Replies: 1 comment 14 replies
-
Can you post the both the working and broken PDDL files? |
Beta Was this translation helpful? Give feedback.
14 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hi there,
So I'm trying to get the functionality for traces with dynamic goals working, but I'm getting some issues on what I believe to be the tarski side of things. I set up a simple test where I just take part of the original goal, just to see if I could successfully re-write the problem file and generate a plan. (I'm just testing with the blocks problem/domain right now):
https://github.com/QuMuLab/macq/blob/3efb33ad767f9a62649445e2a23341c63652a766/macq/generate/pddl/generator.py#L220
When I try to compute a plan though, I get an error at this line, that
resp
does not generate a plan: https://github.com/QuMuLab/macq/blob/3efb33ad767f9a62649445e2a23341c63652a766/macq/generate/pddl/generator.py#L241The issue being that there was an error parsing the problem because the domain types and problem types don't like up - the domain (original file) is using "default object" while the problem is using "object." I figure that tarski deals with default objects differently, which is causing the mismatch. However, when I tried to also regenerate the domain file in order to match with the format of the problem one, I then get the issue that the domain doesn't have any types at all (think it's caused by this?)
https://github.com/aig-upf/tarski/blob/80a33dc5fe6ad2602b40b1c69f758b91188dd8e1/src/tarski/io/fstrips.py#L238
while problem is still using "object," so I get the same parsing error.
TLDR: main issue looks to be a typing inconsistency with tarski... ;P
Beta Was this translation helpful? Give feedback.
All reactions