> The final product probably doesn't use it. I disagree here, usually projects that are not keen on relying on ROS end up re-implementing a lot of the groundwo

> The final product probably doesn't use it. I disagree here, usually projects that are not keen on relying on ROS end up re-implementing a lot of the groundwork (shared memory, message distribution, parameter server, logging, etc.), usually they are older products (10 or 15+ years old), where ROS wasn't really a thing and migrating to ROS after implementing their own tailored stack is not worth it.