Toy Montezuma's Revenge

This is a reproduction of the Toy Montezuma's Revenge environment described in Deep Abstract Q-Networks (Roderick et al., 2017).

Rewards:

• +100 for picking up a key
• +300 for opening a door
• +10000 for reaching at the goal location

Usage

Demo (play by hand):

python -m mr_pycolab.toy_montezuma

# All rooms are fully observable rather than partially:
python -m mr_pycolab.toy_montezuma --full-observation

OpenAI gym

import mr_pycolab, gym
env = gym.make("ToyMontezumaRevenge-v0")

s = env.reset()   # [11, 11, 5]

actions = ('D', 'U', 'L', 'R', '?')
s, r, done, info = env.step(env.action_space.sample())