/Meta-Reinforcement-Learning-for-Heuristic-Planning

Code for Meta Reinforcement Learning for Heuristic Planning paper

Primary LanguagePython

Meta-Reinforcement-Learning-for-Heuristic-Planning

This repository contains code used in the paper Meta Reinforcement Learning for Heuristic Planning. For the task selection method used in the paper please refer to: ITTS. The envs folder contains the scripts used to translate PDDL files to Open AI gym domains.

Dependencies

  • python 3.*
  • numpy
  • tensorflow v1.7+
  • garage
  • gym
  • scipy
  • pddlpy

Data

For PDDL files used in the experiments please take a look at: http://planning.domains/

Usage

For training the models check rl2.py For testing use model_test.py

Acknowledgments

Part of the code was influenced by :

https://github.com/awjuliani/Meta-RL