Will driverless cars fix Houston’s traffic problems?

Tory Gattis thinks they might.

The second step is understanding the ramifications of coming new technologies — specifically self-driving cars. While the general vehicle fleet will take decades to turn over as people slowly replace their cars, we can expect extremely rapid adoption among taxi services as soon as these vehicles are available in the early 2020s. The economics are simply too compelling: Almost 80 percent of the cost of a ride is the driver. One estimate has the typical ride dropping to $3.25, with shared rides going for $2.43 or even as low as $1 with SUVs carrying up to six passengers at once along a shared route.

Customized SUVs could be made with private individual compartments, so that passengers traveling in generally the same direction could share a ride without interacting. When vehicle pulls up, an indicator could tell you which door to enter for your compartment, then alert you again when it’s time for you to get out based on the destination you put into your smart phone. A private ride combined with shared prices and efficiency: the best of both worlds.

The impact on traffic congestion could be dramatic, as fewer vehicles carry more riders. Analysis by MIT, Stanford, and others estimate that shared rides could reduce the number of vehicles needed to carry the same number of trips by 70 to 90 percent. Quite the silver bullet to reduce traffic congestion! Then there’s the icing on the cake: Automated drivers are expected to dramatically reduce crash injuries and space required for parking, which will free up a tremendous amount of much-needed land in our cities.

All indications are that these super-cheap, point-to-point autonomous taxi services will essentially replace most bus and rail transit: Most trips would be much faster and more direct at nearly the same cost. In fact, transit agencies like METRO may switch their fleets to such vehicles, providing better service to their customers. Helsinki’s transit agency is already a pioneer of this transition, offering on-demand mini-vans available via smart phone app.

Gattis goes on the describe Managed eXpress Lanes — MaX Lanes, for short – as a way to efficiently ferry around all those multi-occupant driverless SUVs. I have no doubt that driverless cars are coming and that they will change how we do cars – indeed, I have good reason to believe that. What I’m less sure about is the effect it will have on how people behave. As cheap as this form of “mass” transportation may be, people aren’t just motivated by price. Status, comfort, convenience, luxury, self-expression – these are all things people look for in their mode of transportation, and there’s no reason to think that will be any different with driverless cars.

In a utopian world, as my colleague Brian Fung writes, driverless cars would make everything more efficient. You’d never have to waste time looking for parking. Cities, in fact, could get rid of it. Vehicles that today sit empty most of their lives could be put to maximum use instead, transporting one passenger after another.

This ideal scenario, though, assumes some kind of all-knowing central dispatcher: a company, or service, that would distribute cars to serve the most people the most effectively, with an omniscient eye on the entire network. And, as more companies dive into this space — General Motors and Lyft announced an eye-popping new partnership Monday — it’s tempting to think we’re witnessing the start of an epic battle for the coming autonomous monopoly.

Who will get there first, winner-take-all? General Motors working with the ride-hailing startup Lyft? Or Ford, as it’s rumored, teaming with Google? Toyota? Or Uber, which seems to think it doesn’t need a traditional automaker ally at all?

“Uber, their whole goal is to minimize the time from request to pickup, and to do that means you have to have a lot of vehicles,” says Dave King, an assistant professor of urban planning at Columbia University. “And if they’re the monopolists, maybe it’s close to being efficient. But they’re not going to be the monopolists.”

None of these companies will, he predicts.

That’s because autonomous cars will require the same market segmentation we already have today (whether consumers want to own these vehicles or share them or use some hybrid service in between).

“If you think the rich people on the Upper East Side of Manhattan are going to get into just a lowly old car, I don’t see that happening,” King says. They’ll want the Audi of autonomous cars, while you may gladly hitch a ride in the Kia equivalent. And even if I’m not truly driving a sports car myself, I may want to ride in something that feels sporty.

“The idea that everybody wants the same experience for personal travel is strange to me, because nobody’s ever wanted that,” King says. “We don’t buy the same bicycle, we don’t buy the same model of car. Some people like the bus, some people like the train.”

Sure, a ride in that automated Uber SUV may be cheap, but I’d bet a lot of people would be willing to pay extra to not have to share the ride. I mean, unless you’re the last person picked up and the first person dropped off, a part of your ride will be spent going to and from other people’s pickups and dropoffs, and who wants to spend all that excess time in a car? Note that this is not the case with plain ol’ mass transit – you get on and off at your stops, with no extra side trips. Also, maybe you like listening to your tunes through the car’s stereo system instead of earbuds. Maybe you just don’t want to be in a car with other people, even if you can be hermetically sealed from them; I confess, this “SUV with individual compartments” idea conjures for me images of being packed into a carton like eggs, or slotted like a LEGO action figure into a car built with the iconic bricks.

And while the egg carton individually-compartmented SUV might be a great deal for daily commuters who travel on predictable schedules, the fact remains that a fair number of our driving trips are spontaneous and unplanned – “Hey, let’s go out to dinner tonight”, “Sure, I’d love to meet you for a movie that’s starting in 30 minutes”, “Dad, I need to go to Jenny’s to work on our science project”, “He’s got a fever and he’s barfing, we need to get him to a doctor”. How much tolerance will people have to wait for a robot car to pick them up? (Side question: How many of these driverless Ubermobiles are going to have to be on standby for this, and where will they be when they’re not in use? Idle cars do not maximize profits, and as with commuting people may not want to make three or four other stops before they get where they want to go, especially if they’re in a hurry.) It’s not like car manufacturers are going to stop marketing to individual owners (*), after all.

And if owning an automated car is as common and pervasive as owning a car is today, then we may not get any reduction in traffic at all.

We’re starting to see professional reports echoing long-standing concerns about how driverless cars will affect our cities. This new one from KPMG, in particular, is getting a lot of press.  It’s actually a focus group study about the transport desires of different generations, but it confirms the thought experiments that many of us have already been laying out for a while.

Much depends on whether these cars are owned or spontaneously hired like taxis, Uber, and Lyft.  A taxi model is definitely better in its congestion impacts, but that doesn’t mean it will happen.  The ownership model is closer to the status quo, and the status quo always has enormous power.  Driverless taxis will not always be available on demand, especially in suburban and rural areas, so a legitimate fear of being stranded will make people in those areas prefer the security of having a car just for them. And of course, that’s just the effect of rational concerns about relying on taxis.  Less rational desires for car ownership, as an expression of identity or symbol of liberty, will also not vanish overnight.

This leads to a nightmare scenario that University of Washington’s Mark Hollenbeck laid out in our recent Seattle Times panel.  Paraphrasing Mark:  A suburban father rides his driverless car to work, maybe dropping his daughter off at school. But rather than park the car downtown, he simply tells it to drive back home to his house in the suburbs.  During the day, it runs some other errands for his family.  At 3 pm, it goes to the school to bring his daughter home or chauffeur her to after-school activities.  Then it’s time for it to drive back into the city to pick up Dad from work.  But then, on a lark, Dad decides to go shopping at a downtown department store after work, so he tells his car to just circle the block for an hour while he shops, before finally hailing it to go home.

This is really easy and obvious behavior for a driverless car owner.  It reduces the number of cars someone needs to own, and reduces pressure on inner city parking, but would cause an explosive growth in vehicle trips, and thus in congestion (not to mention emissions and other impacts).  Just the commute behavior doubles car volumes, because the car now makes a two-way trip for each direction of the commute, instead of just one.  And if everyone shopping downtown has a car circling the block waiting for them, well, that level of congestion will far exceed what’s generated by cars circling for parking today.  It could pretty well shut down the city.

And while increased safety is touted as the single biggest boon from driverless cars, there’s reason to fret about that as well.

For the past five years, my collaborators and I in the Vision Sciences Lab at Harvard University have been exploring the differences in capabilities between people and today’s best AIs. My studies have focused on simple tasks, like detecting a face in a still image, where AIs have become reasonably skilled. But I have become increasingly unsettled by the implications of our research for very challenging AI tasks. I am especially concerned by the implications for the extremely challenging task of driving a car. Self-driving cars have enormous promise. The improvements to traffic, safety, and the mobility of the elderly could be dramatic. But no matter how capable the AI, humans just behave differently.

[…]

The biggest difference in capability between self-driving cars and humans is likely to be theory of mind. Researchers like professor Felix Warneken at Harvard have shown that even very young children have exquisitely tuned senses for the intentions and goals of other people. Warneken and others have argued this is the core of uniquely human intelligence.

Researchers are working to build robots that can mimic our social intelligence. Companies like Emotient and Affectiva currently offer software with some ability to read emotions on faces. But so far no software remotely approaches the ability of humans to constantly and effortlessly guess what other people want to do. The human driving down that narrow street may say to herself “none of these oncoming cars are will let me go unless I’m a little bit pushy” and then act on that instinct, but behaving that way will be one of the greatest challenges of making human-like AI.

The ability to judge intention and respond accordingly is also central to driving. From determining whether a pedestrian is going to jaywalk to slowing down and avoiding a driver who seems drunk or tired, we do it constantly while behind the wheel. Self-driving cars can’t do this now. They likely won’t be able to do it for years. But this isn’t just about routine-but-confusing interactions like that between the Google self-driving car and the Mountain View bus.

Even the best AIs are easy to fool. State-of-the-art object recognition systems can be tricked into thinking a picture of an orange is really an ostrich. Self-driving cars will be no different. They will make errors—which is not so bad on the face of it, as long as they make fewer than humans. But the kinds of errors they make will be errors a human would never make. They will mistake a garbage bag for a running pedestrian. They will mistake a cloud for a truck.

In other words, we may be farther away from achieving the kind of AI needed to make self-driving cabs and buses feasible than we now think.

Of course, none of this may happen. Ultimately, we’re all just guessing. All I’m saying is that we should not rule out the human factor when gaming out how driverless cars will change how we live. We tend to adopt new technology to suit our needs, including the needs we didn’t know we had. Maybe a larger percentage of people in the future will forego buying cars and exist solely on the various forms of mass transit than in the present, but it’s not clear to me that that will happen. We’re a car-centric culture that loves our automobiles and needs to be prioritized, right? I know I’ve heard that argument before. I hope I live long enough to see if that remains the case.

(*) Unless of course the Big Government of the future forbids it. If you listen closely, you can hear Ted Cruz’s head exploding.

Related Posts:

• Help Metro figure out its Regional Transit Plan
• And now a few words from our city transportation planner
• Goodbye, Greenlink