I'm sceptical about driverless cars. Today, most trains still have drivers and rail is a heck of a lot easier to automate, than road vehicles. When driverless trains come into widespread use, then there's a possibility the same will happen with road vehicles.
It's an imaginary, i.e., legal, barrier.
Consider: there aren't as many trains on the road as there are cars.
The cost of developing driverless trains is high.
The cost of defending driverless trains in court is high. (It's likely that a robust solution would succeed in court, but the inevitable fact is, it would be challenged, and that challenge would be very expensive.)
Presumably, one can conclude that the maintenance cost, of having minimal human labor onboard the vehicle, is lesser than the amortized sum of the above steps. Therefore, they do not pursue it.
Trucking has a similar problem, with a greater challenge (roads are more complex, and the driver must cooperate with other human drivers), and as it is right now, they're pinching off the labor as tightly as they can. A truck driver still has more to do than a train driver, but more and more, their activities are proscribed by the constraints of their employer, and the legal obligations around that (labor laws setting mandatory break, for instance).
A general purpose passenger car has somewhat
less liability compared to those other applications: modern cars are quite safe, even when driven very poorly, so the medical and legal risk to the company making them is relatively small. They would still be exposed to class-action lawsuits, but that would only be applicable once a large customer base exists. The early experiments (where we're at now) present only internal risk, and early adopters (where we'll be, fairly soon), present only a moderate risk.
Once the consumer version is perfected, enterprise grade versions will be picked up by all the shipping companies. It doesn't make any sense for them to stick their neck out, when larger vehicles, more valuable cargo, and more homogeneous fleets, are at stake.
What I'm curious to see play out is, how much of the roads will remain human-accessible in the next 50 years? Once human drivers become the minority (and they absolutely will), they will become the marginalized, restricted, and ever-more-vocal group. Will they be limited to certain open-access roads or highways? Will driver qualifications tighten? (Perhaps making US drivers more like Germany's.) What will be the equilibrium fraction once self-driving cars reach saturation -- 50% human? 10%? 1%? I'm also absolutely certain that human drivers will not go away completely; but they will be limited to enthusiasts, whether because they like it that way (as most car enthusiasts today), or because they want to keep doing it that way (traditionalists).
Tim