Skip to content

Latest commit

 

History

History
220 lines (168 loc) · 7.42 KB

README.md

File metadata and controls

220 lines (168 loc) · 7.42 KB

Shimmy

An API conversion tool for popular external reinforcement learning environments to Gymnasium and PettingZoo APIs.

Supported APIs

OpenAI Gym

  • Bindings to convert OpenAI Gym environments to Gymnasium Environments.

Atari Environments for OpenAI Gym

  • Bindings to ALE-py to provide Atari environments in Gymnasium.

DeepMind Control

  • Gymnasium bindings for single agent environments.
  • Pettingzoo bindings for multiagent soccer environments.

DMLab

  • Pettingzoo bindings for all environments.

OpenSpiel

  • Pettingzoo bindings for all environments.

Incoming Projects

The following are a list of existing environment suites that we are looking into bringing into Shimmy. We are actively looking for developers to contribute to this project, if you are interested in helping, please reach out to us.

At a glance

This is an example of using Shimmy to convert DM Control environments into a Gymnasium compatible environment:

import gymnasium as gym
from shimmy.registration import DM_CONTROL_SUITE_ENVS

env_ids = [f"dm_control/{'-'.join(item)}-v0" for item in DM_CONTROL_SUITE_ENVS]
print(env_ids)

env = gym.make(env_ids[0])
env_flatten = gym.wrappers.FlattenObservation(env)
print(env_ids[0])
print("===🌎", env.observation_space)
print("===🕹️", env.action_space)
print("---flattened 🌎", env_flatten.observation_space)
print("---flattened 🕹️", env_flatten.action_space)
['dm_control/acrobot-swingup-v0', 'dm_control/acrobot-swingup_sparse-v0', 'dm_control/ball_in_cup-catch-v0', 'dm_control/cartpole-balance-v0', 'dm_control/cartpole-balance_sparse-v0', 'dm_control/cartpole-swingup-v0', 'dm_control/cartpole-swingup_sparse-v0', 'dm_control/cartpole-two_poles-v0', 'dm_control/cartpole-three_poles-v0', 'dm_control/cheetah-run-v0', 'dm_control/dog-stand-v0', 'dm_control/dog-walk-v0', 'dm_control/dog-trot-v0', 'dm_control/dog-run-v0', 'dm_control/dog-fetch-v0', 'dm_control/finger-spin-v0', 'dm_control/finger-turn_easy-v0', 'dm_control/finger-turn_hard-v0', 'dm_control/fish-upright-v0', 'dm_control/fish-swim-v0', 'dm_control/hopper-stand-v0', 'dm_control/hopper-hop-v0', 'dm_control/humanoid-stand-v0', 'dm_control/humanoid-walk-v0', 'dm_control/humanoid-run-v0', 'dm_control/humanoid-run_pure_state-v0', 'dm_control/humanoid_CMU-stand-v0', 'dm_control/humanoid_CMU-run-v0', 'dm_control/lqr-lqr_2_1-v0', 'dm_control/lqr-lqr_6_2-v0', 'dm_control/manipulator-bring_ball-v0', 'dm_control/manipulator-bring_peg-v0', 'dm_control/manipulator-insert_ball-v0', 'dm_control/manipulator-insert_peg-v0', 'dm_control/pendulum-swingup-v0', 'dm_control/point_mass-easy-v0', 'dm_control/point_mass-hard-v0', 'dm_control/quadruped-walk-v0', 'dm_control/quadruped-run-v0', 'dm_control/quadruped-escape-v0', 'dm_control/quadruped-fetch-v0', 'dm_control/reacher-easy-v0', 'dm_control/reacher-hard-v0', 'dm_control/stacker-stack_2-v0', 'dm_control/stacker-stack_4-v0', 'dm_control/swimmer-swimmer6-v0', 'dm_control/swimmer-swimmer15-v0', 'dm_control/walker-stand-v0', 'dm_control/walker-walk-v0', 'dm_control/walker-run-v0']
dm_control/acrobot-swingup-v0
===🌎 Dict('orientations': Box(-inf, inf, (4,), float64), 'velocity': Box(-inf, inf, (2,), float64))
===🕹️ Box(-1.0, 1.0, (1,), float64)
---flattened 🌎 Box(-inf, inf, (6,), float64)
---flattened 🕹️ Box(-1.0, 1.0, (1,), float64)

For most usage, we recommend applying the gym.wrappers.FlattenObservation(env) wrapper to reduce the Dict observation space to a Box observation space.

Installation and Usage

To install Shimmy from PyPI:

pip install shimmy

Out of the box, Shimmy doesn't install any of the dependencies required for the environments it supports. To install them, you'll have to install the optional extras. All single agent environments have registration under the Gymnasium API, while all multiagent environments must be wrapped using the corresponding compatibility wrappers.

OpenAI Gym

Installation

pip install shimmy[gym]

Usage

import gymnasium as gym

env = gym.make("GymV22CompatibilityV0", env_name="...")

Atari Environments

Installation

pip install shimmy[atari]

Usage

import gymnasium as gym

env = gym.make("ALE/Pong-v5")

DM Control

Installation

pip install shimmy[dm-control]

Usage (Multi agent)

from dm_control.locomotion import soccer as dm_soccer
from shimmy.dm_control_multiagent_compatibility import (
    DmControlMultiAgentCompatibilityV0,
)

walker_type = dm_soccer.WalkerType.BOXHEAD,

env = dm_soccer.load(
    team_size=2,
    time_limit=10.0,
    disable_walker_contacts=False,
    enable_field_box=True,
    terminate_on_goal=False,
    walker_type=walker_type,
)

env = DmControlMultiAgentCompatibilityV0(env)

Usage (Single agent)

import gymnasium as gym

env = gym.make("dm_control/acrobot_swingup_sparse-v0")

DM Lab

Installation

Courtesy to Danijar Hafner for providing this install script.

#!/bin/sh
set -eu

# Dependencies
apt-get update && apt-get install -y \
    build-essential curl freeglut3 gettext git libffi-dev libglu1-mesa \
    libglu1-mesa-dev libjpeg-dev liblua5.1-0-dev libosmesa6-dev \
    libsdl2-dev lua5.1 pkg-config python-setuptools python3-dev \
    software-properties-common unzip zip zlib1g-dev g++
pip3 install numpy

# Bazel
apt-get install -y apt-transport-https curl gnupg
curl -fsSL https://bazel.build/bazel-release.pub.gpg | gpg --dearmor > bazel.gpg
mv bazel.gpg /etc/apt/trusted.gpg.d/
echo "deb [arch=amd64] https://storage.googleapis.com/bazel-apt stable jdk1.8" | tee /etc/apt/sources.list.d/bazel.list
apt-get update && apt-get install -y bazel

# Build
git clone https://github.com/deepmind/lab.git
cd lab
echo 'build --cxxopt=-std=c++17' > .bazelrc
bazel build -c opt //python/pip_package:build_pip_package
./bazel-bin/python/pip_package/build_pip_package /tmp/dmlab_pkg
pip3 install --force-reinstall /tmp/dmlab_pkg/deepmind_lab-*.whl
cd ..
rm -rf lab

Usage

import deepmind_lab

from shimmy.dm_lab_compatibility import DmLabCompatibilityV0

observations = ["RGBD"]
config = {"width": "640", "height": "480", "botCount": "2"}
renderer = "hardware"

env = deepmind_lab.Lab("lt_chasm", observations, config=config, renderer=renderer)
env = DmLabCompatibilityV0(env)

For Developers and Testing Only

pip install shimmy[testing]

To just install everything

pip install shimmy[all, testing]

OpenSpiel

Installation

pip install shimmy[pettingzoo]

Usage

import pyspiel
from shimmy.openspiel_compatibility import OpenspielCompatibilityV0

env = pyspiel.load_game("2048")
env = OpenspielCompatibilityV0(game=env, render_mode=None)

Citation

If you use this in your research, please cite:

@software{shimmy2022github,
  author = {{Jun Jet Tai, Mark Towers} and Elliot Tower and Jordan Terry},
  title = {Shimmy: Gymnasium and Pettingzoo Wrappers for Commonly Used Environments},
  url = {http://github.com/Farama-Foundation/Shimmy},
  version = {0.2.0},
  year = {2022},
}```