IPC website update: domains, baseline planners & more

2 views
Skip to first unread message

Malte Helmert

unread,
Mar 11, 2009, 5:21:37 PM3/11/09
to IPC-2008
Dear IPC-2008 group,

with the ICAPS deadline approaching, we get more and more requests for
the IPC-2008 domains, so we finally took the time to make the
following long overdue updates to the competition website:

1. Competition domains now available for download.

Go to http://ipc.informatik.uni-freiburg.de/Domains for details.
WARNING: There is a bug in the numeric formulations of temporal
openstacks. Don't use them. (The warning also appears on the
referenced web page.)

2. Full competition results now available for download.

It's a huge archive and it's very raw data (all plans, all planner
logs, all everything), so for most purposes the summarized results we
presented at ICAPS 2008 (which remain available at
http://ipc.informatik.uni-freiburg.de/Presentations) should be better.
Still, if you want the gory details, they are now available from the
IPC-2008 home page at http://ipc.informatik.uni-freiburg.de/ under
"Post-IPC"/"Raw competition results". (For those of you who have
contacted me individually about this before: it's the same
data.tar.bz2 file you already know.)

3. Planner page overhauled.

The page which contains the source code and papers for the
participating planners has been overhauled. Most importantly, we've
added links for the source code of the three baseline planners. Go to
http://ipc.informatik.uni-freiburg.de/Planners for details. All
participants are encouraged to add links to the participants' and/or
planner homepages to their planner entries. See the entry for LAMA for
an example.

4. Minor terminology prettifications.

The competition is now called "IPC-2008" everywhere. (Previously, it
was some mixture of "IPC-6", "IPC6" and "IPC-2008".) We think the year
notation is better than the counter notation because it is most
memorable. (Who knows off-hand that ICAPS 2009 is the 19th ICAPS?) The
tracks previously referred to as "optimal tracks" are now called
"optimization tracks" everywhere, which makes more sense.

5. Acknowledgements.

Last not least, we put up an acknowledgements page for the many people
who contributed to the organization of IPC-2008:
http://ipc.informatik.uni-freiburg.de/OrganizerAcknowledgements.

Best regards,

Malte

Malte Helmert

unread,
Mar 11, 2009, 5:37:55 PM3/11/09
to IPC-2008
On Mar 11, 10:21 pm, Malte Helmert <malte.helm...@googlemail.com>
wrote:
> Dear IPC-2008 group,
>
> with the ICAPS deadline approaching, we get more and more requests for
> the IPC-2008 domains, so we finally took the time to make the
> following long overdue updates to the competition website:

If I talk about what we *have* done, maybe I should also briefly talk
about what *remains* to be done. These are the two major things
related to IPC-2008 that we haven't done yet, but are planning to do:

1. Descriptions of the domains and instances that were used.

We are planning to write about this in more detail at some point,
perhaps in the form of a competition paper like the ones that were
written for the last IPCs. There is no timeline for this yet. Note
that for some domains, the IPC website already provides some
additional information, which is linked from http://ipc.informatik.uni-freiburg.de/Domains.

2. Things related to PDDL 3.1.

We hope to provide a more comprehensive description of the language
extensions in PDDL 3.1 (most importantly, object fluents) at some
point. There is a bit of a chicken-and-egg problem with object
fluents. If planners don't make use of the feature, there's little
incentive to write domains that use it. If no benchmarks are available
with this feature, few people will implement a planner for it. We hope
to eventually break this cycle by providing object-fluent formulations
of all IPC-2008 competition domains. One of the main things holding
this up at the moment is tool support, which is a difficult problem.
So don't expect any updates on this topic soon.

Best regards,

Malte
Reply all
Reply to author
Forward
0 new messages