The Arcade Learning Environment (ALE) is a simple object-oriented framework that allows researchers and hobbyists to develop AI agents for Atari 2600 games. It is built on top of the Atari 2600 emulator Stella and separates the details of emulation from agent design. This video depicts over 50 games currently supported in the ALE.
For an overview of our goals for the ALE read The Arcade Learning Environment: An Evaluation Platform for General Agents. If you use ALE in your research, we ask that you please cite this paper in reference to the environment (BibTeX entry at the end of this document). Also, if you have any questions or comments about the ALE, please contact us through our mailing list.
Feedback and suggestions are welcome and may be addressed to any active member of the ALE team.
- Object-oriented framework with support to add agents and games.
- Emulation core uncoupled from rendering and sound generation modules for fast emulation with minimal library dependencies.
- Automatic extraction of game score and end-of-game signal for more than 50 Atari 2600 games.
- Multi-platform code (compiled and tested under OS X and several Linux distributions, with Cygwin support).
- Communication between agents and emulation core can be accomplished through pipes, allowing for cross-language development (sample Java code included).
- Python development is supported through ctypes.
- Agents programmed in C++ have access to all features in the ALE.
- Visualization tools.
Install main dependences:
sudo apt-get install libsdl1.2-dev libsdl-gfx1.2-dev libsdl-image1.2-dev cmake
Compilation:
$ mkdir build && cd build
$ cmake -DUSE_SDL=ON -DUSE_RLGLUE=OFF -DBUILD_EXAMPLES=ON ..
$ make -j 4
To install python module:
$ pip install .
or
$ pip install --user .
Getting the ALE to work on Visual Studio requires a bit of extra wrangling. You may wish to use IslandMan93's Visual Studio port of the ALE.
For more details and installation instructions, see the manual. To ask questions and discuss, please join the ALE-users group.
Releases before v.0.5 are available for download in our previous website. For the latest releases, please check our releases page.
Execute ./ale -help for more details; alternatively, see documentation available at http://www.arcadelearningenvironment.org.
-random_seed [n] -- sets the random seed; defaults to the current time
-game_controller [fifo|fifo_named] -- specifies how agents interact
with the ALE; see Java agent documentation for details
-config [file] -- specifies a configuration file, from which additional
parameters are read
-run_length_encoding [false|true] -- determine whether run-length encoding is
used to send data over pipes; irrelevant when an internal agent is
being used
-max_num_frames_per_episode [n] -- sets the maximum number of frames per
episode. Once this number is reached, a new episode will start. Currently
implemented for all agents when using pipes (fifo/fifo_named)
-max_num_frames [n] -- sets the maximum number of frames (independent of how
many episodes are played)
If you use the ALE in your research, we ask that you please cite the following.
M. G. Bellemare, Y. Naddaf, J. Veness and M. Bowling. The Arcade Learning Environment: An Evaluation Platform for General Agents, Journal of Artificial Intelligence Research, Volume 47, pages 253-279, 2013.
In BibTeX format:
@Article{bellemare13arcade,
author = {{Bellemare}, M.~G. and {Naddaf}, Y. and {Veness}, J. and {Bowling}, M.},
title = {The Arcade Learning Environment: An Evaluation Platform for General Agents},
journal = {Journal of Artificial Intelligence Research},
year = "2013",
month = "jun",
volume = "47",
pages = "253--279",
}