QuMuLab / pdkb-planning

Project for converting multi-agent epistemic planning problems to automated planning problems. Also includes utilities for dealing with Proper Doxastic Knowledge Bases (PDKBs).

Multi-agent Epistemic Planning With Proper Doxastic Knowledge Bases

This repository houses the code for solving Multi-agent Epistemic Planning (MEP) problems using Proper Doxastic Knowledge Bases (PDKB). There are a number of components that may be useful on their own, including working directing with PDKB's, augmenting the MEP formalism, or even just creating / dealing with KD45 kripke structures.

Eventually, more documentation will be added to the project, but feel free to contact me if you have any questions. For a demo and more information on the project, [click here].

Getting started

Using Docker

docker build -t pdkbplanning:latest .
docker run -it pdkbplanning

You end up in a shell running in the Docker container, which has all the required tools installed. Then you can run the planner on an example PKDBDDL problem:

python3 -m pdkb.planner /MEP/pdkb-planning/examples/planning/grapevine/prob-paper1.pdkbddl

Requirements

• Graphviz
• NetworkX
• Pygraphviz