==============================================================================
                  2027 International Planning Competition
                             Classical Tracks

                   Preliminary Schedule and Call for Domains

                    https://ipc2027-classical.github.io
                    ipc2027-...@googlegroups.com
==============================================================================

We are delighted to announce the classical track of the 2027 International
Planning Competition (IPC 2027). The competition is organized in the context
of the International Conference on Automated Planning and Scheduling (ICAPS)
and will conclude at ICAPS 2027.

The IPC empirically evaluates state-of-the-art planning systems on a range of
benchmark problems. Its goals are to promote planning research, highlight
challenges for the planning community, and provide new and interesting
benchmarks for future research.


Call for domains
------------------------------------------------------------------------------

We invite submissions of planning domains for the competition. If you would
like to contribute a domain, or already have one to share, please contact us at
ipc2027-...@googlegroups.com before the domain submission deadline below.
An informal email expressing your interest is sufficient at this stage. We will
work with domain submitters to assess and, where necessary, adapt their domains
for use in the IPC.

Submissions should ideally include a problem generator whose parameters allow
the difficulty of instances to be controlled and scaled. Domains related to
real-world applications are particularly welcome, although this is not a
requirement. If a domain-specific solver or another way method for finding high-
quality or optimal solutions, or of proving instances unsolvable, is available,
please include it as well.

We will consider all submissions and select a high-quality subset for use in
the competition. In particular, we are interested in domains that

 - model a real-world application or another compelling planning problem,
 - come with a generator offering meaningful control over instance difficulty,
 - exhibit interesting intrinsic challenges rather than difficulty arising
   only from scaling up a simple problem,
 - include unsolvable instances that are challenging to recognize or prove
   unsolvable,
 - pose substantial challenges for grounding, for example by producing large
   grounded representations even from compact PDDL input, and
 - provide a way to obtain high-quality or optimal solutions, or to certify
   unsolvability, for example via a domain-specific solver or suitable generator
   construction.

We recognize that participants whose submitted domains are selected may have
some advantage on those domains. We regard this as an incentive to contribute
strong domain proposals.

One goal of the IPC is to create a new, publicly available benchmark suite for
future research. We therefore plan to publish all selected domains, generators,
and related material in a public repository. By submitting a domain, you agree
that we may do so. We strongly recommend releasing submissions under a
permissive license, so that others can use and build upon them.

If you do not have a domain yourself but know someone with an interesting
planning problem, please forward this call to them. If you are unsure whether a
domain would be suitable, do not hesitate to contact us.


PDDL fragment and tracks
------------------------------------------------------------------------------

IPC 2027 will use the same subset of PDDL as the recent classical tracks. Domains
may use STRIPS, action costs, negative preconditions, and conditional effects,
including their combination with universally quantified effects. We will also consider
domains with disjunctive preconditions and existential quantification. Domains using
axioms are welcome as well.

As in the previous IPC, we are especially interested in domains and instances
that pose challenges for grounding, since many classical planners instantiate
the entire task before search. We also explicitly welcome domains with
unsolvable instances, particularly when proving unsolvability presents a
meaningful challenge.

The precise set of tracks has not yet been finalized. We currently expect
optimal, satisficing, and agile tracks along the lines of previous editions,
possibly with adaptations or extensions depending on the submissions received.


Preliminary schedule
------------------------------------------------------------------------------

Call for domains:                         September 2026
Call for participation:                   Autumn 2026
Domain submission deadline:               Early 2027
Planner registration:                     Early 2027
Feature stop (final planner submission):  March-April 2027
Planner abstract submission deadline:     May 2027
Contest run:                              May-June 2027
Results announced:                        ICAPS 2027

We will announce exact dates and submission instructions by email and on the
competition website: https://ipc2027-classical.github.io.


Organizers
------------------------------------------------------------------------------

Clemens Büchner (University of Basel)
Daniel Gnad (Heidelberg University & Linköping University)
Arnaud Lequen (Linköping University)
David Speck (University of Basel)

Contact: ipc2027-classical@googlegroups.com

==============================================================================