Skip to content

README.md

Latest commit

 

History

History
56 lines (44 loc) · 2.07 KB

README.md

File metadata and controls

56 lines (44 loc) · 2.07 KB

Retro-Learning-Environment

A learning framework based on the Arcade Learning Environment (ALE) and LibRetro (Stella for Atari and SNES9X for the Super Nintendo Entertainment System).

The environment provides an inreface to training and evaluating AI algorithms against different console games using its screen as input.

The currently supported games can be found in the src/games/supported directory . Some popular games include: Mortal Kombat, Super Mario All Stars, F-Zero, Castle Wolfenstein and Gradius III.

A paper is available for RLE at http://arxiv.org/abs/1611.02205. If you use RLE in your publication, please use the following BibTex entry:

@article{bhonker2016playing,
  title = {Playing SNES in the Retro Learning Environment},
  author = {Bhonker, Nadav and Rozenberg, Shai and Hubara, Itay},
  journal = {arXiv preprint arXiv:1611.02205},
  year = {2016}
}

=============================== Quick Start

Install main dependences:

sudo apt-get install libsdl1.2-dev libsdl-gfx1.2-dev libsdl-image1.2-dev cmake

To use the shared_library interface:

$ mkdir build && cd build
$ cmake -DUSE_SDL=ON -DBUILD_EXAMPLES=ON ..
$ make -j 4

To install the python interface:

$ pip install .
or
$ pip install --user .

To install the lua (Torch) interface, the additional alewrap module is required:

luarocks install https://raw.githubusercontent.com/nadavbh12/Retro-Learning-Environment/master/ale-2-0.rockspec
luarocks install https://raw.githubusercontent.com/nadavbh12/alewrap/master/alewrap-0-0.rockspec

=============================== DQN Implementations Using RLE

Acknowledgements

  • @mgbellemare for his work on ALE and his useful advice.
  • @Alcaro and the @libretro community for their assistance in incorporating their work into our framework.