Handling the pick-up "rush" when everybody leaves at once

Taxi line at a (sucky) airport

The discussion on cars circling to avoid parking fees (short-answer, they won't) leads to a few other interesting issues I want to cover.

Today, I want to talk about the pick up rush. Earlier, I covered what should happen with Lyft and Uber at airports to parallel what happens with taxi lines. Let's look at the evening rush hour and related situations.

As the work day ends, everybody wants to get home at roughly the same time. Some will do it in regular cars or on transit like today. Some will want robotaxi service. Some will want to take their private robocar home. And others will be open to self-driving transit forms which could come at a much reduced cost.

If we look at those in private robocars, robotaxis and regular taxis, we face the problem that there may not be enough "pick up curb" for all the people wanting pick-up. We see that at airports, which is why the "cell phone waiting lot" was created. Buildings have far less pick-up space than airports. We also see it at schools where most of the children are driven home by their parents.

The problem is people want to avoid waiting for long at the curb. They would rather summon their vehicle in advance, and have it waiting for them when they get there. This can quickly fill up the space. Most people will try to time it well, but many will signal for their car when they are almost ready to leave, and then be inaccurate in their estimate of when they would actually leave. Some will summon cars parked in cheaper parking 10 minutes away from their building, and thus have much more trouble matching arrival at the pick-up spot, though their phone will be aware of the approaching ride and giving the signal to get to the spot.

With robotaxis, we might imagine we can solve the problem with the "taxi line" concept. We certainly don't right now, as Uber/Lyft allocate a specific vehicle to a specific customer rather than having the next customer take the next available vehicle waiting in a line. We could try to implement a waiting area, but the problem is there can be 6 or 7 different services from Uber, another 5 from Lyft, and several other competing companies. Each class of service in theory needs its own line, and few curbs or even pick-up areas have enough room. The uniformity of taxis, allowing the line, is one of the few benefits of the way old taxis are regulated.

It's easier to do if the building has a parking facility. The facility can make sure there are enough spare spots in its parking lot for use as pick-up spots. But many buildings have only a little curb space in front of them, if that.

When it comes to private robocars, we have a bunch more questions. Each car is there for its owner and needs a place to wait if the owner is not there yet. If there is no place to wait, the car needs to go somewhere. If short term parking can be rented, that's OK, but if it can't, the car's only option is to keep driving -- which is going to increase congestion at the busiest time of the day. Indeed, it might start "circling" not because it does not want to pay for parking, but because there isn't any to be had.

Ideally, private robocars that parked nearby would not begin their journey until it was certain their owner was ready to get in. The question is, how would one enforce that? If you make a punishment for the car showing up and waiting, cars will just circle.

For private robocars parked more distantly, the owner will not be waiting when they summon. They will summon some time in advance of when they want to actually leave. The owner's phone will signal them when they must go down to the pick-up spot, but again, it's not easy to enforce violations of that. Cars won't have a "circle" command, but they will be commanded to travel to the pick-up zone for their owner, and if that zone is full, they will have no choice but to move on.

I can think of a few solutions, the most obvious of which is general road management and congestion charging that makes it expensive to circle. Even with the owner waiting, however, there may be times when no pick-up spots are free because volume of traffic is so heavy.

Designated pick-up locations

If you have used Uber/Lyft in certain towns like Las Vegas, or at airports, you learn they have designated pick-up areas. If you want to get picked up at a particular hotel, you must go to the designated area, you can't get picked up at side doors or on the street. Cities are already frustrated by Ubers blocking traffic to do pick-up/drop-off and are marking specific spots for this. This is likely to grow, possibly to the assignment of specific stalls for passenger exchange which are allocated at the busy locations.

Each building which is in a "congestion area" that needs a pick-up zone would have a computer to manage its pick-up spots. (Most probably they would just contract out to somebody else.) Cars heading for that building would, at the appropriate times, request a slot reservation, and an amount of time they might wait at the spot. The passenger's phone would also talk to this server to track how close the passenger is to the spot.

Here are some sample rules which might improve things:

  • If a building's pick-up area never fills up completely, it sets its own rules -- it might not even have management software at all. Buildings would be encouraged to have sufficient amount of pick-up area. Pick up areas should ideally use perpendicular or parallel stopping stalls with a pass-through lane.
  • If a private pick-up area fills up during a designated congestion period, such that cars can't go in, it must follow the rules of a "public pick-up area." Public curbside area is automatically under public rules at these times.
  • In a congested public pick-up area, if an incoming vehicle gets within one minute of the area and the passenger is not closer in time to the pick-up area, the vehicle must find the closest parking and park there, and may leave once the passenger has arrived at the pick-up area. It may not divert back before it has parked, even if the passenger has arrived -- this is a penalty for summoning too early during a congestion period.
  • If, under that rule a pick-up area still fills up, then no vehicle may be summoned except by a person already present in the area.
  • If, under that rule a pick-up area still fills up, then the building must limit who can use the pick-up area until it doesn't. Others should be directed to a larger pick-up area (such as one in a parking garage.) Examples of restrictions might be allowing use of the internal area by the employee of the month, or by carpoolers or robo-transit riders, or the disabled, or top executives.

Cars doing pick-up would need to get permission to enter a pick-up area before accepting one as a destination inside a congested zone. Such permission would be granted by both the building manager (who might allocate who gets to use it) and or the city, which would manage pick-up zones on public streets.

In addition, buildings might work together, so that if one pick-up area is full and there are spots next door, vehicles go there and the passengers walk a short distance if able.

I am interested in contributions from readers on good solutions. Particularly ones that can work in a more distributed fashion with less central control. The above plans can actually be highly distributed, in that each building can manage its own zones, and each block can be managed mostly independently of the others, but it's still better if there can be an emergent system. The other open question is the best way to enforce any rules that are put in place.

Reader suggestions

Harvie Branscomb suggests an interesting idea for highly congested pick-up spots, namely enforced carpooling. If a taxi (or a private car) wants priority at the curb, require it to take other people who are waiting to a place along the way that is less congested or more spacious for them to meet their cars. People could insist on private pickup, but might have to wait. It would be nice to do this today for school pick-up, which is very crowded in the USA. But people are paranoid about their kids getting in the cars of others, even other parents.

Robotaxi or van away from work

One more extreme option is to discourage or disallow pickup by privately owned robocars in highly congested areas which don't have adequate pick-up spots for everybody. This is a variant of road metering, but it has a specific target.

The private robocar would park/wait outside the congested area, in a space which has ample nearby drop-off spots. Private owners would be required to take the next available robotaxi or similar to the staging area, where they would then quickly transfer to their private robocar. This is probably not needed for drop-off

Even better if the people take a group vehicle, such as a van, to the staging area. This is efficient and reduces congestion on the road. Indeed, a building or block might have specific designated staging areas. If every robocar owner in the building is heading to the staging area, it means you can fill a van quickly during peak demand times. This depends on how many directions of travel there are from the building, since people want a staging area along their route home. If there's only one direction, you could even have something larger than a van. It's essential the departures to the staging area are very frequent, to keep the convenience that led people to use a private car. Solo robotaxis of course would leave on demand and be almost as convenient has having your own car pick you up.

As an added benefit, the parking out at the staging area will be much cheaper than the CBD. Those who want the convenience of quick pickup by their car will need to park near their location -- and pay a premium for it. If your car parks close, it is not a problem to go down to the pick-up area and wait a short time for it to arrive.

When it's time to empty a stadium, the use of vans going to staging areas may be the only system that works. We can't have private cars and taxis all go up to the doors of a stadium when 40,000 people leave at once.

At the other end would be the use of robocar transit. In that article I outline a group transport option which works to frictionlessly put people together in vans on the shared parts of their trips. Such services could seriously reduce congestion, and thus it would make sense to give them priority at pick-up spots.

In the morning

As I noted, drop-off is a different story. Drop-off should always be quick, there is never any chance of the passenger not being present. It is still possible to overflow a drop-off area but it should usually not last for long. If it becomes a problem, a subset of the above techniques could be used. In addition, if a car approaches a drop-off area and there is no available space, it can just go do any other nearby area if the passenger is able to walk short distances.


Those of us who are parents in suburbia are well aware of the inconveniences of this sort of situation. Drop off generally takes about 5-10 minutes. Pick up can take 30-45 minutes - every school day.

Adding multiple drop-off/pick-up locations is one of the best solutions. This is generally done by the schools for drop-off, and is part of what cuts down that time. I haven't seen it done for pick up, though it would be feasible, even moreso when dealing with robocars.

I think having some sort of nearby parking lot is going to be the fastest way to handle going home. That might be too expensive in the busy cities, although there I expect some sort of mass transit to be used (at least for the "last mile") anyway.

Issuing parking spaces would be centralized: Your car is told by the parking lot software exactly where to park. It might be told to move if that becomes necessary. "Your car" might mean your personal car, or it might mean the car operated by whatever robotaxi company wishes to use the lot.

As far as pricing and incentives, use of the parking lot would be free for up to 15 minutes for anyone who had the equivalent of a parking validation. But there would also be some sort of point system so that the people (or companies) who used the parking spot for the least amount of time would get the closest and most convenient parking spaces in the future. This would be part of the algorithm that determines where to tell each car to park, and would be set up so that the best spots have the highest volume of traffic going through them.

Outside of rush hour, the lots would be used for longer term parking. Depending on the portion of personally owned vehicles and robotaxis, the demand for longer term parking may or may not exceed the capacity of the lot. If it does, another point system giving priority to the people who tend to leave the earliest (plus the people willing to commit to leaving early that day or paying a fine) might work. During a transition period the farthest or least convenient spots could be used for non-robocar parking, and/or some spots could be used for valet parking.

There would, of course, be one or more traditional pick-up/drop-off spots in addition to the lot(s). This would be usable by travelers with special needs, and if there's enough space could also be used by individuals or companies who wanted to pay for VIP access. But it wouldn't be open to the general public. "No stopping or standing without a permit" if it's a public street.

Since what we're talking about is mostly changing people who used to drive to work and had to park to people who are getting picked up (by their own car or a taxi) there obviously has to have been parking nearby. When we have people switching from transit to taxi, then you might have no parking nearby and need managed pick-up spots.

One challenge not quite solved is that right at 5pm there is still high demand to park in the parking lots (for people not yet leaving) and also high demand for pick-up spots. In theory, the parked cars should be leaving but there might be a very short period of undersupply. When lots rent parking, they will be asking, "Do you need to stay past 5" and ideally the whole lot would not fill with people all wanting that.

Yes, although over the longer term it would be nice to be able to reduce the size of the current parking lots, at least barring an increase in the number of people at that location.

If 50% of commuters switch from human-driven cars to robocars, it might not be possible to cut the parking lot size by 50%, but maybe you can cut it by 35% if you are efficient enough with the space you use.

Likewise, for places still growing (which is most places), you might be able to double the number of workers while increasing parking by only 50%.

If you have a lot of people switching from transit to taxi, that's going to be a problem. I'm not sure that managed pick-up spots can handle that, without taking up a lot more space than they currently do. Which I think means that people just won't switch from transit to taxi, in high traffic places where parking lots are not feasible. From transit to roboshuttle (possibly with a transfer to taxi), maybe.

(I wonder how roboshuttles will work in terms of letting the shuttle know that everyone is ready to go. Along the same lines, I wonder how it will work in terms of making sure everyone has a "ticket." Are roboshuttles going to need human attendants in order to do this?)

The cut is much more dramatic than you suspect. Some fraction switch to robotaxis -- they don't park very much at all, preferring to go and pick up another customer. For those who use private robocars, those cars will park more densely than "valet" style, fitting 2.5 to 5 times as many cars per square foot in the lot, unless parking for just a few minutes.

Transit riders (and many car riders) will switch to robotic vanpool transit, also eliminating parking. Some transit riders may switch to taxi service, but they should not cause parking problems.

So yes, if 50% are still parking manual cars, you still need 50% of the parking, but the rest of the people probably fit in only 10-15% more space.

Shuttles will handle rider verification through smartphones, I suspect. No ticket collector needed. I believe they will operate on demand, so you don't so much let the shuttle know that everybody is ready to go, rather, you use the shuttles at a time where it takes only a short time to gather 16 people at the staging lot all going to roughly the same area. Perhaps only a minute or two to gather them.

Smartphones, sure. But how do you make sure there are exactly 16 people on the shuttle, and that those 16 people are all people with e-tickets? Even if you do identify a cheater, what do you do about it?

I guess one adequate answer is to accept that it's possible, and rely on deterrence (especially the possibility that the cheater might find the police waiting for them at the first stop) to keep it to a minimum. Still, this will work better in some locations than others.

Vandalism will likely also be a problem for shuttles operating in some areas, especially if the costs are low.

I will admit that most of my shuttle plans are multi-stage, so you ride a solo car to the shuttle at a staging lot, rather than getting on the shuttle right at your point of origination. You're right, if you are getting on the shuttle directly there is an enforcement question. Since seats are reserved, somebody else is going to say "you're sitting in my seat" and they will have the e-ticket to prove that, but there is no driver or transit cop there to resolve the dispute.

So it's a problem. However, is it a big problem? If you don't have a ticket, you don't even know where the shuttle is going. So just why are you trying to get on and pick a fight with the guy whose seat you took. (Or even if there is an empty seat and you get on just at the end, what is the point?)

The system is going to know you did it, and probably take a picture of you. If you do it a lot (again, why?) they might come after you. If there is a transit cop equivalent around the 7th time you do it, you may get arrested.

And of course the simple solution is that you send the transit cop to the destination of the shuttle. They arrest you when it stops. If you displaced somebody from their seat, send a private car for them, tell them not to fight. I guess you could also let the other passengers decide if they want to ride with the stowaway (which might be creepy) or delay their ride. Needs more thought.

I have to admit I hadn't considered assigned seating. I had envisioned more of a "standing room only" type shuttle.

Generally, phone-summoned travel is reserved seat travel. With UberPool it doesn't assign you a specific seat but counts how many people it puts in the vehicle.

This does not mean you could not have non-assigned-seat shuttles for ad-hoc travel, like from a busy building out to a staging area. But then, we don't really care if people hitch a ride on that too much.

My vision of robotransit though includes different price levels. At all but the cheapest level that means guaranteed nice seats, because people want that. They dislike that rush hour transit often is standing room only. And those would generally be assigned seats. In particular to make transit efficient, you want to match the vehicle size to the passenger load closely, which means if there are 12 people you send a vehicle with not much more than 12 seats.

You can also send standing vehicles and it is efficient -- but also unpleasant.

Makes sense. Another option, which might be more efficient and more idiot-proof, would be to assign seats at the time of boarding: Scan your phone (NFC) and get message on a screen (e.g. "Seat 4A" or "No Reservation: Call 555-867-5309").

It definitely needs to be really simple for people, unless you have a human attendant directing things.

I am definitely imagining the use of most transport in the future to be more like Uber, that people use their phones to summon up a ride. After that, the phone and the servers it talk do do the rest, including seat assignments. The idea of sitting around at a station, waiting for something to pull up, and then getting on it and scanning your ticket card (or phone based ticket system) is the transit way of thinking. One could use the transit approach if it made sense, but I am not sure it makes much sense other than for people who don't have phones. Even if you are standing on the curb and shuttles are pulling up, it still is more efficient to tell your phone you want a ride, and to be assigned one. It lets the system running the shuttles plan more efficiently for where to put everybody.

I didn't mean to exclude the Uber method of asking for a ride using your phone and being assigned one.

Once the car arrives, there's presumably some point at which you tell the car who you are and the car confirms that you're in the right car. There are lots of ways to handle that interaction. Scanning a phone using NFC and having a message pop up on a screen seems like a good way to do that. I guess you could skip scanning the phone and just have the message on the screen say "Are you Brad? [Yes] [No]". Just assuming it's always the right person that got into the vehicle doesn't seem good. I don't know about you, but I've gotten into the wrong Uber vehicle before. Fortunately there was a human driver there to let me know that I screwed up.

Yes, I expect the taxis and vans to have displays on them saying the names of who is supposed to get in. (Or something as simple as the coloured lights that Uber and Lyft use.) And then a bluetooth match-up when you get inside. NFC is also possible. The vehicles will have sensors to know if somebody is in a seat (as cars do already today.)

The harder question was the one about somebody trying to "sneak on." You want to avoid fights, so the simplest thing is probably to get a ride for the displaced person and then have the vehicle do a quick stop next to a police officer somewhere along the route, as long as it does not appear that will put other passengers at risk. Ideally at the end of the ride but then the police have to travel which may not be practical.

However, once it becomes clear that if you sneak on, you 100% get arrested, it won't happen much.

Add new comment