I want to enhance two other ideas I have talked about. The first was the early adoption of self-driving cars for parking. As I noted, long before we will accept these cars on the road we'll be willing to accept automatic parking technology in specially equipped parking lots that lets us get something that's effectively valet parking.

I also wrote about teleoperation of drive-by-wire cars for valet parking as a way to get this even earlier.

Valet parking has a lot of advantages. (I often joke, "I want to be a Valet. They get all the best parking spots" when I see a Valet Parking Only sign.) We've given up to 60% of our real estate to cars, a lot of that to parking. It's not just denser, though. It can make a lot of sense at transportation hubs like airports, where people are carrying things and want to drive right up close with their car and walk right in. This is particularly valuable in my concept of the minimalist airport, where you just drive your car up to the fence at the back of the airport and walk through a security gate at the fence right onto your plane, leaving a valet to move your car somewhere, since you can't keep it at the gate.

But valet parking breaks down if you have to move the cars very far, because the longer it takes to do this, the fewer cars you can handle per valet, and if the flow is imbalanced, you also have to get valets back quickly even if there isn't another car that needs to come back. Valet parking works best of all when you can predict the need for your car a few minutes in advance and signal it from your cell phone. (I stayed at a hotel once with nothing but valet parking. The rooms were far enough from the door, however, that if you called from your room phone, your car was often there when you got to the lobby.)

So I'm now imagining that as cars get more and more drive-by-wire features, that a standardized data connection be created (like a trailer hitch brake connection, but even more standard) so that it's possible to plug in a "valet unit." This means the cars would not have any extra costs, but the parking lots would be able to plug in units to assist in the automated moving of the cars. For example, consider a car with DBW (drive-by-wire) steering, throttle and brakes. It could not park itself (or it might be able to parallel park) but the parking lot could plug in a unit that had the right technology. For example, if there is a guidewire under or along the road, the unit could have sensors able to detect it. The car would not need to standardize. Fancier self-driving units, with cameras, lidar and the rest could also be made available in the valet unit. Or for teleoperated valet parking, the valet unit could have the cameras needed for the remote operator. Ideally the data bus would be wireless. You would throw a switch (with the key in) to pair up with the valet unit, and then it could sit on the roof or the hood to do the job. (Though it would be nice if it could get power from the car.)

Even the simplest cars could make use of a valet unit which is effectively a tow vehicle. Consider a small tow vehicle which simply hooks onto the vehicles tow hook. The car itself is left in neutral, but the tow unit gets control of the brakes. Most cars have power brakes, so drive-by-wire brakes would be a very simple addition to the typical car. No more complex than the remotely controlled brakes of a trailer. The tow unit would then test the brakes (applying them should cause tension on the tow line, otherwise slow down very gradually and signal a problem, and don't be going downhill) and then tow the car to its parking location, which might be some distance away. No backing up of course. In this case a permanent human valet would probably take control of the car for its final parking -- and removing and re-connecting the tow vehicle on demand.

Now none of this need be very fast in places where you get lots of advanced warning of need for the car, like airports where you have hours of notice. The valet units could drive the cars to their spots at 2 miles/hour if need be, to assure as much safety as required. In malls and offices, the tow to the parking area can be slow, but the trip back needs to be fast if the person wants their car right away.

Of course, if the car can give control of throttle and steering as well as brakes, you don't need an actual tow vehicle. Though you can still go as slow as you like, at least at airports. Valet units might start with simple following of guide-wires, and progress to active vision systems. Even if used in sealed off lots, they still would need basic sensors to avoid hitting things, but at very slow speeds the response time can be quite slow. It's also possible that the alleyway to and from the valet area could have other external sensors not on the valet units which detect any obstruction and signal all units to pause until it is dealt with. In that case cars could return quickly along such an alleyway, knowing it's safe to go faster.

Tow units could be battery or gasoline powered, but if they can plug into the car's electrical system with sufficient current, they could run off the car's battery or alternator. Again, they don't need to be very fast so they don't need the sort of current electric cars deliver.

Naturally, if having more DBW in your car gets you better parking or better rates, people will want it. Many cars already have a lot of this, particularly hybrids and electrics. And the cars that can self-parallel park have all of it. Any car with cruise control has some DBW control of throttle. It may be possible to make a simple tow vehicle that uses the car's throttle for power but pulls the car by its tow-hook to get steering. This tow vehicle would not need nearly as much power and could work with minor modifications on most cars today.

Having a valet at the remote lot is reasonably efficient because they don't have to move with the cars as they travel from the owner drop-off to the remote lot, they only deal with cars coming in and out, like the valets at the owner point. It's not out of the question that owners could even connect valet units themselves, especially if it's just a matter of sticking it on the roof with a suction cup and enabling it.