It’s harder to manage the stoppage distance of trains over long distances than of any other vehicle type. Trains need to identify stations or obstacles well in advance before making a calculated stop. To make matters worse, the friction between a train’s metal wheels and metal tracks is much lower than that between a car’s tyre and the road. So coming to a quick stop needs human oversight.
Human presence does not override physics. Advantage at the moment of a person is image recognition - but that may not be true in the future as advances are made in this area.
Additionally, railway workers across the world are unionized, and would not want to lose their jobs to automation. Plus, many people feel UTOs might be risky, and as such, there’s not a lot of public support for these vehicles at the moment. All these factors tell us that completely automated, long distance passenger trains may not become a mainstay at least in the foreseeable future.
Doesn't seem to be an issue in Sydney where the new metro is going in. Whoever makes fully automated commuter rail possible will make a lot of money!