Autonomous driving is encountering many problems to become popular in the case of cars, because the situations that can arise on the roads are the most varied and unpredictable. But this is not the case in the case of other types of vehicles such as boats.
The boats always move on the surface of the sea, and in the sea there is very little traffic and there are no "pedestrians", because the bathers are close to the shore where the boats do not sail.
On the other hand, a ship needs a lot of time and space to change course, and the more so the larger the ship, which is why its control is very delicate and Artificial Intelligence is a good help.
I have seen a control system called Orca AI that has performed autonomous navigation tests with a ship of more than 700 tons, and has traveled almost 500 miles through an area with intense navigation (from Tokyo Bay to Ise Bay) without requiring human intervention.
The system has 18 cameras that provide a 360º view to analyze the trajectories of all surrounding ships, and it has infrared capability for fog and low visibility. And also the support of Artificial Intelligence to learn with time by itself.
A video of testing in the Panama Canal, and other places with high fog and low visibility for humans, but not for infrared cameras, can be viewed at: https://youtu.be/M1BezS_2Jbs
There are several companies that are working on this issue, and it is more than likely that in a short time they will begin to see applications on ships, as they are systems that quickly pay for themselves considering how much it costs to have specialized people on ships.
It is not about leaving a ship to sail alone, but about leaving it with less crew, and therefore at a lower cost.
And in the case of pleasure boats, almost half a million are manufactured every year in the world, and they can be a good market niche for simpler autonomous navigation systems that allow their "skippers" to navigate from one place to another with ease and total comfort while having a beer.
EN 
ES_ES 