Outsourced valet parking with drive-by-wire cars
There already are some drive-by-wire cars being sold, including a few (in Japan) that can parallel park themselves. And while I fear that anti-terrorist worries may stand in the way of self-driving and automatic cars, one early application, before we can get full self-driving, would be tele-operated cars, the the remote driver in an inexpensive place, like Mexico.
Now I don't know if the world is ready, safety-wise for a remote chauffeur in a car driving down a public street, where it could hit another car or pedestrian, even if the video was very high-res and the latency quite low. But parking is another story. I think a remote driver could readily park a car in a valet lot kept clear of pedestrians. In fact, because you can drive very slowly to do this, one can even tolerate longer latencies, perhaps all the way to India. The remote operator might actually have a better view for parking, with small low-res cameras mounted right at the bumpers for a view the seated driver can't have. They can also have automatic assists (already found in some cars) to warn about near approach to other cars.
The win of valet parking is large -- I think at least half the space in a typical parking lot is taken up with lanes and inter-car spacing. In addition, a human-free garage can have some floors only 5' high for the regular cars, or use those jacks around found in some valet garages that stack 2 cars on top of one another. So I'm talking possibly almost 4 times the density. You still need some lanes of course, except for cars you are certain won't be needed on short notice (such as at airports, train stations etc.)
The wins of remote valet parking include the ability to space cars closely (no need to open the doors to get out) and eventually to have the 5' high floors. In addition, remote operators can switch from vehicle to vehicle instantly -- they don't have to run to the car to get it. They can switch from garage to garage instantly, meaning their services would be 100% utilized.
Somebody who parks a lot would readily justify the cost of the remotely operable car (drive-by-wire, cameras, security system, wireless ethernet) with the reduced cost of parking, and also the convenience. Electric cars will typically be drive-by-wire anyway. You would pull up to the mall, office tower, train station or airport, and place your car in the entrance lane of the remote valet parking area, which will also be much closer to where you want to go than ordinary parking spots. You would get out, and sync control of your car over to the lot after approving their digital certificate. You car could wait for a bit there, and when cars in front are clear and an operator is ready, the operator would drive the vehicle to a parking spot appropriate to your estimated return range.
After all, people spend $150 to $300 per month for urban spaces, and $10 to $30 per day for airport spaces. Cut 40% from that price and you can justify this type of car. In the garage, the cost is the network connection, and a private entrance and area that grows as the number of cars that can do this grows.
A few minutes before you want to leave, you would either use the internet or call a special number on your cell phone, which would signal for an operator to extract your car. A team of operators would quickly move the cars blocking your car into temporary spots, and then bring your car to the entrance, where you can regain control with your key or code. In addition, you could have a signal to say you're going to leave "soon" which moves your car to a holding area near the entrance where it can then be extracted on 30 seconds notice.
Eventually, we might start to accept the telechaufeur on an ordinary street, which would really improve the market for a car like this. However, once human life is at risk, the safety and insurance bar is a lot higher. For that, the chauffeur would probably need to be local, and the network connectivity highly redundant and with a failure rate lower than mechanical failure of the car. (While you could require the passenger sits in the driver's seat to take over at any moment, you know that some would take the opportunity to sleep.) It's possible that as a step to the crash-avoiding cars we need, we might be able to create a car that, on loss of signal, begins a fail-safe process, staying in lane, and breaking at a modest rate until the in-car driver can take over. However, it's harder to handle tricky moments like lane-changes.
One could imagine a telechauffeur strictly for highway driving on highways with well marked lane markers. We're almost at the point where a computer can handle that, and I think a computer could handle the gap between loss of signal from the telechauffeur and takeover by the person in-car. The highway would need to have fiber optic cable running along with with radio nodes every few hundred feet, with dual systems to attain that level of reliability. This part may be far enough in the future that we wait for the computer-driven car instead.
I think, however, that local telechauffeurs making $15/hour would be economical. Unlike regular chauffeurs, these could immediately pick up another client after releasing your car back to you, so they would be 100% utilized, though work would primarily come at rush hour -- a nice moonlighting job. Many would pay a rate like this in order to be able to work or read during the commute.