PDDL-RM
Generating Reward Machines from high-level PDDL models using Partial-Ordered Plan (Adapted from POP-GEN, QRM)
Installation
git clone https://github.com/yanxi0830/pop-rm.git
cd pop-rm
conda create --name py27 python=2.7 # Python 2 for now, TODO: change to Python 3
conda activate py27
python -m pip install -U matplotlib
pip install networkx==2.2
# Gurobi
# Need to get license..
conda config --add channels http://conda.anaconda.org/gurobi
conda install gurobi
grbgetkey LICENSE-CODE
Run
cd src
python run.py --domain ../domains/craft/domain.pddl --problem ../domains/craft/t9.pddl --plan ../domains/craft/t9.plan
Example
Given the task of getting gold, and the rules of Minecraft described in PDDL. This package encodes the actions sequences into a finite state machine.
