All in all I think we are at the top of the hype phase for what regards self driving cars. We have seen a few of them winning challenges, some driving on highways and some negotiating traffic in urban settings.
At the 2014 Automated Vehicles Symposium a poll was taken among the 200+ participants. The consensus was that by 2020 we will be able to buy a car that drives itself in most situation with override option to take control of it. Interestingly there was also a generalised feeling that by 2030 you will get a car that will drive itself but you won't have the option of taking control. You will be just a passenger and please do not speak to the driver!
As I said, my take is that we are at the peak of the hype curve and in the coming years we are going to see several drawbacks that will push those dates further down the lane.
What I ca see is that the number of gimmicks (if I may say so) that will provide assisted driving will increase over the next years: self parking cars, cars that can keep a cruising speed over the whole trip to get us to a given place at a certain time, cars that can interact with the infrastructure to select a slot and avoid queuing (like slowing down if there is a block road ahead so that you get there when there is a way through, driving assisted vision (like when driving in fog or letting you see beyond an obstacle), self braking, overtaking controlled by the car and so on.
These are all things that I see coming relatively soon. Car makers will be pushing them starting with top of the line models, will charge good premium dollars to start with, and progressively they will become cheaper, part of bundles and then standard equipment.
For a full self driving car, even one that hands over the controls when the situation gets tough, I think we should wait 10 years more, at least, than what these expert are predicting.
In a news reporting the polls I mentioned before there is an interesting observation: we have today most civil aircrafts equipped with automated pilot and most of them indeed fly on the automated pilot. If the system ever has to disconnect, for some reason, the pilot has at least several seconds to react (even in the approach phase) whilst in a car the hand over would have to be managed in a fraction of a second and this is not feasible. So the challenges facing self driving cars are tougher then the ones that had to be overcome in civil aviation.
Of course, I might underestimate the potential of technology evolution and it only takes 6 more years to see if I am wrong. But here it is a bit about technology, and a lot about other aspects, including regulation, responsibility, accountability and not least acceptability. Yes, to me it is not going to happen in the next six years.