This idea has been around a lot longer than you might think. I first heard it from a guy at Sperry 15 years ago. They were part of a project that borrowed Aegis Cruiser tech to make quasi autonomous cargo ships. The bridge was run with a form of artificial intelligence, that made logic based decisions while consulting nav, radar and engines. If it detected an apparent collision on radar, it would determine if there were any bottom hazards if it chose to change course. If there were, it would slow it’s engine speed until the vessel passed. When the ship reached it’s destination, it would take on a crew to take her to the dock
The owners’ vision was that a full crew would take the ship to sea, and then disembark at the sea buoy. A Captain and a Chief would remain on board, if taking manual control became necessary. Each one had touch screen monitors in their quarters (yes, even in the heads) that could bring up any information, and even use it to control the vessel. Their jobs were to respond to emergencies, or shut the vessel down and prepare for tow if necessary.
The guy said the idea was pitched to the IMO, who quickly shot it down. They felt that technology just hadn’t reach that level of reliability yet. For the past fifteen years, I’ve nothing about automated ships… until now. I really hope they put this back on the shelf where it belongs. It makes for interesting Popular Science reading, but I still don’t think we’re at the level of reliability needed to makes this happen.