Jan. 7, 2010: Web page created
Playing games successfully requires players to devise action sequences that accomplish certain goals. In other words, they are solving planning problems. With such a tight relationship between action planning and playing games, it is surprising that planning and AI in games research have evolved for several decades without much interaction.
The goal of this second ICAPS workshop on planning in games is to bring researchers in both fields together to discuss how to plan actions in large game state spaces, ranging from complex abstract games such as go and chess to modern video game genres like role playing and real-time strategy games. These games test the limits of traditional game-tree search approaches, and also involve many current topics of interest in planning such as: temporal and spatial reasoning, resource management, cost-based planning, and imperfect information.
We therefore think it is a good idea to organize this workshop and invite submissions on all aspects of planning in the domain of games. Relevant topics include, but are not limited to:
We ask authors to submit technical papers in PDF format. Papers should be formatted in accordance with the AAAI style template and may be at most 8 pages long, including figures and bibliography. Visit http://www.aaai.org/Publications/Author/author.php for formatting instructions. Please submit papers using EasyChair.
Please note that all submitted papers will be carefully peer-reviewed by multiple reviewers, and that low-quality or off-topic papers will not be accepted. Also note that all workshop participants must register for the main ICAPS-10 conference.ICAPS Workshop on Planning in Games (2007)
(presentations + QA : 27 minutes each, 1h for invited talk, 10 minutes break)
Invited PresentationSpeaker: Jeff Orkin
TITLE: The Evolution of Planning in Games
Over the past five years in the game industry, planning has gained acceptance as a viable, or even superior, solution for generating character behavior in games. No longer considered an exotic approach, planning has proven itself in numerous successful commercial titles. Approaches to planning in games have evolved from STRIPS-like to hierarchical planners. Challenges for AI in games today include authoring increasingly complex behavior, implementing effective co-operative teammates, and managing non-linear storytelling. Recent academic planning research holds potential solutions, including work on planners for story generation and interactive narratives, and data-mining and case-based planning for RTS games. In this talk, I will describe how planning has evolved in commercial games, and where there are opportunities to apply academic research to solve current industry problems. Finally, I will describe my current work on Collective AI, where we are generating social behavior and dialogue for co-operative characters with a case-based planning system that draws from data mined from over 15,000 players of The Restaurant Game.
Jeff Orkin is a game developer, AI researcher, and PhD candidate in the Cognitive Machines Group at the MIT Media Lab. Jeff's research focuses on Collective Artificial Intelligence -- a combination of -- -- crowd-sourcing, data-mining, and case-based planning designed to empower non-experts to author socially rich digital actors who can play roles in games, simulations, and virtual worlds. These socially intelligent characters are capable of communication and co-operation with humans and other characters. Prior to enrolling at the Media Lab, Jeff developed several generations of AI systems in the game industry. As a Senior Engineer at Monolith Productions, Jeff focused on goal-oriented autonomous character behavior and planning, while developing AI systems for the award winning titles No One Lives Forever 2 (NOLF2) and F.E.A.R. Jeff is a Contributing Author and Section Editor of the AI Game Programming Wisdom book series, has presented at the Game Developer's Conference, AIIDE, and AAMAS, and holds a Master's degree in Computer Science from the University of Washington and Bachelor's degree in Computer Science from Tufts University with a minor in Studio Art.
Yngvi Bjornsson, Reykjavik University, Iceland