/hiro-robot-envs

A collection of robotic environments (which inherit from gym.Env) for reinforcement learning algorithms.

Primary LanguagePythonMIT LicenseMIT

hiro-robot-envs

A collection of robotic environments (which inherit from gym.Env) for reinforcement learning algorithms.

This repository is a Python package that allows you to use the environments within this repo.

Quick Start

  • Note - requires at least Python 3.5 or above.

Before installing this package, you will need to install Mujoco, which requires some extra steps. You can check them out here.

git clone https://github.com/HIRO-group/hiro-robot-envs
cd hiro-robot-envs
pip3 install .

Current Supported Environments

  • AntEnvWithGoal - creates the Mujoco ant environment, conditioned on a goal.

Example

After installing the package, here's an example of using the ant environment.

from hiro_robot_envs.envs import AntEnvWithGoal
from hiro_robot_envs.envs import create_maze_env

env = AntEnvWithGoal(create_maze_env('AntMaze'), 'AntMaze')

# do things you would normally do with an env below: