docker-README.md

drake-ros Docker README

This is a README for using drake-ros with docker. It supports the following platforms -

• Ubuntu 22.04 both amd64 and arm64
• Apple Silicon Macs: Via BUILD_DRAKE_FROM_SOURCE The docker container would allow you to both visualize in rviz2 (on M1 Macs as well via noVNC!) and develop.

Prerequisites:

• Docker must be installed on your system. If you don't have Docker installed, follow the installation instructions at https://docs.docker.com/engine/install/.

• For NVIDIA GPU support, you need to have NVIDIA Container Toolkit installed. Follow the instructions at https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html to install it.

  • Rocker: rocker installation is required for NVIDIA GPU support. A simple way to install it is - pip3 install rocker Find out more about it here.
  • nvidia-docker2: Installation Instructions here

Instructions:

1. Clone the repo and go to the directory containing the Dockerfile.
2. Build the Docker image by running either -

   docker build --build-arg ARCH=arm64v8/ --build-arg BUILD_DRAKE_FROM_SOURCE=true -t drake-ros:arm64 .
   (On arm64 systems, support is only available via building the source code.)
   
   docker build --build-arg ARCH=amd64/ --build-arg BUILD_DRAKE_FROM_SOURCE=false -t drake-ros:x86_64 .
   (Recommended for x86 systems since it's significantly faster unless you want to build the latest drake)
   

NOTE: When building drake from source make sure you are alotting at least 8 gigs of RAM and limiting the number of cores (rule of thumb is 1 job per 8GB RAM) accordingly to the docker container as well. Otherwise, your builds will fail.

Machines with Nvidia GPUs -

1. Start a Docker container with the drake-ros image using rocker. The following command mounts the current working directory (the root of the drake-ros repository) inside the container at /drake_ros_ws/src/, enables NVIDIA GPU support, and sets up X11 forwarding:

   $ rocker --nvidia --x11 --volume "$(pwd):/drake_ros_ws/src/" -- drake-ros

If needed, replace drake-ros with the name of the Docker image you built earlier. Alternatively, you can also run an equivalent docker run command.

Machines without NVIDIA GPUs (Integrated CPU graphics, Apple Silicon etc.) -

1. Use the provided docker compose via - DRAKE_ROS_CONTAINER_NAME=my_custom_container_name docker compose up
2. In a new terminal, run the following to get a bash session running in the drake-ros container - docker exec -it drake_ros_integration /bin/bash
3. Go to http://localhost:8080/vnc.html to see any visualizations that you may run. Example - rviz2

Note - Since your drake-ros directory is mounted to the overlay ROS2 workspace in the container, any changes you make internally will be reflected on your host system. This is done to make development easier. You can read more about it here.