This repository builds and uploads the Docker container for CUAD's ROS Setup
In a terminal, run docker run -it --rm -p 10000 ghcr.io/cuautodrone/cuad-ros:latest.
If you want to make it such that the container is not deleted once you exit it, run docker run -it -p 10000 ghcr.io/cuautodrone/cuad-ros:latest instead.
Then, in a second terminal, run docker ps to find which port on your computer port 10000 in the container has been bound to.
To run GUI apps, go to localhost:port in a browser, where port is the port number.
If you need to run multiple GUI apps at once, and don't want to open a second terminal inside the container, you can run the first command with a & character at the end, which will cause it to run in the background.
To open a second shell within the container, get the container id (this can be found by running docker ps) then run docker exec -it <container_id> /bin/sh where <container_id> is the container id found previously.
Additionally, sim_vehicle.py can be found in /ardupilot/Tools/autotest/.
Note: The timezone in the container is hardcoded to America/New_York. If you want to change the timezone, run sudo dpkg-reconfigure tzdata in the container.
First, run mkfifo to create a named pipe in a directory you plan on sharing with the container, using mkfifo /path/to/named/pipe where /path/to/named/pipe is the path to the named pipe you want to create (This probably won't work on Windows, I'm not sure how to created named pipes on Windows). Next, pass your camera to the named pipe using a command such as ffmpeg -re -y -f avfoundation -framerate 30 -i "FaceTime HD Camera" -preset ultrafast -vcodec libx264 -tune zerolatency -vf mpdecimate,setpts=N/FRAME_RATE/TB -f h264 -flush_packets 0 - | pv -B 10000000 > /path/to/named/pipe. You may have to replace the -f avfoundation and -i "FaceTime HD Camera" parts of the command with different values, which you can find by looking up "ffmpeg get input from camera" followed by your operating system (the above example is for a camera called "FaceTime HD Camera", which is the default name of the built-in camera on Mac Laptops, on macOS). Additionally, you may need to tweak the value passed to the -B parameter of pv. This value affects the size of the camera buffer (in bytes). Larger values may lead to more lag, but if the value is too small, the buffer can't even hold a full frame. The value above should work. Then, in your code, you should replace the 0 passed to cv2.VideoCapture with the path to your named pipe within the container, which should be enclosed in single quotes (for example, cv2.VideoCapture('/home/user/HandGestureTest/camera')). Additionally, when running the Docker Container, in order to share a local directory with the container, you should pass the -v /path/on/host:/path/in/container flag to docker run, where /path/on/host is the path on your computer that you want to share and /path/in/container is the path that you want the shared directory to be at in the container.
To fix this, you will have to authenticate Docker to the GitHub Container Registry. To do this, follow the instructions at this link. However, I would recommend providing the PAT directly from your clipboard if possible so that it doesn't end up in your shell history. If you do this, do not run the command that has export in it. On macOS, one can paste from their clipboard by replacing echo $CR_PAT with pbpaste.
In the project directory, run docker buildx bake --build-arg default.args.ROS_APT_SOURCE_VERSION=$(curl -s https://api.github.com/repos/ros-infrastructure/ros-apt-source/releases/latest | grep -F "tag_name" | awk -F\" '{print $4}').
Then, if you want to clear the build cache afterwards, run docker buildx prune --all.
Any new changes should first be tested by opening a Pull Request. After opening a Pull Request, the ci-pr workflow will run, which will prepare the necessary files and build the necessary wxpython wheel. Once the ci-pr workflow finishes, and the Pull Request is reviewed, manually dispatch the container's build for the test tag. This must be done within 30 days of when ci-pr is run, or else the necessary build info from ci-pr will no longer be available for ci-pr-dispatched to download. To do this, first go to the ci-pr workflow run for the same Pull Request and obtain the Run ID. This can be obtained by looking at the URL of the run, which should look something like https://github.com/CUAutoDrone/cuad-docker-ros/actions/runs/<run-id>, where <run-id> is the Run ID. Then, go to https://github.com/CUAutoDrone/cuad-docker-ros/actions/workflows/docker-ci-pr-dispatch.yml and click "Run workflow". Under "Use workflow from", select the Pull Request's branch, and under "Workflow Run ID from meta bake definition generation", provide the Run ID you obtained above. If you have already run ci-pr-dispatched on one Pull Request, please do not run it on another Pull Request until after the first Pull Request is merged as it will overwrite the image built for first pull request. Once the CI build is finished and you are ready to merge your changes, please go to https://github.com/CUAutoDrone/cuad-docker-ros/actions/workflows/docker-ci-merge.yml and select "Run workflow", setting "Use workflow from" to the Pull Request's branch. This will copy the image with the test tag to the latest tag, thus promoting the changes to the stable image, before automatically merging the PR. Whenever a commit is pushed directly to the main branch, an image is built with the main tag.
.dockerbuild files can be found in the build summaries from the ci-pr-dispatch action.