Artificial Intelligence Project

Objective of the Project

The aim of the project is the implementation (using Python 2.7 as interpreter) of a Planner for solving an Artificial Intelligence planning problem with SAT algorithms (SAT based planner).

Given an instance specified by the initial and the goal states, as well as the actions that may be performed and a horizon length, the algorithm exploits the frame axioms and builds a Propositional Formula. The boolean formula is solved using Minisat. If the latter is satisfiable, the ENHSP will find an assignment which would be translated into the desired plan (a sequence of actions) to reach the objective. Otherwise a new encoding will be generated reflecting a longer plan length.

How to run

Clone the repository:

git clone https://github.com/MLagomarsino/Artificial_Intelligence_project.git

Install the required libraries and the Minisat solver:

cd ~/Artificial_Intelligence_project
pip install -r z3
pip install -r satispy
sudo apt-get update
sudo apt-get install minisat

From the workspace folder, install and compile the ENHSP module with the following commands:

cd code/enhsp
./install
./compile

Moreover in bin folder:

cd ..
cd bin
./validate

Run the project using Python 2.7 as interpreter. From terminal with command:

plan.py [-h] [-domain <path_pddl_domain> <path_pddl_problem>] [-linear] [-parallel] [-pprint]

Or loading for example the following configuration in PyCharm:

Script path: ../Artificial_Intelligence_project/code/plan.py

Parameters: -domain ../Artificial_Intelligence_project/domains/blockworld/domain.pddl ../Artificial_Intelligence_project/domains/blockworld/instances/pb3.pddl

Working directory: ../Artificial_Intelligence_project/code

Author

Name	ID	E-mail
Marta Lagomarsino	4213518	[email protected]