All of our code is stored in robosub-ros, the public Github repository linked above. Our software stack is built on top of Robot Operating System (ROS), and we do all development in Python and C++. Our code is organized into two main packages: /onboard, which contains all code that runs directly on the robot during competition (e.g. computer vision, task planning, controls), and /landside, which contains code that doesn’t need to run on the robot (e.g. simulation). Each package is thoroughly documented with a README which contains detailed descriptions and setup instructions. Feel free to explore the codebase, and maybe even try cloning and running it locally!
Detecto is our custom open-source computer vision library which has received over 30,000 downloads in Github. Developed on top of a Faster R-CNN Resnet 50 architecture, Detecto allows our robot to locate and classify any object in its field of view.
Are you curious to learn more about how our robot works? If so, you have come to the right place. Below are comprehensive general overviews of each subsytem aboard Cthulhu v2 (Mechanical, Electrical, and Software). Enjoy!