Dear David,
I'm not sure I completely understand your problem description but it
sounds like there is a combinatorial explosion during grounding. There
is no explicit limit on the number of types or objects but using too
many of them can lead to number of ground actions that is so high, that
it becomes problematic to even represent the problem. Have you computed
(on a theoretical level), how many possibilities there are to
instantiate an action? For example, if you have 2k action schemas, each
with 3 parameters and there are 20 possible choices for each parameter,
you'd get 2000 * 20^3 ground actions. A back-of-the-envelope computation
like this can probably give a hint whether the problem is still
realistically groundable.
If grounding is the problem but the actual search in the problem is
easy, a lifted planner might be an option. We are working on something
like this and could use additional benchmark domains. If you don't mind
sharing your domain, we could try running it in the lifted planner. If
that also doesn't help (for example, if both search and grounding are
difficult), then you might have to adapt the model.
Cheers
Florian
> --
> 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/91a2d342-23f3-4b22-8bb3-96026a9b7ee4n%40googlegroups.com
> <
https://groups.google.com/d/msgid/fast-downward/91a2d342-23f3-4b22-8bb3-96026a9b7ee4n%40googlegroups.com?utm_medium=email&utm_source=footer>.