Where else would you test a submarine? Space? The desert perhaps?

Robotic subs are a weird concept. Submarines typically maintain their stealth by cutting off their communications for days or weeks. How does a robotic submarine make decisions without a constant stream of communications? And if the submarine is meant to deliver a payload and then return home, isn't that just a modified torpedo, then? Like how a drone is to a missile. If they maintain communications for days or weeks, then they are liable to be hacked. It'll be interesting to see how this plays out.