You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Would it be technically possible to have a module in AI-Toolbox capable of loading MDP/POMDP models in RDDL and/or PPDDL format (which seems to be the "official" formats for the international planning competitions IPC) ?
There is a lot of planning domains / models in these formats and it would be cool to be able to easily test the AI-Toolbox algorithms on them.
The text was updated successfully, but these errors were encountered:
I have to admit I'm not really familiar with these formats. If it's just a matter of parsing a file and extracting transition probabilities and expected rewards, I don't see why not. If there's additional metadata, or the MDPs have properties we don't yet map, we'd have to write an additional model class to handle everything.
All in all, I don't think there is any technical reason why something like this can't be added to the library. Would you be interested in trying? I'd be glad to help you with the codebase and how I like to organize things (and possibly to help write any additional model classes if they are needed), but I don't think I'd be able to help with the parser itself.
Would it be technically possible to have a module in AI-Toolbox capable of loading MDP/POMDP models in RDDL and/or PPDDL format (which seems to be the "official" formats for the international planning competitions IPC) ?
There is a lot of planning domains / models in these formats and it would be cool to be able to easily test the AI-Toolbox algorithms on them.
The text was updated successfully, but these errors were encountered: