I obviously don’t know the detail of how these system will work, but as far as I understand the plan is not to have any manual controls per se (I.e. no Wheel house or Engine Control room)
The operators initially manning the Yara Birkeland will be housed in temporary container(s) and presumably take control through portable means. Whether that will be a laptop computer and internal WiFi, or some sort of “control box” connected by wires I don’t know.
In worst case they should at least be able to shut down all systems by hitting an ESD (Jesus button?) and call for help.
In autonomous mode the worst case scenario is where some unfriendly person/pirates gets on board and try to take “manual” control. This can be eliminated by not having any such possibilities (No manual controls)
Somebody hacking into the onboard computer, disrupting or diverting control from the Operating Centre is the other main concern. I 'm not sure how that will be solved, but certainly it is one of the main thing those who develop the systems are looking into VERY seriously.
As far as I know the bean counters are not in control of the development, which is largely funded by the EU and individual Governments, mainly in Europe, but also in China, Japan and Singapore to mention those most involved. (Google, I assume, are paid for their services and are not controlling any purse strings).
The main commercial player here in Europe is Rolls-Royce, Siemens and DNV-GL all with deep pockets. They are also involved with development in the other countries mentioned, together with major shipping companies like NYK and COSCO.
On the purely maritime aspect side I know for a fact that experienced mariners are involved, not just Engineers and Computer experts. At least here in Norway, where much of the development work is concentrated. (Actually “next door” to me)
Exiting time for some, scary for others. Luckily I’m only watching from the sideline, trying to stay informed.