I think a robo-Uber is more likely. Mostly because the designated highway sections would require concerted public actions and overcoming a lot of NIMBYs.
Robo-Uber requires to solve technical problems (and some regulation). Designated lanes are more of a social problem. (We could have had automatic driving on designated roads for decades; just make the cars run on rails, too.)
On the other hand highways are much simpler and could be completely self-driven with today's technology, probably much more safely than a human driver. This takes people out of high-stakes boring driving where the vast majority of fatalities take place and give them an easier, safer commute.
I can't react to debris in the road, a pothole, or a deer crossing the road in 1 ms like an automated driver can and at 80 MPH that's where it really matters. A 25 MPH collision is basically just a matter of money, as with today's cars there's almost no mortal danger. Hitting a deer at highway speeds is a life-threatening problem.
You can make a flatbed-type powered sled that people drive their cars onto and it drives you from exit A to exit B. It can enclose and lock the wheels of your car for safety. This can even be electric or hybrid to reduce costs with a long, flat Tesla-like battery under the flatbed. If I could drive my car onto a system like this and take a nap for 5 hours on the highway I would do this in a heartbeat because I have confidence that today's technology can drive a restricted access road better than I can in good conditions.
I don't think you need to assume switching driving modes to a flatbed which would introduce bottlenecks and reduce flexibility compared to individually autonomous vehicles.
>I can't react to debris in the road, a pothole, or a deer crossing the road in 1 ms like an automated driver can and at 80 MPH that's where it really matters.
The laws of physics still apply but, yes, in general a computer could do a better job--at least on average--than a person. (Although I suspect there will be some things a computer is very good at and a few where it tends to make errors that a human might avoid.)
I agree with your basic point though that autonomous driving on significant sections of highway is by itself a big win even though it doesn't fundamentally change how most people own and drive cars.
I'd heard that the actual reliable sensors that can see through rain and such are like $50k+ so are not affordable in private cars, but a service that uses it all the time could. I wouldn't want to fall asleep with today's consumer-grade hardware, but you're right that even an autopilot that has to be supervised still has most of the safety benefits.
>Mostly because the designated highway sections would require concerted public actions and overcoming a lot of NIMBYs.
But it's much closer to being technically feasible than the general case.
That said, I don't really disagree with your point which is one reason why it's further out than a lot of enthusiasts would like to think. The likely scenario is that (initially) high-end luxury cars will get features that could allow them to operate autonomously on designated sections of highway. The problem, as you point out, is that it would take concerted public action to change laws and designate those sections (perhaps requiring some sort of radio beacons to enforce the designation). That's a fair bit of public action, and probably at least some money, to let "rich" people fiddle with their phones while tooling down 280 or wherever. I'm not sure it's something I, as a politician, would be anxious to champion.
I suppose there could be commercial benefits as well--trucking and so forth--but given that you still need to pay a human to sit in the vehicle, it's unclear that highway-only self-driving would given businesses much of an economic win.
Robo-Uber requires to solve technical problems (and some regulation). Designated lanes are more of a social problem. (We could have had automatic driving on designated roads for decades; just make the cars run on rails, too.)