Imagine, now, that self-driving combined with Uber pool, so that people who need to get from and to the same destinations at the same time (as is typical for commutes) would be able to share the ride.
Imagine then making those vehicles larger to enable not only a decrease in traffic, but an increase in housing density, local infrastructure in walking distance, and economic growth.
Imagine if we then realized we could charge that elongated car as it's driving if we put electric lines above streets.
Oh, and why not increase rolling efficiency while we're at that, and make the task of self-driving simpler by embedding signaling infrastructure into the road itself.
You just imagined buses, trolley buses, and streetcars respectively.
Which is what makes the most sense, and will continue making the most sense for vehicles that don't sit in your garage most of the time.
There's more to car ownership than being able to get from A to B in principle.
Once you add immediate availability, individual vehicle preferences, having your stuff in the car already, the ability to keep your stuff there once you arrive at your destination, never having to think about small, but normal usage damage (especially if you have kids), or, conversely, having the vehicle just as clean as you want it...
...you will see that garages aren't going anywhere any time soon, self-driving cars or not.
And that we'd be wise to develop actual public transport infrastructure for the people who really do only care about getting from A to B, with some acceptable wait time. It's a win-win for everyone.
My pessimistic prognosis is that the development of the self-driving car will make the taxis slightly cheaper, but instead of paying your Uber driver, you're going to be paying engineers, maintenance, and cleaning staff.