A spate of recent stories and opinion pieces have expressed concern over the following issues:

  1. Doing too much “deadheading,” travel with no passengers on board.
  2. Deliberate deadheading, “circling,” just to avoid parking.
  3. Using a city’s precious curbs “for free” and hogging them.
  4. Filling up public EV charging stations.

In some cases, concern arises because these are robots, not people, or that they are owned by corporations. Almost everything they do, however, is in the service of their riders, who are generally local residents, like all the other people using the roads and curbs. Let’s look at all these issues. Most of these issues have come up first with Waymo, since they have effectively the only decent sized robotaxi service, but they apply to all companies.

Data reported by the CPUC indicates that Waymo vehicles are empty around half the time, more than people expect. Traditional taxis in New York City have no passengers 40% to 60% of their miles, depending on which source you read. It’s better in dense Manhattan than the outer reaches of the city. You should expect robotaxis (and Ubers) to do better because they don’t need to drive around to find their next fare, but that’s not yet the case, but the numbers are still not too different. (Today’s robotaxis have depots located some distance from the busiest areas, which may account for this.) Waymo’s deadheading number has been improving over time, and it’s expected to improve more as the fleet size grows and ride volume grows. It has currently stabilized at around 42% empty.

The primary empty miles come from going to pick up a passenger and, if no fare is assigned, moving towards zones where passengers are expected to be. That allows serving them with shorter wait times, which is essential. There are also empty miles to and from depots, charging stations and other service locations. Ubers and taxis also need this, though the driver cleans the vehicle and does not need to travel somewhere to do it, usually.

That taxis (robot or otherwise) need to deadhead is an expected property, not a flaw. It’s inherent in having vehicles which serve multiple riders with one vehicle. Private cars serve only one driver, and never deadhead, but that means they just sit parked all day, rather than serving others. All that parking uses up a lot of land in the city. A private car is in use only 5-10% of the day, parked the rest. A taxi will be in use perhaps 50% of the day. That’s a big win for land resources, but it comes with the need for deadheading. You can’t get one without the other. You can reduce deadheading by increasing the fleet size, so that trips between rides are shorter. But that means needing more parking.

Yes, vehicles will also be idle a lot of the day, that’s another thing you can’t avoid. You need enough vehicles in your fleet to serve rush hours, so there is almost no way to avoid having much of the fleet idle during less busy times, especially at night. I say “almost” because if the vehicles have 4-5 seats, as they often do, you can in theory do much more ride-pooling at rush hours, with more solo rides off-peak. That can greatly reduce the gap between peak and off-peak usage, and thus allow a smaller fleet and less idle time. Idle time’s cost is essentially parking and a bit of capital, and parking is much less than you think it is.

From time to time, people will imagine that a robot, not wanting to pay for parking, would just circle around, clogging the streets, until it gets another fare. This is different from the vehicle doing “predictive positioning," which all taxis will do. The reality is this is very unlikely. Cars have a cost to run per hour that’s pretty high. Gasoline cars probably cost about $20/hour. Electric cars are cheaper, but robotaxi operations have a cost, and it’s unlikely we will ever see their per hour operating cost get so low that it’s cheaper than parking. That’s true even for today’s cost (which is about $5/hour if you don’t need to park in the most expensive lots in town right next to the premium office buildings) and definitely not true compared to the real future cost of parking which will be less than $1/hour (for robotaxis, which can park much, much cheaper than human driven cars.)

Waymo has said they will not circle . It’s not out of the question some companies would in their early days, because they are still testing and learning, and they do that by having the vehicles out there driving, encountering problems. If they get this value from driving they might circle just to make use of this asset. This is not something they will do long term.

If a company does start circling, it’s a pretty easy thing for a city to bad. Robotaxis are much easier to enforce laws upon. You catch one doing it and you can make life very bad for the robotaxi company for deliberately breaking the rule. It’s a hard thing to hide.

One thing which might cause vehicles to circle is advertising. There are already human driven vehicles that circle in certain places to display ads. That is antisocial, and it could make sense to ban it, or at least doing it in any zone with traffic. It’s often said that the roads should be for people, not cars. But of course the taxis are serving the people who ride in them, and they are as much people as anybody else. Ads, however, are not serving people.

Essential to taxi service is what we call PuDo--Pick Up and Drop Off. PuDo is for people, and those people tend to want to do it in convenient places, which are often on the curb. Private cars, taxis, ride-hails and robotaxis all want to do PuDo, but always in the service of a human.

The curb is also often used for parking and standing/waiting. It may be metered parking, or free parking or even non-parking spaces that allow standing and waiting as long as the “driver” is with the vehicle. There have been suggestions by some that robotaxis should not get to use these spaces to wait or stage.

This was partly in response to a study that examined the consequences of restricting access to the curb for robotaxis, and the answer was you’ll get a lot more deadheading .

One principle violated here is that all cars should be treated the same. If a private car can use a space, so should a taxi. If a taxi can use it, so should a robotaxi be able to. There would need to be a compelling reason to treat the robotaxis differently from other cars.

Now, and Donald Shoup argued in “The High Cost of Free Parking,” ideally none of this space should be free. But it should not be overly expensive either, nor granted to the people in private cars but not to the people riding in ride-hails robotaxis. One strong reason for that is that whatever you think of them now, robotaxis are likely to be the best actors on the road. It’s very difficult to enforce rules on private drivers or even Uber drivers. Robots, on the other hand can be programmed to obey rules, as long as you have authority over the corporations running them. They absolutely will obey. The robots obey because they are robots, and the corporations, though they have great lobby and legal power, also have much more to lose if they face enforcement actions. The robotaxis will, over time, become by far the best actors on the roads, and you don’t want to punish the best actors.

Airports have not figured that out. While they say they charge TNCs an robotaxis a fee to come to the airport in order to manage traffic, their real motive is revenue. Either way, it’s a mistake. Robotaxis could promise that every car that brings somebody to the airport will take somebody else out of the airport if anybody is waiting. That means half the traffic, since all the private cars (that don’t pay) only take somebody one way. The airports punish the vehicles that can be the best actors.

One ironic problem is that while it would be extremely simple to charge robotaxis a fee to use the curbs for PuDo and waiting, it’s pretty hard to charge private drivers. Metered parking tries to charge private drivers for long term stays, but it’s not everywhere, and enforcement and compliance are not universal. Robotaxis would be again, the best actors, never not paying.

All robotaxis today are electric vehicles, which should be celebrated. Not all drivers are willing to buy an EV for their personal car, but nobody finds any downsides to an electric powertrain in their taxi, and the fleet owners, and society, get lots of upsides. Because charging takes more time than fossil fill-up, robotaxis like to charge when they are waiting. They would like to charge close to where they are needed--to avoid deadheading back more distant chargers, like their depots.

But when they do use more local chargers (which today they can only do if there’s a human around, such as a human safety driver in a test fleet) it turns out they are a large chunk of the EV population in a downtown, doing a lot more miles per day. So they are going to be seen in numbers at the chargers. If you get to a charger and see most or all of the stalls are filled with some fleet, robotaxi or otherwise, you’re going to be annoyed by the wait. You’re going to be particularly annoyed at the fleet that got there ahead of you.

It’s not clear how to quell that emotional reaction. Fleet vehicles have just as much right to be customers of chargers as anybody else. If there’s more EV charging demand, that’s a good thing, not a bad one, but it does mean the city needs more places to charge. It will get that, but slowly. The fleet does have the ability to build its own depots, but that’s actually a very slow and expensive process--it can take a year to get the electrical service in place. But if we force the vehicles to use only their own depots, that means they have to deadhead more, and this time not to serve any human, other than the human that wants that charging stall.

Of course, public charging should not exist in cities, but gasoline thinking drives people who buy an EV without a place to charge it at home or work to rely on public fast chargers. Unlike humans, many robotaxis will drive all day, using up more than is in the battery, so they need a charge.

At the same time, it’s better if robots don’t make humans wait, even if their owner has the same right as anybody. So they should perhaps go to more remote charging, even though it means more deadheading and more cost. It’s courteous, but it’ a courtesy no human would do, so who should pay for it?

The world we want has charging all over, and nobody has to deadhead or detour too far to get it. Fortunately, since robocars need them, we’ll see “full serve” stations where a robot or human plugs a car in. This mean that private cars with limited self-drive ability will be able to use them, as well as robotaxis. Visits will be confirmed before departing, so there will be no lines or waiting. Your car will just always been full, as if by magic, and you won’t care who is using what charger. We don’t have that day yet, but the robotaxis are hastening its arrival.

Streets and Resources are for People

Earlier, I noted that it is often written that our streets and their resources are for people, not cars, and that’s true. But that’s all sorts of people, including people on foot, people on bikes, people in transit, people in cars, people in robotaxis, and cargo on its way to people in trucks and robots. All of these things are to serve people. Even deadheading with nobody in the car is trying to serve the next person who needs a ride. The true question is whether something is consuming an inordinately larger share of resources per person. When that happens, the resources should not be free (for anybody) but they should also be fairly priced. It is not just a matter of “if it’s robots doing it, it’s not for people so it should be stopped.” More nuance is required, particularly when there are other public benefits to a technology.