Repository files navigation

PyGame Learning Environment (PLE) is a learning environment, mimicking theArcade Learning Environment interface, allowing a quick start to Reinforcement Learning in Python. The goal of PLE is allow practitioners to focus design of models and experiments instead of environment design.

PLE hopes to eventually build an expansive library of games.

Accepting PRs for games.

Docs for the project can befound here. They are currently WIP.

Available games can be found in thedocs.

APLE instance requires a game exposing a set of control methods. To see the required methods look atple/games/base.py.

Here's an example of importing Pong from the games library within PLE:

fromple.games.pongimportPonggame=Pong()

Next we configure and initialize PLE:

frompleimportPLEp=PLE(game,fps=30,display_screen=True,force_fps=False)p.init()

The options above instruct PLE to display the game screen, withdisplay_screen, while allowing PyGame to select the appropriate delay timing between frames to ensure 30fps withforce_fps.

You are free to use any agent with the PLE. Below we create a fictional agent and grab the valid actions:

myAgent=MyAgent(p.getActionSet())

We can now have our agent, with the help of PLE, interact with the game over a certain number of frames:

nb_frames=1000reward=0.0forfinrange(nb_frames):ifp.game_over():#check if the game is overp.reset_game()obs=p.getScreenRGB()action=myAgent.pickAction(reward,obs)reward=p.act(action)

Just like that we have our agent interacting with our game environment.

PLE requires the following dependencies:

• numpy
• pygame
• pillow

Clone the repo and install with pip.

git clone https://github.com/ntasfi/PyGame-Learning-Environment.gitcd PyGame-Learning-Environment/pip install -e.

Set the following in your code before usage:

os.putenv('SDL_VIDEODRIVER','fbcon')os.environ["SDL_VIDEODRIVER"]="dummy"

Thanks to@wooridle.

cd into thePyGame-Learning-Environment directory and run the following:

git pull

• Documentation is currently in progress.
• Tests
• Parallel Learning (One agent, many game copies)
• Add games
• Generalize the library (eg. add Pyglet support)

If PLE has helped your research please cite it in your publications. Example BibTeX entry:

@misc{tasfi2016PLE,  author = {Tasfi, Norman},  title = {PyGame Learning Environment},  year = {2016},  publisher = {GitHub},  journal = {GitHub repository},  howpublished = {\url{https://github.com/ntasfi/PyGame-Learning-Environment}}}