olam 1.0.2

Online Learning of Action Models (OLAM).

Online Learning of Action Models for PDDL Planning

This repository contains the refactored code of the Online Learning of Action Models (OLAM) algorithm presented at IJCAI 2021, for details about the method please see the paper. The previous code used for the paper experiments is available at this link.

Example Usage

import unified_planning
from unified_planning.io import PDDLReader
from unified_planning.shortcuts import SequentialSimulator
from amlgym.util.util import empty_domain
from olam.OLAM import OLAM

# Disable printing of planning engine credits
unified_planning.shortcuts.get_environment().credits_stream = None

domain_ref_path = "olam/benchmarks/domains/blocksworld.pddl"
problem_path = "olam/benchmarks/problems/blocksworld/1_p00_blocksworld_gen.pddl"
empty_domain_path = empty_domain(domain_ref_path) # remove preconditions/effects
olam = OLAM(empty_domain_path)

sim_problem = PDDLReader().parse_problem(domain_ref_path,
                                         problem_path)
simulator = SequentialSimulator(sim_problem)
learned_domain_str, trajectory = olam.run(simulator, max_steps=100)

print(f"Generated a trajectory with {len(trajectory.observations)} states")
print(f"Domain learned: {learned_domain_str}")

Installation for developers

Clone this repository and install in developer mode:

pip install -e .

Citations

@inproceedings{ijcai2021-566,
  title     = {Online Learning of Action Models for PDDL Planning},
  author    = {Lamanna, Leonardo and Saetti, Alessandro and Serafini, Luciano and Gerevini, Alfonso and Traverso, Paolo},
  booktitle = {Proceedings of the Thirtieth International Joint Conference on
               Artificial Intelligence, {IJCAI-21}},
  publisher = {International Joint Conferences on Artificial Intelligence},
  pages     = {4112--4118},
  year      = {2021},
  doi       = {10.24963/ijcai.2021/566},
}

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgements

This code has been refactored with the help of Ejdis Gjinika