VisionServer  v2.1.1-1-g21dc5465
FRC vision library
Robot Project Integration

Although it is not strictly necessary to develop vision code alongside robot code, combining the two into a single project can be helpful for cross-referencing during development, and compartmentalizing code for similar tasks. Additionally, this makes it easy to publish all season code, both for a robot and for vision, to a single repository.

This tutorial will traverse through the process of integrating the VisionServer library into a robot project.

Filestructure

  1. Create/have a WPILib robot project. More info can be found here.
  2. Setup the vision program directories.
    • In the base of your project, create a folder that will contain all vision code for the raspberry pi (ex. vision/).
    • Now navigate inside this folder, and create 3 folders: one for all source files, one for intermediate files, and one for output files (ex. src/, obj/, and bin/).
    • The filestructure should now resemble something like this:
      > Robot-Project/
      > .gradle/
      > .vscode/
      > .wpilib/
      > gradle/
      > src/
      > vision/
      > bin/
      > obj/
      > src/
      > other build files...

Project Files

  1. Clone VisionServer. While still inside the base vision directory, open a terminal and enter git clone https://github.com/FRC3407/VisionServer --recurse-submodules to clone this repo and its submodules.
    • If your base project is already a git repo, and you want to add VisionServer as a submodule (definitely recommended for this config), then navigate to the base project folder and git submodule add https://github.com/FRC3407/VisionServer VISION_PROGRAM_RELATIVE_PATH/VisionServer - where for this tutorial VISION_PROGRAM_RELATIVE_PATH would be vision (as created in step 1).
    • The filestructure should now resemble something like this:
      > Robot-Project/
      > ...
      > vision/
      > bin/
      > obj/
      > src/
      > VisionServer/
      > other build files...
  2. Create source files (begin development). Within the src/ subdirectory, create your source files. This step is intentionally vague because the this is not a programming guide.
    • See the link at the end for an example project.
  3. Setup the build system. This could involve make, cmake, or even gradle. I personally use a makefile similar to that used for building VS itself.
    • For linking to VisionServer, the shared and static library binaries can be found in ...VisionServer/lib-vs/out/. Headers are also contained in the include/ subdirectory, although it is more convenient to just include straight from the source folder instead (...VisionServer/lib-vs/src/).
    • BUILD.md contains more info about linking to and building the library itself.
    • All libraries that VS depends on can also be conveniently utilized by a user-project (ex. OpenCV), and are located under ...VisionServer/lib-vs/lib/ and ...VisionServer/lib-vs/indlude/.
    • See the link at the end for an example project.
  4. Add the robot-side API files into your robot project. Within the base project directory, navigate to where your source files are located (ex. src/main/java/frc/robot/ for Java), and git clone https://github.com/S1ink/VisionServer-Robot API_DIR - where API_DIR represents the directory name where the repo will be cloned (ex. vision is once again a good name).
    • Again, if submodules are preferred, then this should be added as a submodule from the base project directory: git submodule add https://github.com/S1ink/VisionServer-Robot SRC_DIR_PATH/API_DIR.
    • Note that for Java projects the "package" line at the top of the added files will need to be updated based on the directory names used above.

Additional Automations (Optional)

  1. Configure VSCode workspace and tasks. This is the same process as described in step 4 in SETUP.md, except that include paths may need to be adjusted based on the build system present. A build task can be constructed to invoke whatever build system is used with the correct parameters, and allows for a keybound build command.
  2. Rebuild VisionServer before each project build. If you plan on editing the library itself during development, you can simply create a build script that calls make -C VISION_DIR/VisionServer/lib-vs shared OPT=release or make -C VISION_DIR/VisionServer/lib-vs static OPT=release before the vision program itself is built. This will update ...VisionServer/lib-vs/out/libvs3407.so [and .a] along with recopy all headers into ...VisionServer/lib-vs/out/include/, which can then be used as artifacts for the program build.

Example Project

An example project used during the Rapid React FRC season can be found here.