meagmohit/gym-boxworld

Boxworld Game with Atari rendering in OpenAI Gym

gym-boxworld

A simple boxworld game (with atari rendering) in the gym OpenAI environment

Installation instructions

---

Requirements: gym with atari dependency

git clone https://github.com/meagmohit/gym-boxworld
cd gym-boxworld
sudo pip install -e . #python setup.py install

import gym
import gym_boxworld
env = gym.make('boxworld-v0')
env.render()

Environment Details

---

• boxworld-v0 : Default settings (map_size=6, total_keys=5)
• BoxWorldNoFrameskip-v3 : Default settings (map_size=6, total_keys=5) and tcp_tagging=True

Agent Details

---

• agents/random_agent.py random agent plays game with given random trajectories
• agents/play.py for humans to play the game. Keys w,s,a,d for UP, Down, Left and Right respectively.
• agents/agent_recordEEG_openLoop.py plays the game slow with TCP connection port to allow data stimulations

References

---