/gym-wobble

A simple 1-D cursor-target catch game (with atari rendering) in the gym OpenAI environment

Primary LanguagePython

gym-catch

A simple 1-D cursor-target catch game (with atari rendering) in the gym OpenAI environment

Installation instructions

Requirements: gym with atari dependency

git clone https://github.com/meagmohit/gym-wobble
cd gym-wobble
sudo pip install -e . #python setup.py install
import gym
import gym_wobble
env = gym.make('wobble-v0') # The other option is 'WobbleNoFrameskip-v4'
env.render()

Environment Details

  • wobble-v0 : Default settings (max_timesteps=100, max_dist=3, tcp_tagging=False, tcp_port=15361)
  • WobbleNoFrameskip-v4 : Default settings and max_timesteps=100
  • WobbleNoFrameskip-v3 : Default settings and max_timesteps=100 and tcp_tagging=True

Agent Details

  • agents/random_agent.py random agent plays game with given error probability to take actions (Perr)

References