gym-super-mario-bros 0.9.0

Super Mario Bros. for OpenAI Gym

OpenAI Gym - Super Mario Bros

An OpenAI Gym environment for the original Super Mario Bros. game on the Nintendo Entertainment System (NES).

Installation

The preferred installation of gym-super-mario-bros is from pip:

pip install gym-super-mario-bros

NES Emulator

NESGym uses FCEUX to emulate NES games. Make sure it's installed and in your $PATH.

Unix

sudo apt-get install fceux

Mac

brew install fceux

Usage

import gym_super_mario_bros
env = gym_super_mario_bros.make('SuperMarioBros-v0')

done = True
for step in range(5000):
    if done:
        state = env.reset()
    state, reward, done, info = env.step(env.action_space.sample())

env.close()

NOTE: gym_super_mario_bros.make is just an alias to gym.make for convenience.

Environments

The following environments play the game as a human would. The agent has three lives to make it through the 32 levels of the game. The agent is configured to only see reward-able game-play frames. No cut-scenes, loading screens, etc. are shown to the agent.

Environment	Description
SuperMarioBros-v0	4 frames per action, standard ROM
SuperMarioBros-v1	4 frames per action, custom down-sampled ROM
SuperMarioBrosNoFrameskip-v0	1 frame per action, standard ROM
SuperMarioBrosNoFrameskip-v1	1 frame per action, custom down-sampled ROM

Citation

Please cite gym-super-mario-bros if you use it in your research.

@misc{gym-super-mario-bros,
  author = {Christian Kauten},
  title = {{S}uper {M}ario {B}ros for {O}pen{AI} {G}ym},
  year = {2018},
  publisher = {GitHub},
  howpublished = {\url{https://github.com/Kautenja/gym-super-mario-bros}},
}