W6. International Planning Competition: Past, Present and Future

List of Papers

Chairs

Minh B. Do, Palo Alto Research Center, USA
Alan Fern, Oregon State University, USA
Malte Helmert, Albert-Ludwigs-Universitaet Freiburg, Germany
Ioannis Refanidis, University of Macedonia, Greece

Preface

The International Planning Competition (IPC) started in 1998. It was created to set the common ground for comparison between different planning techniques tackling different planning problems. For this purpose, it has been a success in shaping up the standard planning language (PDDL), identifying important extensions, gathering a set of benchmarks for different planning problems, and providing tools for building and evaluating planners. In the passing of the years several critiques have been raised concerning the necessity and the usefulness of several aspects of the competition. This workshop aims at reviewing the current status of the international planning competition and determine/sketch/prepare the forthcoming event (6th IPC).

Through a number of high-quality papers received and a carefully decided set of open discussions/panels, the workshop aims to treat topics such as:

Language: Review of PDDL 3, assessment, comparison with other modeling languages and proposals for possible extensions.
Evaluation Criteria: Review/critic on the current evaluation approach and proposals for changes/standardization. Whether or not a single objective function is adequate to compare different planners in different tracks, or a multi-objective function is needed. Effective ways to compare fully automated planners, planners with learned or hand-crafted control knowledge.
Domains: Review of current domains, wrt their difficulty/complexity, closeness to real-world applications, critic/review on the hardness of different domains and whether or not the current problem sets are adequate in representing the technical challenges in different classes. Proposal for new domains, particularly ones that can capture most critical constraints in some real-world applications as well as discussion on applications for which planning is suitable.
Existing Tracks: Review of the existing tracks, in particular what worked, what did not, and what needs to be changed/improved in each track.
Learning Track: Develop plans for the new learning track of the competition.
Proposed Tracks: Review proposals for new competition tracks.
Planners: Review of the performance of various types of planners (e.g. model checking, satisfiability, state-space heuristic, local search etc) in the past competitions, possibly wrt various types of domains. Identifying problem structures that are suitable for different types of planner and why.
Review of the evolution of planners and planning techniques throughout the competition history. What kind of planner do we seek?

Committee

Blai Bonet, Universidad Simón Bolívar Caracas, Venezuela
Hector Geffner, Universitat Pompeu Fabra, Spain
Alfonso Gerevini, Università degli Studi di Brescia, Italy
Patrik Haslum, National ICT Australia, Australia
Jörg Hoffmann, University of Innsbruck, Austria
Subbarao Kambhampati, Arizona State University, USA
Henry Kautz, University of Rochester, USA
Derek Long, University of Strathclyde, UK
Alessandro Saetti, Università degli Studi di Brescia, Italy
David Smith, NASA Ames Research Center, USA
Sylvie Thiebaux, Australian National University, Australia