gym-hexario's Introduction

gym-hexario

OpenAI gym multiagent environment for the game Hexar.io.
By now it only works on linux (and macOS eventually) because it requires the native library SDL2.

installing SDL2

Make sure SDL2 is present on your system. It can be installed with all common package managers, e.g.

sudo pacman -Syu sdl2               # arch based
sudo apt-get install libsdl2-2.0    # debian etc
brew install sdl2                   # macOS

installing gym-hexario

git clone https://github.com/dhansmair/gym-hexario.git
cd gym-hexario
pip3 install .

code example

import gym
import gym_hexario

# create the environment
n_agents = 2
env = gym.make('Hexario-v0', n_agents=n_agents)

# run 100 episodes with random actions
for _ in range(100):

    env.reset('all')
    dones = [False] * n_agents

    while False in dones:
        env.render()

        actions = [env.action_space.sample() for _ in range(n_agents)]
        states, rewards, dones, infos = env.step(actions)

    print('finished episode')

configuration

The environment can be configured through various optional arguments for the gym.make(...) function.

kwargs={
    'n_agents': 1,
    'map_radius': 20,                   # size of the board
    'max_steps': 500,
    'perspective': 'local',             # 'local' | 'global'
    'observation_space': 'features',    # 'features' | 'pixels'
    'observation_radius': 8,
    'action_space': 'discrete',         # 'discrete' | 'continuous'
    'reward_timestep': 0.0,
    'reward_capture': 1.0,
    'reward_claim': 0.0,
    'reward_kill': 100.0
}

env = gym.make('Hexario-v0', **kwargs)

# ...

Recommend Projects