The big opportunity of robocars isn't the cars themselves; it's how they could create a far more efficient transportation system.
When I've thought about driverless cars, which if you believe Sergey Brin, will be available within "several years," I've tended to think of them as a drop-in replacement for our current automobiles. So, you'd buy a VW Automaton and it would sit in your driveway until you wanted to go somewhere. Then, you'd hop in, say, "Take me to Lake Merritt," and then just sit back and pop in the latest Animal Collective while the computer drove.
But maybe that's not what would happen at all. Changes in transportation technology have tended to be accompanied by changes to transportation systems, too. Long-time technologist Brad Templeton argues that this will, in fact, be the case. And he's even got an idea of what the big shift might be. We could enter the age of the "whistlecar." If one can hire a cheap specialized 'robotaxi' (or whistlecar) on demand when one has a special automotive need," Templeton writes, "car users can elect to purchase a vehicle only for their most common needs, rather than trying to meet almost all of them -- or to not purchase at all."
This vision is kind of stunning: imagine the Kiva Systems logistics robots that now speed around major warehouses, but for people. Transportation-as-a-service models could really take off in a world of hyperoptimized robotaxis. Not only would the robotaxis be built differently from normal cars, but people's private vehicles (if they had one) would change as they realized how they could use the new system more effectively.
That is to say: right now, people buy big old SUVs and cars that drive 400 miles on a tank because they are buying for the maximum number of use cases. Really, most people drive their cars a few dozen miles at most and they do it alone. People have WAY more car than they need. So, Templeton's conceit is that if we had roaming driverless vehicles that would show up at your door when you called one, you might be inclined to buy "less car" because you'd get the rest on-demand.
My own thought: perhaps when you bought a small, electric vehicle, you'd get a "service plan" that came with X number of trips in a driverless vehicle of your choosing; your bundle would be the small, energy efficient daily car and access to self-driving vans, trucks, station wagons, and sports cars.
Templeton's theorizing could also answer some of the critiques from transit-oriented environmentalists who see driverless cars as perpetuating the doomed auto-heavy American system. Don't think about the driverless car as a fossil-fuel powered car replacement; think of it as one mode of a radically more efficient system: what could you do now within a system that now has free-floating semi-autonomous people transporters?
Let's say the tech opens up the system to change. Templeton's main argument is that the new system would bounce back on the tech. The design of cars would change because they'd have a new set of uses, possibilities, and constraints. I'll just provide the bullet list here of what might happen to cars in the auto-automobile era. You can check out his post for the details.