DAMIÁN ISLA
Bungie Studios

Halo 3 Objective Trees:
A Declarative Approach to Multiagent Coordination
Encounters in Halo 3 are complex systems with many moving parts. Creating these systems - and scripting the movements and tactical decision-making of the many agents that make them up - is a tough problem. Not only do those many agents need to act tactically "smart", their movements and decisions also need to tell the story of the encounter, and provide a fun experience for the player. And of course, the tools for creating these systems need to lie firmly in the hands of the designers.

In this talk we will introduce Objective Trees, the encounter-scripting technique devised for Halo 3. We will discuss the general problem of encounter design and describe the development of the Objective Tree system, as well as delve into the core decision-making algorithm behind it all. Finally, we will discuss in broad terms what Objective Trees - and other declarative authoring paradigms - might mean for game AI.

Damian Isla is a Gameplay and Artificial Intelligence Lead at Bungie Studios, where he was responsible for the AI for mega-hit first-person shooters HALO 2 and HALO 3. Before coming to Bungie, he earned an M.Eng. at the MIT Media Lab's Synthetic Characters Group, where he did research on learning and behavior for artificial creatures. He has spoken on games and character AI at the Game Developers Conference, the International Joint Conference on Artificial Intelligence (IJCAI), the AI and Interactive Digital Entertainment Conference (AIIDE) and at Siggraph. He has a B.Sc. in Computer Science, again from MIT.

DOUG CHURCH and BORUT PFEIFER
Electronic Arts Los Angeles

In attempting to create high fidelity, believable human characters that can interact with a player more meaningful ways, we run into many problems revolving around an agent's inability to manage the many levels of decision making coordinating mind and body. Human-like agents must be able to effectively manage what would be both conscious and sub-conscious performance on the part of a real person. Current research and production methods of building AI characters are often oriented towards abstractly simulating intelligence, and not offering dramatic interaction possibilities - when we apply these methods towards this goal we run into communication problems conveying an agent's intent to a player because of the complexity of layers of behavior and performance (as well as scalability problems). This talk will look at some of the less addressed problem areas in the research while suggesting possible directions to push forward on this goal, based on Doug & Borut's experience at EA working on one of the Spielberg game projects.

Doug Church has been in the game industry since 1990, working on a variety of PC titles (Ultima Underworld, System Shock, Thief) at what became Looking Glass Studios. He left in late 1999 and consulted on a variety of titles like Ion Storm's Deus Ex, Harmonix's Frequency, some MMP's in Hong Kong. Doug also worked for Eidos where he did design direction, technology investigation, and management. In May 2005, he joined Electronic Arts Los Angeles as an executive producer and is currently working on a game collaboration with Steven Spielberg.

Borut Pfeifer has worked in the game industry since 2000, first on online casino games and then at his own startup White Knuckle Games. Eventually he went on to work as an AI and gameplay programmer at Radical Entertainment (on Scarface:The World Is Yours), and then as lead gameplay programmer on a PS3 launch title at Sony Online (Untold Legends: Dark Kingdom). He currently is at EA working as the AI lead on Doug's team.

ERIC GRUNDSTROM
Electronic Arts / Maxis

The AI of Spore

Spore is a game that takes the player on an epic journey, leading the gamer through a story with five distinct phases-Cell, Creature, Tribe, Civilization, and Space. Each phase has different game play and placed unique demands on the game AI. The approach to developing the various AI systems was heavily influenced by certain aspects of the game design. In particular, non-player characters do not follow the same game rules as the player. Instead, the NPCs are designed to fill specific roles in the player experience. Developing an asymmetric, player-facing AI is a different way of thinking about AI, and has the potential for wider application in games. In most video games the player has a unique role to play, so it only makes sense to acknowledge and embrace that fact when designing an interesting environment and compelling interactions for the gamer.

 Eric Grundstrom has been making computer games for most of his life, and has actually managed to get paid for it during the past twelve years. He is currently a senior software engineer at Electronic Arts, although on Spore he does double duty as an alien wrangler. Eric specializes in simulation and AI, and has worked on several best-selling games including The Sims 2 and SimCity 4. He has a BA in mathematics from Bucknell University, and a PhD in applied mathematics from the University of Maryland, where he researched artificial neural networks.

DAVID COPE
University of California, Santa Cruz

Experiments in Musical Intelligence
David Cope will explain why he created his computer program Experiments in Musical Intelligence, how this program works, why he created over 6,000 music compositions using Experiments in Musical Intelligence, and why he no longer composes using this software. The program will include a musical Turing test, a composed-on-the-spot computer composition, and the world premiere of a new work.

David Cope is Professor Emeritus of Music at the University of California at Santa Cruz, and teaches regularly in the annual Workshop in Algorithmic Computer Music (WACM) held in June-July at UC Santa Cruz. His books on the intersection of music and computer science include Computers and Musical Style, Experiments in Musical Intelligence, The Algorithmic Composer, Virtual Music, Computer Models of Musical Creativity, and Hidden Structure and describe the computer program Experiments in Musical Intelligence which he created in 1981. Experiments in Musical Intelligence’s works are published by Spectrum Press, spectrumpress.com and include Horizons for orchestra, three operas in the style of, and with librettos consisting of, letters by the composers Mozart, Schumann, and Mahler, and a symphony and piano concerto in the style of Mozart, and a seventh Brandenburg Concerto in the style of Bach. Experiments in Musical Intelligence's works are available on five Centaur Records’ albums (Bach by Design, Classical Music Composed by Computer, Virtual Mozart, Virtual Bach, Virtual Rachmaninoff)

STEVE RABIN
Nintendo of America & DigiPen Institute of Technology

The Past, Present, and Future of Game AI
Game AI has progressed at a steady pace since the early days when we were being chased by Blinky, Pinky, Inky, and Clyde. As we stroll through game history, game AI has continued to cheat, steal and slowly progress. Where is it headed and what challenges should we be working on? I'll bring a unique perspective from my time as an game AI programmer, over a decade at Nintendo, as editor of roughly 250 game AI articles in the AI Game Programming Wisdom series, and as an instructor of game AI at DigiPen.

Steve is a Principal Software Engineer at Nintendo of America, where he researches new techniques for Nintendo's next generation systems, develops tools, and supports Nintendo developers. Before Nintendo, Steve worked primarily as an AI engineer at several Seattle start-ups including Gas Powered Games, WizBang Software Productions, and Surreal Software. He managed and edited the AI Game Programming Wisdom series of books, the book Introduction to Game Development, and has over a dozen articles published in the Game Programming Gems series. He's spoken at the Game Developers Conference and has moderated the AI roundtables. Steve teaches artificial intelligence at both the University of Washington Extension and at the DigiPen Institute of Technology. He earned a B.S. in Computer Engineering and a M.S. in Computer Science, both from the University of Washington.