Pyperplan is a lightweight STRIPS planner written in Python.
Please note that Pyperplan deliberately prefers clean code over fast code. It is designed to be used as a teaching or prototyping tool. If you use it for paper experiments, please state clearly that Pyperplan does not offer state-of-the-art performance.
It was developed during the planning practical course at Albert-Ludwigs-Universität Freiburg during the winter term 2010/2011 and is published under the terms of the GNU General Public License 3 (GPLv3).
Pyperplan supports the following PDDL fragment: STRIPS without action costs.
Pyperplan requires Python >= 3.7.
From the Python package index (PyPI):
pip install pyperplan
From inside a repository clone:
pip install --editable .
This makes the pyperplan command available globally or in your virtual
environment (recommended).
The pyperplan executable accepts two arguments: a PDDL domain file and a
PDDL problem file. Example:
pyperplan benchmarks/tpp/domain.pddl benchmarks/tpp/task01.pddl
The domain file can be omitted, in which case the planner will attempt to guess its name based on the problem file. If a plan is found, it is stored alongside the problem file with a .soln extension.
By default, the planner performs a blind breadth-first search, which does not scale very well. Heuristic search algorithms are available. For example, to use greedy-best-first search with the FF heuristic, run
pyperplan -H hff -s gbf DOMAIN PROBLEM
For a list of available search algorithms and heuristics, run
pyperplan --help
For more information on using the planner and how to extend it to do more fancy stuff, see the documentation.
Pyperplan follows the semantics that all types other than the universal supertype object (which is mentioned as such in the PDDL 1.2 paper) need to be explicitly introduced.
Pyperplan is hosted on GitHub: https://github.com/aibasel/pyperplan
The original authors of Pyperplan are, in alphabetical order:
- Yusra Alkhazraji
- Matthias Frorath
- Markus Grützner
- Thomas Liebetraut
- Manuela Ortlieb
- Jendrik Seipp
- Tobias Springenberg
- Philip Stahl
- Jan Wülfing
The instructors of the course in which Pyperplan was created were Malte Helmert and Robert Mattmüller.
For questions and feedback, please start a new discussion.
Please cite Pyperplan using
@Misc{alkhazraji-et-al-zenodo2020,
author = "Yusra Alkhazraji and Matthias Frorath and Markus Gr{\"u}tzner
and Malte Helmert and Thomas Liebetraut and Robert Mattm{\"u}ller
and Manuela Ortlieb and Jendrik Seipp and Tobias Springenberg and
Philip Stahl and Jan W{\"u}lfing",
title = "Pyperplan",
publisher = "Zenodo",
year = "2020",
doi = "10.5281/zenodo.3700819",
url = "https://doi.org/10.5281/zenodo.3700819",
howpublished = "\url{https://doi.org/10.5281/zenodo.3700819}"
}