Problems in running fast-downward with my pddl domain / problem

[Mike Joung]

Hi everyone, I am trying  to implement the domain and the problem for the stacking container problem in PDDL. The implementation is based on the paper " M. Salido, O. Sapena, and F. Barber, ‘The container stacking problem: an artificial intelligence planning-based approach’, The International Workshop on Harbour, Maritime and Multimodal Logistics Modelling and Simulation, (2009) ". When I run fast-downward, using the command:

./fast-downward.py 'PDDL/container_stacking/container_stacking_domain.pddl' 'PDDL/container_stacking/container_stacking_problem.pddl' --search "astar(lmcut())" 

I got the following report:

INFO     Running translator.
INFO     translator stdin: None
INFO     translator time limit: None
INFO     translator memory limit: None
INFO     translator command line string: /usr/bin/python3 /home/kevin/Desktop/PlanningReasoning/downward/builds/release/bin/translate/translate.py PDDL/container_stacking/container_stacking_domain.pddl PDDL/container_stacking/container_stacking_problem.pddl --sas-file output.sas
Parsing...
b'Undeclared predicate: <\n'
translate exit code: 30

Driver aborting after translate

I cannot understand what is the undeclared predicate. To be clearer I attached to this message also the pddl domain and problem files I passed as argument to the execution command.

Thank you for getting attention to my post.

Best regards

[Christian Muise]

Hi Mike,

To share easier, I'd suggest posting your PDDL on the http://editor.planning.domains site (you can use the Session menu to save a session and get a read-only link). Regardless, it looks as though the domain is using a numeric comparator (the "<" symbol) when the planner doesn't support such an operation. I would suggest looking into which planner was actually used for the research.

Best,

Christian

--
You received this message because you are subscribed to the Google Groups "Fast Downward" group.
To unsubscribe from this group and stop receiving emails from it, send an email to fast-downwar...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/fast-downward/775c826b-b49c-4495-b6fb-d8b2ff59e8f0%40googlegroups.com.

[Malte Helmert]

Dear Mike,

I think you forget to attach the PDDL files. From the error message, it
looks like the PDDL file uses the "<" ("less than") predicate, which
suggests that this is a numerical planning problem. Fast Downward only
supports PDDL level 1, which excludes numerical planning features other
than action costs.

So it looks like what you need is a numerical planner. I cannot give you
recommendations myself, but perhaps others reading here can.

If this is *not* a numerical planning problem, I think we would need to
take a closer look at the PDDL files.

Best regards,
Malte

[Mike Joung]

Hi Christian,

Thank you for the answer. You're right, indeed the pddl files should already have been sent in attached to my previous message, but probably some trouble occurred when I posted my question.

Here is the link to the pddl editor session: http://editor.planning.domains/#read_session=GKD1FNtUQ7

Thank you again for the answer.

Best regards

To unsubscribe from this group and stop receiving emails from it, send an email to fast-d...@googlegroups.com.

[Christian Muise]

Sure thing. As both Malte and I point out, the issue is coming from the type of planning formalism it uses. FastDownward is not a suitable choice for numeric planning.

Cheers,

Christian

To unsubscribe from this group and stop receiving emails from it, send an email to fast-downwar...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/fast-downward/25ab4046-e0cc-4113-8686-56459f760488%40googlegroups.com.

[Mike Joung]

Hi,

I reaaly thanks a lot Malte and Christian, for the answers.

Thus in fast-downward is not recommended or not even possibile to define and evaluate functions / fluents to compute the cost of an action or check for some constraint (as the height of containers stacks, in the considered planning problem) ?

Best regards

To view this discussion on the web visit https://groups.google.com/d/msgid/fast-downward/25ab4046-e0cc-4113-8686-56459f760488%40googlegroups.com.

[Malte Helmert]

Hi Mike,

numerical fluents can be used in Fast Downward to define the costs of
actions, but *only* for this. Action costs are the only form of numbers
supported by Fast Downward. Other uses will lead to error messages like
the one you have seen, since this part of PDDL is not implemented in
Fast Downward.

PDDL is a language with many different levels of expressivity. The one
supported by Fast Downward is called "level 1" and doesn't include
numbers except for action costs and doesn't support temporal planning
features. The same is true for the majority of planning system.

Christian's advice is good: if your starting point is a planning domain
from a specific paper, then it would make sense to look at the planners
discussed in that paper, assuming that it includes some form of
experimental evaluation.

Best regards,
Malte

> <http://www.google.com/url?q=http%3A%2F%2Feditor.planning.domains&sa=D&sntz=1&usg=AFQjCNEsPFEnBRde2oHT2c9IcpD6AuvcLA>

> <https://groups.google.com/d/msgid/fast-downward/775c826b-b49c-4495-b6fb-d8b2ff59e8f0%40googlegroups.com?utm_medium=email&utm_source=footer>.

>
> --
> You received this message because you are subscribed to the
> Google Groups "Fast Downward" group.
> To unsubscribe from this group and stop receiving emails from

> it, send an email to fast-d...@googlegroups.com <javascript:>.

> To view this discussion on the web visit

> https://groups.google.com/d/msgid/fast-downward/25ab4046-e0cc-4113-8686-56459f760488%40googlegroups.com
> <https://groups.google.com/d/msgid/fast-downward/25ab4046-e0cc-4113-8686-56459f760488%40googlegroups.com?utm_medium=email&utm_source=footer>.

>
> --
> You received this message because you are subscribed to the Google
> Groups "Fast Downward" group.
> To unsubscribe from this group and stop receiving emails from it, send

> an email to fast-downwar...@googlegroups.com
> <mailto:fast-downwar...@googlegroups.com>.

> To view this discussion on the web visit

> https://groups.google.com/d/msgid/fast-downward/ee2eb96c-c255-4569-8537-bdb63de2a707%40googlegroups.com
> <https://groups.google.com/d/msgid/fast-downward/ee2eb96c-c255-4569-8537-bdb63de2a707%40googlegroups.com?utm_medium=email&utm_source=footer>.

[Mike Joung]

Hi,

Thank you for the clarifications.

I wrnt on in searching about the issue, and I found this old post, in which a method to 'move around' the problem is suggested (I report the link to the post and the relative answer: https://groups.google.com/d/msg/fast-downward/DPSiqAZglsU/qhWko3afqxEJ ).
Thus I would ask if this method could be applied also to my problem.

I already tried to apply it, but I am having some errors, though about syntactic issues, so I do not know if I'm applyting the proposed method correctly.

I updated the pddl files posted previously, if thay could be useful to make the issue clearer.

Thanks again for the answers.

Best regards

> an email to fast-d...@googlegroups.com
> <mailto:fast-downward+unsub...@googlegroups.com>.