This is only unpopular because so many people would rather have free, congested roads and expensive, slow transit than just paying for the roads. This will clearly change before most of Seattle's new plans are completed: self-driving cars will clog up our roads waiting for riders unless we charge them to be on the roads.
Building bus rapid transit makes much more sense because it is felixible and less expensive. But politicians don't like photo ops with buses nearly as much as with trains.
However, once you are going to the trouble of building dedicated right-of-way, the incremental cost of making it rail isn't too bad compared to the enormous efficiency/scalability of rail over buses.
Also, people don't feel safe driving at 60mph in a densely packed road, so they don't. This should be no problem for a computer, so you could have a densely packed highway but traffic still moving at very high speeds. Basically, flow velocity could remain constant as the highway becomes more congested.
On city streets, your car would route around traffic light timing, in coordination with other vehicles, to distribute traffic and prevent unnecessary stopping.
Edit: One caveat worth mentioning is that these optimizations would be most effective if a road is 100% automated cars, though maybe dedicated lanes are a good transition mechanism. Once the safety benefits are clear, I think adoption will be very fast. The societal cost of car accidents is absolutely massive.
Not only do they not feel safe, they aren't safe driving bumper-to-bumper.
Computers will have better reaction times, so much of that goes away. But it's still dangerous. If two cars ahead of me slows, and the car ahead doesn't slow and collides, it will decelerate very quickly, perhaps too quickly for my brakes at 60mph.
I'm all for self driving cars, but I'd like to see solutions for the problem today, not just letting things get worse until something magically makes it better.