Docker ROS NVIDIA

Overview

Dockerfiles to build images that have ROS (1 or 2) with NVIDIA support and with GUI support (e.g. Gazebo and RViz).

The build.bash and the run.bash files are used to automatically build and run the image.

This repository is intended to be used as a base for other projects that require ROS, NVIDIA, and GUI support in Docker. You can clone it, remove the .git file to start a new repository, and modify the build.bash, run.bash, and Dockerfile files to suit your needs.

If you find this useful, you can cite this repository and/or me in your work.

Preliminaries

Install Docker Community Edition (ex Docker Engine). You can follow the installation method through apt. Note that it makes you verify the installation by running sudo docker run hello-world. It is better to avoid running this command with sudo and instead follow the post installation steps first and then run the command without sudo.

Follow with the post-installation steps for Linux. This will allow you to run Docker without sudo.

Install NVIDIA Container Toolkit (nvidia-docker2).

Usage

The docker base image and the ROS version can be changed by modifying the BASE_IMAGE, BASE_TAG, and ROS_NUMBER bash variables in build.bash. To change the image name, modify IMAGE_NAME both in build.bash and run.bash.

Build the docker image (use the -r option to update the underlying images):

./docker/build.bash [-r]

Run the container:

./docker/run.bash

The workspace directory should be the folder containing run.bash and build.bash. It is mounted in the Docker container on startup.

Build the workspace inside the Docker container with colcon or catkin to avoid permission problems. The workspace's setup.bash is automatically sourced when the container is opened; thus it will fail the first time the container is run.

Take a look at https://docs.docker.com/develop/develop-images/dockerfile_best-practices/ before modifying the Dockerfile according to your needs.

To use VS Code with Docker, you can use the Dev Containers extension to attach VS Code to a running container. For having autocomplete, linting, etc. take a look at https://github.com/athackst/vscode_ros2_workspace and in particular to c_cpp_properties.json and settings.json in .vscode.

Troubleshooting

• No GPU available in Docker: running nvidia-smi in the Docker container returns Failed to initialize NVML: Unknown Error.
  Solution:
  • Run sudo nano /etc/nvidia-container-runtime/config.toml, set no-cgroups = false, and save (Ctrl + X and then Y).
  • Restart Docker with sudo systemctl restart docker.

Author

Davide De Benedittis

Acknowledgments

• Baptiste Busch: creation of a ROS workspace in Docker
• Official NVIDIA CUDA Docker Images: NVIDIA Drivers in Docker. Uses the 12.x.x-base-ubuntuxx.xx version. Change it to your liking.