Even if the development of autonomous vehicles eventually reaches the level you're hoping it will, and none of the obstacles which might prevent the realization

Even if the development of autonomous vehicles eventually reaches the level you're hoping it will, and none of the obstacles which might prevent the realization of your vision actually occur, induced demand means that any reduction in the level of congestion will be temporary. Anyway, you're suggesting that we could get a 2x or 3x improvement in 20-30 years by completely overhauling all of our transportation infrastructure: but we can get a 10x improvement in the same span of time at a lower cost with less risk using proven technology. I'm happy to see people proceed with research on autonomous cars, and can certainly see the value of an Uber/Lyft type service using robot drivers, but it would be a mistake to sit and wait for speculative technology research to maybe solve the problem when we can for sure solve the problem if we just get going with what we already have. What's the worst that can happen, we end up with lots of good options? Robot cars might make cabbies obsolete, and it is likely that they will change the way we use other transportation technologies, assuming they work the way people are hoping they will, but they won't make light rail obsolete any more than the invention of the jet engine made the automobile obsolete.