Robocars should make a serious change in the economics and dynamics of parking. There are many reasons for this, but one of the foremost is that a robocar doesn't "park" as we think of it, what they would do is more accurately called "standing" because there is always a "driver" attending the robocar. More on that below.
Ideally, robocars should result in a serious reduction in the number of cars present in a city. Robocars can really enable car-sharing and the use of hired robotaxis, to the point that far fewer cars are needed, and far more people decide they don't need to own a car. I predict many people may hire out their own private cars as robotaxis, and groups of people will come together to form "car clubs" which own a pool of cars.
The result would be a major reduction in the number of cars needed in an area, which is almost entirely a reduction in the number that are "parked," since the number that are using the roads stays the same or even goes up. This alone should alleviate any under-supply of parking for some time to come.
However, let's imagine that everybody still wants to own a car, and so lots of parking is needed. Robocars still can change the equation in many ways.
Robocars should encourage the use of half-width, 2/3 length single person "jaunt" cars, which are perfectly good for a large proportion of trips. These would fit almost 3 into a present-day parking space. (Five, if you account for the valet parking density factor noted below.)
There are also designs such as the Twill and MIT City Car which fold to stand upright when parked, or even stack against other cars. These cars can park in 1/2 to 1/3 of their driving footprint.
Robocar parking will also be dynamic. There is no need for marked out spaces that are all the size of a large car. Short cars can take a short amount of space, and boats can take more. Cars can move in concert to fill in gaps.
In dense areas, robocars can park themselves fairly far from their owner's destination, as long as the owner can give some warning (or even some vague schedule) as to when the car is needed next.
This allows two things. First, cars need not crowd dense downtown cores. If there is not enough parking downtown, those cars that have a sense of when their owner will need them can wait/park elsewhere, where it's cheaper.
Two, this allows "load balancing" of parking lots. Right now most buildings try to have enough parking for their peak demand time. Free spaces in a lot half a mile away are of little value to users of a building. With robocars, every lot can be fully utilized as space fills up. Every lot is thus "serving" a very wide area.
Robocars don't park, so they can wait where they block things, notably low-usage driveways, fire hydrants and other cars. As long as there is just a short notice that somebody wants to use the driveway, the robocar can clear away. In most cases, people leaving and returning to their homes will never even notice a robocar was waiting in front of, or even in their driveway, because their house could broadcast a signal requiring any vehicles blocking the way to leave.
Blocking driveways not only frees up a tremendous amount of parking space in front of those driveways, it also frees up other space that was previously unusable because the distance between driveways and other no-parking zones does not match the length of cars.
Indeed, robocars can also park fairly close in a line, using exactly as much space as their length. There should be no need for pre-marked parking spaces which are the length of a typical long car. Robocars will fit tightly together, and could move on request to close or open gaps.
They will probably be better parallel parkers than we humans are, and thus be able to deal with a smaller gap between cars. Some might even be able to turn all their wheels and go in sideways.
Robocars should be able to outdo even the best parking valet when it comes to parking densely in a lot. This is not just because of their ability to do precise moves in close quarters, it's because they can coordinate, often even without radio. For example, if rearranging a valet lot requires moving every single car, that's something the robocars can all do at once, and an almost impossible task for human valets.
Of course, robocar lots may want to use radio to organize so that cars that may be needed soon can get out more easily, and cars with known (and later) need times will be more blocked, but in truth no car will be very blocked, because of the robocars ability to move in concert.
The combination of smaller cars and super-valet parking should allow typical parking lots to hold 2x-5x more vehicles than they can today.
If we ever need to build more parking lots, lots designed for robocars could be even denser. For example, they could have sections with very low ceilings -- because humans almost never go in there. A parking structure could thus have twice as many storeys.
Robocars might not only stand blocking driveways. At low-traffic times (which are exactly the times that the most parking is needed) robocars could double-park or even triple park on the streets. One could imagine a street with 6 lanes at rush hour that, during the night or middle of the day is reduced to just 2 lanes with 4 lanes of parking. Or even to a single one-way lane. This works well because cars have to be somewhere. They either will be moving in the driving lanes or waiting in the waiting lanes. The total amount of space remains similar (parked cars are of course denser on the ground than moving ones.)
Robocar double parking doesn't mean a problem for the cars on the curb lane, even if the cars are quadruple parked. Of course, cars would be encouraged to try to organize themselves so that cars needed soonest are on the outside, and long-wait cars are on the inside, but that just reduces the amount of moving.
That's because robocars could, when double-parked on a block of moderate length, always leave one "gap" in the line of cars. This gap could be in the middle, or it could be a more natural gap at one of the ends. If there's a frequently used driveway, it would be the gap.
When a car on the inside needs to leave, it would just ask the robocars in the outer lane to move so that the gap moves next to the departing car. Since all the cars can move at once, in concert, this would take just a few seconds, especially for electric cars. If there are multiple blocking lanes, all those lanes would, at the same time, move their gap to be in the right place for the car that wants to leave. It doesn't matter how many lanes there are, this happens in the same amount of time -- a few seconds.
Through the gaps, the existing car departs. Then the car in the double-park area which has the longest expected wait for its master, and which fits, should take the spot on the curb lane that was vacated. This time the cars in the curb lane will do the moving, to move the gap next to that car, and it can then take he spot.
If you have a very large number of parking lanes -- now we're talking more about a parking lot than a street, it might make sense to leave the lanes more sparse, but collapse them when possible.
Thus, with a very small cost in gaps, and possibly none at all, it's possible to do double and triple parking on any city street with almost no delay for a car on the curb which is summoned to pick up a passenger.
All of this could be self-organizing, but it also makes sense for each block that allows this to have a parking control computer which tells cars in the area what to do. It could announce when double-parking is OK, and when all cars should leave, even those on the curb, because more traffic capacity is needed.
Robocars parked on the sides of the road would also keep their basic sensors on, to detect pedestrians who are walking between the parked cars in order to jaywalk. Cars on the road can be signaled about the presence of such jaywalkers. This will prevent any jaywalker from entering the road unseen from behind a parked robocar.
(The "move in concert" behaviour does not need communication between cars. If the car behind you starts moving towards you, the natural reaction of a robocar will be to also move to even the gaps and see what the car in front starts doing. If the car in front does not move, everybody stops. If the car in front moves, it propagates forward until you reach a car that can't or won't move. Fancier signals like "this is temproary" do require coordination.)
In spite of the surplus of parking that robocars could provide, I think we should eliminate free parking. We should replace it, however with a dynamic market in very cheap parking.
Right now cars are subsidized greatly through the provision of free parking.
I believe parking should be sold at real market rates. Those rates might not be too high, but obviously they should cover the cost of the land. Administration of the parking, all done via computers, should become very cheap.
There would be, pardon the pun, a "spot" market in parking. Spaces would have fluctuating prices based on bid and ask. Cars would have software to bid on parking or seek parking that offers the best combination of price and location. People who want to keep their car close to them will pay more than those who are willing to have it park further away. Some will only have their car park close to them when it gets near their anticipated departure time (for example from work.)
There is a risk that this might encourage excess movement of vehicles. The prices will have to be balanced, based on proper externality fees on the energy, so that the right decision for the environment (both urban and atmospheric) is done.
One way to have "mostly free" parking (since I think there is going to be a lot of spare capacity) is to use a 2nd price auction. In a 2nd price auction, the winner pays the price of the second highest bid. However, if there is only one bidder, there is no second bid, and the price is thus zero, or perhaps some minimum floor price. As a result, as long as there is a surplus, the commodity is free. As soon as there is contention due to high demand, it suddenly has a market price.
This sort of system may make sense for space, such as street parking, that is already there and won't be put to any other use during the day. However, if the price of parking becomes too cheap, we will see parking lots converted to other uses, be they parks, roads or human occupation. Over time, the parking supply will drop until demand and supply are better matched.
But that's a long time, since there is so much supply, when you consider that every spare driveway, every emptied garage spot and every spare spot in every existing parking lot can be made available if there's financial incentive. Today there are 3 full-sized car spots, with access lanes, for every car in the USA.
Of course, some of the methods described above require that you have an entire line of robocars. Truly stationary HDVs won't move on command.
HDVs of course must not double park, but one could have a situation where HDVs park on the curb, and robocars double park next to them. The robocars would notice the HDV turning on its turn signal to exit, and could move to immediately offer the driver a gap. The robocars could also leave a gap to allow HDVs wishing to park on the curb a way in if an HDV stops near an open spot on the curb, as one would for parallel parking, and activates the turn signal.
HDVs in the curb lane would prevent the simple algorithm of moving a vacated curb lane spot for quick access by the car with the longest predicted parking time, but this can still be managed with slightly more complex movements in the double parked lanes, or by placing a slightl less ideal car in the vacated spot, or leaving it open for another HDV.
Indeed, if the curb is full of robocars, and an HDV approaches, stops and signals the desire to parallel park, and the block's parking control computer has authorized double parking on the block, then a robocar along the curb (and not blocking a driveway or hydrant) could vacate that spot and move to a double parked position.
The early robocars will not be able to count on special synchronized parking actions with other robocars. However, parking lots can slowly allocate small, then growing sections of their lots to robocars to valet park in, and in fact can bid on the business of selling spare space for waiting to interested robocars. Unlike ordinary lots, all lots within a fairly wide distance can compete (online) for a robocar's business, and set up a simple system to identify robocars that paid online for space. A robocar would try to park close to its owner but would accept attractive lower bids further away.
A parking lot need not be an official parking business. Any parking lot, even one for a store, that usually has free spaces could advertise space at a low price. Even if the store does not know if their lot is full or not, and does not want to put up a webcam that can figure that out, the system could just generate a database of how often it is full to help cars decide if it is worth going there to check it out. Robocars could promise as part of the deal to leave if the lot actually becomes full so that HDVs are never without a spot. The robocars will of course use only the less desireable spots, further from the door. It's a win-win for everybody.
Robocars crusing metered parking could find empty spots easily enough but could not use them until cities put in means to pay for them. Many cities are going to centralized metering and even to being able to buy parking on the internet already.
Robocars might find themselves preferring "standing" places where they can't park and don't have to pay to park. For example, a driveway with a car in it can be used, because the robocar can see the car's lights come on when it starts and move away quickly. Some homeowners might decide to scare the cars away but this would become tiring and not very usful. A waiting robocar in your driveway would cause only a slight delay when traffic is so heavy it can't pull out right away -- in which case neither can you.
When rush hour comes, Robocars should obey the orders of a block's parking control computer to clear all parking, and enable all lanes for traffic. However, HDVs won't do that. Of course, they will be violating the parking rules that already exist on streets where parking is forbidden during rush hour, but that doesn't help move the traffic.
Fortunately, robocar techniques allow the creation of a robot tow truck, (perhaps Towbot?) which can make it easy to clear the streets. When a car blocking traffic is reported or found, a tow truck could be summoned. An operator is probably needed to attach the tow mechanism to the illegally parked car, but after that the tow truck should be able to tow the car to the impound lot without a human. The tow operator can then move directly to the next job, where another robot tow truck can be placed. How to afford all these robot tow trucks? Expensive parking tickets, presumably. Though frankly, the faster rate of enforcement should strongly discourage people parking on rush hour streets.