I expect a startup was the only option to execute on the idea with a decent level of autonomy. Google/Alphabet already have well established teams in the areas of mapping & autonomous driving so you can't just go ahead and start your own multi-engineer effort.