Madrid, Spain, pedro@fdi. ucm.es D. Hunter Hale The University of North Carolina at Charlotte, 9201 University City Blvd., Charlotte, NC 28223, USA, dhhale@uncc.edu Frederick W.P. Heckel The University of North Carolina at Charlotte, 9201 University City Blvd., Charlotte, NC 28223, USA, fheckel@uncc.edu Chad Hogg Department of Computer Science & Engineering, Lehigh University, Bethlehem, PA 18015, USA, cmh204@lehigh.edu xi

12 xii Contributors Arnav Jhala University of California Santa Cruz, Santa Cruz, CA 95064, USA, Ramon Lawrence Computer Science, University of British Columbia Okanagan, 3333 University Way, Kelowna, BC V1V 1V7, Canada, Stephen Lee-Urban Department of Computer Science & Engineering, Lehigh University, Bethlehem, PA 18015, USA, David Llansó Complutense University of Madrid, Madrid, Spain, ucm.es Héctor Muñoz-Avila Department of Computer Science & Engineering, Lehigh University, Bethlehem, PA 18015, USA, Santiago Ontañón Artificial Intelligence Research Institute (IIIA-CSIC), Campus UAB, Bellaterra, Spain, Ashwin Ram CCL, Cognitive Computing Lab, Georgia Institute of Technology, Atlanta, GA 30332, USA, Chinmay Rao Penn State University, University Park, PA 16801, USA, Mark Riedl Georgia Institute of Technology, Atlanta, GA 30332, USA, Megan Smith Department of Computer Science & Engineering, Lehigh University, Bethlehem, PA 18015, USA, Nathan R. Sturtevant Department of Computer Science, University of Denver, Denver, CO 80208, USA, David Thue University of Alberta, Edmonton, AB, Canada T6G 2E8, R. Michael Young North Carolina State University, Raleigh, NC 27695, USA, G. Michael Youngblood The University of North Carolina at Charlotte, 9201 University City Blvd., Charlotte, NC 28223, USA, Joseph Zupko Demiurge Studios, Cambridge, MA 02139, USA,