Workshop on Roles of Planning in Games (RPG)

ICAPS 2026 Workshop
Dublin, Ireland June 28, 2026

RPG Website

Aim and Scope of the Workshop

Games and virtual environments have long served as test beds for Artificial Intelligence (AI), providing research challenges and benchmarks for assessment of intelligent systems performance. The specific application of Automated Planning and Scheduling (APS) technologies in games has been gaining popularity for uses both within the gameplay experience and for designing them.

The Roles of Planning in Games (RPG) workshop aims to bring together researchers and practitioners to explore the unique challenges and contributions that rise from applying APS techniques to games. The focus is specifically on the aspects that directly interest the ICAPS community such as reasoning, resource management, hierarchical goal decomposition, and search-based methods. The focus on games will also encourage designers and developers to present their needs, wants, and challenges to the ICAPS community while having an opportunity to learn more about the APS resources available to them. This workshop will provide a dedicated space at the intersection of APS research and game applications, fostering collaboration between both communities and ideally connecting industry with academics.

Please send any inquiries to: rpg26@easychair.org

Important Dates

  • Paper submission deadline: May 4th May 11th, 2026
  • Notification of acceptance: June 11th, 2026
  • Workshop date: June 28 (14.30 - 18.00 IST), 2026

The reference timezone for all deadlines is UTC-12. That is, as long as there is still some place anywhere in the world where the deadline has not yet passed, you are on time!

Schedule

  • 14:30 - 14:35 | Welcome
  • 14:35 - 15:20 | Invited Talk: Rogelio Cardona-Rivera — Planning for the Design of Experiences
  • 15:20 - 16:00 | Paper Presentations (Theme: Planning-based Tools for Games)
    • 15:20 - 15:40 | From Domain to Action: A Cohesive Planning Pipeline for Game AI in Unity Engine — Marek Marchlewicz
    • 15:40 - 16:00 | Dynamic Scene Reconstruction for Prototyping Planning-Based Game Environments — Albaraa Othman, Emanuele De Pellegrin, and Ron Petrick
  • 16:00 - 16:20 | Coffee Break
  • 16:20 - 17:10 | Paper Presentations (Theme: Applications of Games for Planning) - now with correct order!
    • 16:20 - 16:35 | Baba Is Axioms: A Challenging Benchmark for Classical Planning with Axioms — Ronny Rochwerg, Allix Fletcher, Samantha Papais, Cynthia Wang, Connor Little, Ting Hu, and Christian Muise
    • 16:35 - 16:50| A Good Snowman Is (also Numerically) Hard to Plan — Matteo Paparo, Francesco Doria, Marco Maratea, and Mauro Vallati
    • 16:50 - 17:10 | Active Information Gathering in Social Deduction Game — Haochong Luo, Chenyuan Zhang, and Guang Hu
  • 17:10 - 17:55 | Invited Talk: Jeff Orkin — Planning in Games in the Age of LLMs
  • 17:55 - 18:00 | Closing

Keynote Speakers

Rogelio Cardona-Rivera (University of Utah)

  • Talk Title: Planning for the Design of Experiences
  • Abstract: Automated planning has long been viewed as a problem-solving paradigm for generating sequences of actions that achieve specified goals. In the context of games, however, its role can be far richer: beyond decision-making, it can serve as a representational, organizational, and interactional framework. In this talk, I will offer a reflective view on my prior work for planning the design of playful experiences and chart potential futures centered on the idea of planning as foundational to shaping how games are designed, experienced, and understood.
Dr. Rogelio E. Cardona-Rivera

Bio: Dr. Rogelio E. Cardona-Rivera is a Puerto Rican computational game design scientist. Rogelio co-founded the Division of Games at the University of Utah, where they serve as an Assistant Professor (with courtesy appointments in Computer Science and Psychology). Prior to joining the faculty at Utah, Dr. Cardona-Rivera was a research scientist at Sandia National Laboratories and Disney Research.
At Utah, Rogelio directs the Laboratory for Quantitative Experience Design, a community of scholars who build plan-based models of storygames and investigate their relation to player psychology and experience. Prof. Cardona-Rivera obtained the BS degree in Computer Engineering from the University of Puerto Rico at Mayaguez and the PhD degree in Computer Science from North Carolina State University in the United States. Rogelio has authored over 30 articles and books and has obtained nearly 1.5 million (USD) in support for their research projects toward establishing a science of game design, including CAREER and collaborative grants from the U.S. National Science Foundation, sponsorships from the games industry via Oculus VR and Activision | Blizzard, and grants from the Department of Energy, IARPA, and the US Naval Research Laboratory.

Jeff Orkin (Bitpart AI)

  • Talk Title: Planning in Games in the Age of LLMs
  • Abstract: Bitpart is building a multi-agent, multi-modal hierarchical planning system for games, to puppeteer groups of NPCs in living worlds, and narrative experiences. This talk will explore our approach, discuss technical design decisions we’ve made, and challenges we’ve faced implementing a powerful planner. In the process, we will consider how LLMs and planners can work together, to balance the pros and cons of each, and achieve the best of both worlds.
Dr. Jeff Orkin

Bio: Dr. Jeff Orkin is an AI researcher, entrepreneur, and game industry veteran. He is the Co-founder and CEO of Bitpart AI, a startup focused on building advanced NPC systems that combine Large Language Models (LLMs) with automated planning to enable more dynamic and believable game worlds. Prior to Bitpart AI, he founded Giant Otter Technologies, a conversational AI company that emerged from his PhD research at the MIT Media Lab, where he worked on learning character behavior from human gameplay data.
Jeff spent over a decade in the game industry as an AI specialist at Sierra and Monolith Productions, contributing to landmark titles such as No One Lives Forever 2 and F.E.A.R., the latter of which is widely recognized for pioneering the use of Goal-Oriented Action Planning (GOAP) in commercial games. His work has been influential in shaping modern game AI systems that balance planning, behavior trees, and emergent agent behavior.
He has co-authored volumes in the AI Game Programming Wisdom series, served as Associate Editor for the IEEE Transactions on Computational Intelligence and AI in Games, and has presented at major conferences including GDC, AIIDE, IEEE CIG, AAAI, AAMAS, ICAPS, and SARA. He holds a Master’s degree in Computer Science from the University of Washington and a Bachelor’s degree in Computer Science from Tufts University, with a minor in Studio Art.

Accepted Papers

  • A Good Snowman Is (also Numerically) Hard to Plan, Matteo Paparo, Francesco Doria, Marco Maratea and Mauro Vallati
  • Active Information Gathering in Social Deduction Game, Haochong Luo, Chenyuan Zhang and Guang Hu
  • From Domain to Action: A Cohesive Planning Pipeline for Game AI in Unity Engine, Marek Marchlewicz
  • Baba Is Axioms: A Challenging Benchmark for Classical Planning with Axioms, Ronny Rochwerg, Allix Fletcher, Samantha Papais, Cynthia Wang, Connor Little, Ting Hu and Christian Muise
  • Dynamic Scene Reconstruction for Prototyping Planning-Based Game Environments, Albaraa Othman, Emanuele De Pellegrin and Ron Petrick

Topics of Interest

Topics include, but are not limited to:

  • Decision Making in Games: Domain-specific heuristics, knowledge representation, search methods, and planners for playing and understanding existing games.
  • Real-Time Planning: Techniques for interleaving planning and execution under strict time and memory constraints.
  • Planning for Quality Assurance: Using automated planners to generate test coverage, detect soft-locks, and validate playable content autonomously.
  • Neurosymbolic AI in Games: Combining neuro-inspired models with symbolic planning to create NPCs that are both conversational and goal-directed.
  • Hierarchical Planning: Advances in hierarchical planning (e.g. Hierarchical Task Networks) tailored for controlling agent behaviours or developing structured game content.
  • Plan, Activity, and Goal Recognition: Inferring player intent to adapt game difficulty or content dynamically.
  • Procedural Content Generation: Constraint-based planning for generating game content such as solvable levels, puzzles, and narratives.
  • Planning for Game Design: Applications of APS to game or level design for videogames and/or tabletop games.
  • Representing Games as Planning Problems: Modeling game domains for (semi-) automated problem solving systems.
  • Tools for Planning in Games: Software or libraries using games that make use of planning.

Submission Instructions

Submit papers via EasyChair using the ICAPS 2026 paper template.

  • Regular papers: up to 8 pages + unlimited references
  • Short / position papers: up to 4 pages + unlimited references
  • All submissions must be in PDF format

At least one author of each accepted paper must register for the ICAPS 2026 workshop day. Accepted papers will be made available on the workshop website.

Program Committee

  • Lucas Bergholz, University of Brasilia
  • Andy Edmondson, Heriot-Watt University
  • Todd Neller, Gettysburg College
  • Ron Petrick, Heriot-Watt University
  • Ioannis Refanidis, University of Macedonia
  • Bruno César Ribas, University of Brasilia

Organizing Committee