README.md

mujoco_sim_docker

This repository contains docker images for deploying ROS packages for MuJoCo with https://github.com/K4R-IAI/K4R_Development

0. Prerequisites: https://docs.docker.com/compose/install

No ROS required, No Unreal Engine required, just a PC with a good NVIDIA Graphic Card :)

Check your docker-compose and docker-py versions:

$ docker-compose -v
docker-compose version 1.29.2, build 5becea4c
$ pip3 freeze | grep docker
docker==5.0.0

1. If you want to use a package simulator from Unreal Engine, download it at https://seafile.zfn.uni-bremen.de/f/a8ac0f55af3c4620bc56/ and unzip it

2. Clone this repository and go into the checked out folder:

cd mujoco_sim_docker

3. Follow these steps to build docker images:

3.1. Clone the packages with wstool

sudo apt install ros-${ROS_DISTRO}-mk # This step is to make sure the mk from rospack is there
./setup_ros_packages.bash

3.2. Install nvidia-docker2 to enable GPU access with Compose (sudo required)

./setup_rviz.bash

3.3. Build the docker images

source .env # This step is to make sure the ROS_DISTRO is noetic
docker-compose build

4. Using the docker images

4.1. Start ros-core and mujoco-ws

docker-compose up mujoco-ws

4.2. Start the simulation (either by clicking Play in Unreal Engine or from the unziped package)

sh Waterfront.sh

4.3. You can visualize everything with rviz from docker container

./run_rviz.bash

The result should look like this:

5. Stop docker containers

5.1. Cancel every terminals with Ctrl + C

5.2. Remove docker containers and network

docker-compose down