Skip to content
This repository has been archived by the owner on Mar 24, 2020. It is now read-only.

An OpenAI Gym environment with complex puzzle maps, highly adequate for reinforcement learning approaches.

Notifications You must be signed in to change notification settings

david-simoes-93/GeoFriends2

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

14 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

GeoFriends2

An OpenAI Gym environment with complex puzzle maps, highly adequate for reinforcement learning approaches, based on Geometry Friends. You can find a more recent version of GeoFriends2-v2.

To ready the environment, we recommend using VirtualEnv. You will need the PyGame and Gym environments.

python3 -m venv venv
source venv/bin/activate
pip install --upgrade pip
pip install pygame gym
python TestEnv.py

We also included some deep reinforcement learning examples, using Asynchronous Advantage Actor-Critic and Asynchronous 1-step and n-step Q-Learning. For those, you will need TensorFlow 1.1+, and SciPy. You can run both algorithms locally with

export PYTHONPATH=$(pwd)
python A3C/A3C-LocalThreads.py
python DQN/DQN-LocalThreads.py

Or distributed with the scripts (for example, 12 processes):

./start-a3c.sh 12
./start-dqn.sh 12

We have published a paper in IJCNN18 with more details and our results, titled "Guided Deep Reinforcement Learning in the GeoFriends2 Environment".

About

An OpenAI Gym environment with complex puzzle maps, highly adequate for reinforcement learning approaches.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published