pbcplus2pomdp

Python scripts that translate probabilistic action descriptions in pBC+ (encodede as lpmln programs) to POMDP models. The outputs are .pomdp files that can be used as input to POMDP solver APPL (https://github.com/AdaCompNUS/sarsop).

Usage

Noncompositional generation:

$ python pbcplus2pomdp.py path/to/lpmln/files discount_factor(0-1)

Compositional generation:

$ python pbcplus2pomdp_compositional.py path/to/lpmln/file/with/no/action path/to/lpmln/action/description/files discount_factor(0-1)

For example,

$ python pbcplus2pomdp.py examples/tiger.lpmln 0.95

$ python pbcplus2pomdp_compositional.py examples/dialog/dialog-comp/dialog_noaction.lpmln examples/dialog/dialog-comp/actions 0.95

System Dependencies

This Python script requires the following system to be installed

• Python 2.7
• clingo python library: https://github.com/potassco/clingo/blob/master/INSTALL.md
• lpmln2asp system: http://reasoning.eas.asu.edu/lpmln/index.html

Examples

The folder "examples" contains lpmln encodings for several example pBC+ action description.

• tiger.lpmln: the well-known tiger example in POMDP literature

• two_tiger.lpmln: a variation of the tiger example where there are two tigers and three doors

• dialog: an example from 1

References

[1] Zhang, Shiqi, Piyush Khandelwal, and Peter Stone. "Dynamically constructed (po) mdps for adaptive robot planning." Thirty-First AAAI Conference on Artificial Intelligence. 2017.