<aside> ⏰ Estimated time to complete: 30 min

</aside>

Minimal install vs Full install

There are two ways of installing RCareWorld.

Full Install

The codebase should be organized according to the following structure

rcareworld_workspace
- rcare_py
- rcare_unity

You can use the following commands (on Linux systems) to create the parent rcareworld_workspace directory.

mkdir rcareworld_workspace
cd rcareworld_workspace

Install Python module

Follow the following steps to install the python module. If you would like to perform reinforcement learning, you need to also install rl-baselines3-zoo in rcare_py folder.

<aside> 💡 Also available on pypi and docker

</aside>

We recommend using conda for environment management. Check this for installing conda. WSL Ubuntu users can use the wget link to install from the command line.

# Create and activate conda environment
conda create -n rcare python=3.8
conda activate rcare

# OPTION1: clone Python repo (with https)
git clone <https://github.com/empriselab/RCareWorld.git> rcare_py

# OPTION2: clone Python repo (with ssh)
git clone [email protected]:empriselab/RCareWorld.git rcare_py

cd rcare_py
cd pyrcareworld

pip install -r requirements.txt
pip install -e .
cd ..

<aside> 💡 KNOWN ISSUE

Running pip install -e . might fail with

gym error message.png

This is caused by compatibility issues from different versions of gym/wheel. A fix would be to pip install a different version of setup tools:

pip install setuptools==65.5.0 "wheel<0.40.0”

</aside>

(Optional) Install rl-baseline3-zoo-1.1.0

Download the source code for rl-baseline3-zoo-1.1.0 from GitHub.

wget <https://github.com/DLR-RM/rl-baselines3-zoo/archive/refs/tags/v1.1.0.zip>
unzip v1.1.0.zip
cd rl-baselines3-zoo/
pip install -e .