Developing with Docker
This will differ on which operating system you have installed, this guide is for linux-based systems. Please take a look at the official Docker Get Docker guide. There is also a guide from ROS called getting started with ROS and Docker. On Ubuntu one should be able to do the following to get docker:
curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo gpg --dearmor -o /usr/share/keyrings/docker-archive-keyring.gpg echo "deb [arch=amd64 signed-by=/usr/share/keyrings/docker-archive-keyring.gpg] https://download.docker.com/linux/ubuntu $(lsb_release -cs) stable" | sudo tee /etc/apt/sources.list.d/docker.list > /dev/null sudo apt-get update sudo apt-get install docker-ce docker-ce-cli containerd.io
distribution=$(. /etc/os-release;echo $ID$VERSION_ID) \ && curl -s -L https://nvidia.github.io/nvidia-docker/gpgkey | sudo apt-key add - \ && curl -s -L https://nvidia.github.io/nvidia-docker/$distribution/nvidia-docker.list | sudo tee /etc/apt/sources.list.d/nvidia-docker.list sudo apt-get update sudo apt-get install -y nvidia-docker2 sudo systemctl restart docker sudo docker run --rm --gpus all nvidia/cuda:11.0-base nvidia-smi #to verify install
From this point we should be able to "test" that everything is working ok. First on the host machine we need to allow for x11 windows to connect.
We can now run the following command which should open gazebo GUI on your main desktop window.
docker run -it --net=host --gpus all \ --env="NVIDIA_DRIVER_CAPABILITIES=all" \ --env="DISPLAY" \ --env="QT_X11_NO_MITSHM=1" \ --volume="/tmp/.X11-unix:/tmp/.X11-unix:rw" \ osrf/ros:noetic-desktop-full \ bash -it -c "roslaunch gazebo_ros empty_world.launch"
Alternatively we can launch directly into a bash shell and run commands from in there. This basically gives you a terminal in the docker container.
docker run -it --net=host --gpus all \ --env="NVIDIA_DRIVER_CAPABILITIES=all" \ --env="DISPLAY" \ --env="QT_X11_NO_MITSHM=1" \ --volume="/tmp/.X11-unix:/tmp/.X11-unix:rw" \ osrf/ros:noetic-desktop-full \ bash # you should be able to launch rviz once in bash rviz
Clone the OpenVINS repository, build the container and then launch it. The Dockerfile will not build the repo by default, thus you will need to build the project.
mkdir -p ~/workspace/catkin_ws_ov/src cd ~/workspace/catkin_ws_ov/src git clone https://github.com/rpng/open_vins.git cd open_vins docker build -t openvins . cd ~/workspace/catkin_ws_ov
If the dockerfile breaks, you can remove the image and reinstall using the above
docker image list docker image rm openvins --force
From here it is a good idea to create a nice helper command which will launch the docker and also pass the GUI to your host machine. Here you can append it to the bottom of the ~/.bashrc so that we always have it on startup or just run the two commands on each restart
nano ~/.bashrc # add to the bashrc file xhost + &> /dev/null export DOCKER_CATKINWS=/home/username/workspace/catkin_ws_ov export DOCKER_DATASETS=/home/username/datasets alias ov_docker="docker run -it --net=host --gpus all \ --env=\"NVIDIA_DRIVER_CAPABILITIES=all\" --env=\"DISPLAY\" \ --env=\"QT_X11_NO_MITSHM=1\" --volume=\"/tmp/.X11-unix:/tmp/.X11-unix:rw\" \ --mount type=bind,source=$DOCKER_CATKINWS,target=/catkin_ws \ --mount type=bind,source=$DOCKER_DATASETS,target=/datasets \ openvins $1" # save and exit source ~/.bashrc
Now we can launch RVIZ and also compile the OpenVINS codebase. From two different terminals on the host machine one can run the following:
ov_docker roscore ov_docker rosrun rviz rviz -d /catkin_ws/src/open_vins/ov_msckf/launch/display.rviz ov_docker bash
Now once inside the docker with the bash shell we can build and launch an example simulation:
cd catkin_ws catkin build source devel/setup.bash roslaunch ov_msckf pgeneva_sim.launch
Jetbrains provides some instructions on their side and a youtube video. Basically, Clion needs to be configured to use an external compile service and this service needs to be exposed from the docker container. I still recommend users compile with
catkin build directly in the docker, but this will allow for debugging and syntax insights.
/ blog.jetbrains.com/ clion/ 2020/ 01/ using-docker-with-clion/
/ www.youtube.com/ watch?v=h69XLiMtCT8
After building the OpenVINS image (as above) we can do the following which will start a detached process in the docker. This process will allow us to connect Clion to it.
docker run -d --cap-add sys_ptrace -p127.0.0.1:2222:22 --name clion_remote_env openvins
We can now change Clion to use the docker remote:
- In short, you should add a new Toolchain entry in settings under Build, Execution, Deployment as a Remote Host type.
- Click in the Credentials section and fill out the SSH credentials we set-up in the Dockerfile
- Host: localhost
- Port: 2222
- Username: user
- Password: password
- Add a CMake profile that uses this toolchain and you’re done.
- Change build target to be this new CMake profile (optionally just edit / delete the default)
To add support for ROS you will need to manually set environmental variables in the CMake profile. These were generated by going into the ROS workspace, building a package, and then looking at
printenv output. It should be under
Settings > Build,Execution,Deployment > CMake > (your profile) > Environment. This might be a brittle method, but not sure what else to do... (also see this blog post).
When you build in Clion you should see in
docker stats that the
clion_remote_env is building the files and maxing out the CPU during this process. Clion should send the source files to the remote server and then on build should build and run it remotely within the docker container. A user might also want to edit
Build,Execution,Deployment > Deployment settings to exclude certain folders from copying over. See this jetbrains documentation page for more details.