Clifford Whinston on driverless cars

by on July 19, 2012 at 5:06 am in Current Affairs, Economics, Law | Permalink

Here is one good point of many:

Driverless cars don’t need the same wide lanes, which would allow highway authorities to reconfigure roads to allow travel speeds to be raised during peak travel periods. All that is needed would be illuminated lane dividers that can increase the number of lanes available. Driverless cars could take advantage of the extra lane capacity to reduce congestion and delays.

Another design flaw is that highways have been built in terms of width and thickness to accommodate both cars and trucks. The smaller volume of trucks should be handled with one or two wide lanes with a road surface about a foot thick, to withstand trucks’ weight and axle pressure. But the much larger volume of cars—which apply much less axle pressure that damages pavement—need more and narrower lanes that are only a few inches thick.

Building highways that separate cars and trucks by directing them to lanes with the appropriate thickness would save taxpayers a bundle. It would also favor the technology of driverless cars because they would not have to distinguish between cars and trucks and to adjust speeds and positions accordingly.

The full piece is here.

Al Brown July 19, 2012 at 6:38 am

We may be able to squeeze more robotic vehicles in a stretch of road. But robotic vehicles will also shrink in size.

Robotic cabs will become more common without the need to pay a driver. They’ll be cheaper than owning a car, because you’re sharing the capital cost with a bunch of other people.

And cabs will shrink without a need for a front seat or even a steering wheel. And without a human need to accelerate so much, engines and the space needed for them will also shrink.

And before we even get to robotic vehicles, traffic flow can improve in other ways. We can start publishing traffic light schedules so cars can adjust their speeds to avoid stopping and starting. That is, what the schedule for a traffic light will be for the next 5 minutes. If cars and signals start talking to each other, we can save tons of fuel and time.

In multiple ways, travel by car is going to get a lot more efficient.

Marian Kechlibar July 19, 2012 at 7:28 am

I, for one, would love to be a member of an automatic car company, if it meant that I could try out various cars at will.

“Hey, it’s Monday evening, let’s get a BMW…”

“Today, I feel like saving money – call me a Kia!”

“I need to move a piece of furniture – send me a small Volkswagen truck.”

Etc. etc.

Sigivald July 19, 2012 at 7:06 pm

What makes you think that there’d be much variety?

The BMW is mostly uninteresting if you’re A) not driving it yourself and also B) not showing off how rich you are by driving it, no?

I suspect it’d be much like ZipCar or the like now – a few options of sizes, and that’s it.

(Not only the above but also the logistics of running the company suggest that; who wants to deal with BMW and Kia maintenance differences and the like? Buy a fleet of “X compact”, “Y sedan”, “Z minivan”, “A smallish cargo vehicle” and go.

Luxury renters aren’t even your market if you’re that company; they’ll have another company entirely, just like Hertz won’t rent you a 760iL – they don’t even have one*.

* They actually do, but only at “selected airports in Europe” via the Prestige collection – and likewise Big Local AutoCar Rental isn’t going to bother with a BMW, because who’d bother renting it? For practical purposes you can’t get a luxury car from a car rental company without going to a specialist.

The logic is even stronger for a car you’re not driving, because the notional extra performance? You won’t get any access to it, now, will you? They’ll give you a Kia with cushier seats.)

Mark Thorson July 19, 2012 at 9:45 am

Yes, the cars and lanes will shrink. It will be like digital TV. In the demos, only show video that takes up the full bandwidth of the channel. But when it comes time to implement the system, knock the bandwidth down to the minimum so more channels can be squeezed into the same spectrum. Driverless cars will be like that. It will be like riding in an egg packed in a carton full of eggs. It will not be like riding on today’s wide roads in today’s (relatively) sparse traffic.

Matthew N (from Driverless Car HQ) July 19, 2012 at 10:21 am

Google “LIt Motorcycle”. It presents a great vision as to what one passenger transportation will probably look like.

The Original D July 19, 2012 at 9:13 pm

Whoa, that’s cool. It would be nice if it could seat two, one behind the other like a regular motorcycle.

Mark Thorson July 19, 2012 at 11:36 pm

Nope, the driverless car doesn’t need windows. Windows are heavy, expensive, and compromise the safety of the unibody. If people want to see the outside, they can look at their screens, but most people won’t want to. The computers can drive the car uncomfortably close to other vehicles, and you don’t want to know just how close that is while it’s happening.

Also, passengers in the driverless car won’t face forward. They’re safer in most collision scenarios facing the rear.

Danny July 19, 2012 at 10:44 am

It is true that they will get more efficient, but there is still an upper bound to that efficiency, especially space efficiency. Driving closer together and having smaller cars will make a difference on highways, but almost no difference at all on city streets, where road capacity is determined more by vehicle size than reaction times. Your extra foot of road space from a slightly smaller vehicle would mean, at most, an extra car or two per block. The biggest benefit to cities, which aren’t going away any time soon, would be a drop in the cost of taxis, which could provide more point to point mobility without adding parking costs. But even taxis have their limitations of efficiency…namely the location problem. People tend to move in similar directions, which means taxis spend an inordinate amount of time relocating with an empty cab.

Other than that, high capacity transportation needs will continue to be skewed towards mass transportation: planes, trains, buses, etc.

Major July 19, 2012 at 8:46 pm

It is true that they will get more efficient, but there is still an upper bound to that efficiency, especially space efficiency. Driving closer together and having smaller cars will make a difference on highways, but almost no difference at all on city streets, where road capacity is determined more by vehicle size than reaction times. Your extra foot of road space from a slightly smaller vehicle would mean, at most, an extra car or two per block.

This is nonsense. Even at just 20 mph, the average human reaction time equates to 20 feet of travel. So even on congested city streets, cars frequently need to be separated by tens of feet at a minimum to provide safe stopping distances. That requirement would be greatly reduced in driverless cars. And that’s just one of the ways in which driverless cars will increase road capacity. They’ll also increase capacity by reducing the rate of accidents and reducing bad human driving practises (tailgating, sudden lane changes, sudden changes of speed, etc.). Accidents and bad driving are major causes of disruptions to traffic flow.

Other than that, high capacity transportation needs will continue to be skewed towards mass transportation: planes, trains, buses, etc.

No one is going to bother walking to bus stops and train stations and waiting around for buses and trains when for the same price or less they can get a faster, more comfortable, more private, on-demand, door-to-door ride in a driverless taxi. Driverless taxis will decimate the market for mass transit.

Jim July 19, 2012 at 7:05 am

Perhaps a more interesting question is what will happen to car travel in cities. If driverless cars are programmed to stop automatically (or at least slow dramatically), then pedestrians may feel a lot more ready to jaywalk. Currently you know that if you cross the road in traffic you might get killed. If you know you won’t, wouldn’t you do it all the time? In crowded cities with a lot of pedestrians, driverless cars might be extremely slow.

Some cities might react by harshly punishing jaywalking so that robocars can speed their way through. We may see complete separation of robocar routes from normal traffic (elevated robo-roads, maybe?) Other cities might welcome the smoother flow of pedestrians (and cyclists too, when people realise it has become much safer to ride a bike). Or perhaps nobody will use driverless cars in city centres – but if they are used to not driving themselves around anymore on inter-city roads, why would anybody want to do it on intra-city roads?

jmo July 19, 2012 at 9:13 am

Couple the hi res optical sensors and face recognition technology the car would send the imagine automatically to the authorities who would mail the perp a ticket for $50 or $100.

jmo July 19, 2012 at 9:14 am

Sorry that should read:

Couple the hi res optical sensors presumably required in a self driving car and face recognition technology

doctorpat July 20, 2012 at 1:16 am

For every month your fines are overdue, the robot cars are allowed to actually hit you once.

Bónapart Ó Cúnasa July 19, 2012 at 7:32 am

Call them “personal buses”, not “driverless cars” – much more effective way to bring the enviromental lobby round in support, at least on my side of the Atlantic where greens have a knee-jerk negative reaction to anything called a car…

Matthew N (from Driverless Car HQ) July 19, 2012 at 10:20 am

This is actually quite smart…. not to mention cheeky. I love it.

doctorpat July 20, 2012 at 1:18 am

But non-greens have a knee-jerk negative reaction to anything called a bus…

Track-free private railway carriages?

Bill Harshaw July 19, 2012 at 8:05 am

Re: “Building highways….” Ridiculous, unless one specifies a time frame. Consider how long it takes our stock of vehicles to turn over, consider how long it took for the Prius to become popular, consider that we’re still using 19th century tunnels on our east coast railroads. We’re talking decades and quarter centuries before this becomes realized.

Greg G July 19, 2012 at 8:55 am

The psychological obstacles to driverless cars may prove more of a problem than the technological obstacles. When people start getting killed because driverless cars occasionally malfunction, many people will want to outlaw them (and award huge damage settlements) even though they might have been less outraged by the same or greater carnage being caused by idiotic human drivers.

jmo July 19, 2012 at 9:19 am

many people will want to outlaw them

But not more than half of the people. If the fatality rate for driverless cars is 0.01 per 100 million miles traveled and the fatality rate for human piloted cars is 1.1 then that is just something you can’t argue with.

Craig July 19, 2012 at 9:56 am

Well, you CAN argue with it, and some people certainly will…you just can’t WIN the argument. Self-driving cars will be mandatory on the Interstate highway system within my lifetime. Our grandchildren will look back with astonishment at the idea that we let human beings drive cars at anything faster than a jogging pace.

anon July 19, 2012 at 10:08 am

What about RVs?

Seems like there is a lot lower hanging fruit than driverless cars – like driverless trains and driverless intercity freight trucks.

Matthew N (from Driverless Car HQ) July 19, 2012 at 10:23 am

@Anon – completely agree but for whatever reason the main focus is in the consumer space right now.

It seems like a great way to test the technology without passengers.

Danny July 19, 2012 at 10:56 am

Well, in NYC, you have union contracts that call for 3 unionized conductors on a commuter train, whereas the total number of conductors/drivers needed is zero, as evidenced by cities like Copenhagen. Freight trains aren’t much better. Driverless trucks are the exception, as they are mostly deunionized, but then again, labor costs are now being dwarfed by other costs. If you can increase MPG by 1mpg for a long haul truck, you can pay for almost 6 drivers.

Marian Kechlibar July 19, 2012 at 11:57 am

It is not just driver costs, it is also time. Human drivers need to take breaks, and, at least in Europe, police enforces this.

Fully robotic trucks could run nonstop, and they could probably run a bit faster. The law may trust a human with a truck up to 90 kph, but robot could reasonably do 130.

prognostication July 19, 2012 at 1:59 pm

A failure of the automated operating system on DC Metro in 2009 caused the deadliest crash in the system’s history. FWIW.

mbutu o malley July 19, 2012 at 4:38 pm

@Marian

In the US we already have speed limits of ~120 KPH through large parts of the country, a fully loaded tractor trailer has trouble maintaining this speed due to physical limitations. There would likely have to be substantial redesigns to the trailers and rigs and changes to the planning of loads to allow higher speeds.

Major July 19, 2012 at 8:54 pm

I’m sure there will be driverless freight trucks, and probably a few driverless buses and driverless trains (there already are some driverless trains). For passenger transportation, the overwhelming benefit of driverless operation will come from applying it to cars, not mass transit vehicles.

anon July 19, 2012 at 9:39 am

Quick correction – It’s Cliff Winston, not Whinston

libert July 19, 2012 at 9:40 am

It’s Clifford Winston, not Whinston

Affe July 19, 2012 at 10:35 am

He’s thinking of roads like a Californian. Once the (robotically-driven, no doubt) salt spreaders and temperature shifts on the east coast or in the midwest crack those car-bearing eggshells he envisions, they’ll be back to thick, juicy slabs of sizzling asphalt in no time.

Go Kings, Go! July 19, 2012 at 12:08 pm

That’s funny because I assumed he was ignoring California roads. Here automotorists share lanes with motorcycles so you can’t just shrink lanes. And if you think bureaucrats will just crush our meaningless preferences, you haven’t met the California Highway Patrolman’s union.

Highgamma July 19, 2012 at 10:53 am

Along the lines of Affe’s comment, the driverless car technology that requires the least amount of change of physical infrastructure will win.

Matt Young July 19, 2012 at 1:38 pm

Naw. The appeal of robotic is extreme speeds and large numbers of passengers.

Once we see a large robot that can carry 100 passengers safely at 120 MPH, then there will be a rush to build separated, high speed robotic guideways. Cars will go away, everyone will want to ride the huge robots.

John Mansfield July 19, 2012 at 2:04 pm

Will these driverless cars be one big, interlinked machine working together harmoniously under central control (whether a big computer downtown or common design requirements)? Or will they be be chess computers on wheels each angling for the way to beat the others? Would either option be preferrable?

Cascadian July 19, 2012 at 3:03 pm

I think people are really stuck on the existing transportation options when they think about driverless cars.

I don’t think most roads will get faster, because most roads are in urban areas where high speeds impose costs that people on downtown streets won’t accept. You have pedestrians and cyclists in high numbers in cities, and high speeds aren’t compatible with them. And if people aren’t driving, but are being chauffeured by automatic cars, they’ll think more like pedestrians, so the future trend if anything is away from development thats cater to the driving mentality.

I’m not sure that suburban or rural roads would necessarily get faster, either. Lots of cars coming in and out of driveways and side roads, more difficult terrain, more varied road surfaces, and more slow-moving vehicles like farm equipment mean that there’s less opportunity for speed increases than you might think. Really the only places where automatic cars would add to average speed significantly would be freeways and rural highways. Those represent a lot of mileage but a very small percentage of total trips (a few percent). Cross-country drives would be faster, as would commutes.

But in the case of commutes, capacity still vastly favors transit for the highest traveled corridors, particularly with automation. Buses without dedicated guideways are considered unreliable because they get stuck in traffic. But with automated driving that’s less likely to happen, and buses on freeways become time-competitive with trains, but with more flexibility at the end of each line where ridership density becomes more diffuse. I would expect automated buses to win out over automated cars for commutes, especially with energy costs high for personal vehicles and everyone habituated to spending vehicle time not driving whether or not they use a personal vehicle or a public vehicle.

I suspect that automation will spell the beginning of the end of mass personal vehicle ownership except for the wealthy. Most people will purchase automated car services for trips that aren’t viable on automated public transit. There will be fleets of private automated cars that you can summon with a gesture that is picked up by your personal mobile computer network and relayed instantly to the internet. Think of ZipCar but with automated fleets that have no need to park on city streets. It will make more sense for long-distance trips between cities to take the AutoZipCar to the high speed rail station than to drive the whole way in an automated car, though road trips to more rural destinations will be served by autocar rental agencies.

The trillion-dollar parking infrastructure (in the US alone) will become almost entirely superfluous. Many estimates suggest that there are three spaces for every car. With automated cars, almost that entire 2/3 extra parking becomes superfluous immediately, and with better automated public transit even more gets freed up. So city streets will be mostly empty of cars except those dropping people off. Cities will have to redesign streets to create buffers between pedestrians and autocars. Probably trees and street-side vendors will take on the calming role once played by parked cars. Suburbs will be even more radically transformed as fields of strip mall and shopping mall parking can be converted to new development, sidewalks, or open space.

All of the productive capacity and personal income devoted to cars will be redirected to more productive sectors, creating an economic boom that more than offsets the decline in the volume of personal car sales.

All of that is much more interesting than the vision that everything will stay the same but people will drive faster on smaller lanes.

Major July 19, 2012 at 9:20 pm

“And if people aren’t driving, but are being chauffeured by automatic cars, they’ll think more like pedestrians, so the future trend if anything is away from development thats cater to the driving mentality.”

Huh? If more people are riding in cars, then more people will be thinking like car passengers, not pedestrians. Pedestrians will become even more marginalized.

“But in the case of commutes, capacity still vastly favors transit for the highest traveled corridors, particularly with automation.”

Commutes are only a small share of total urban trips, and highly-traveled corridors are only a small share of commutes. And mass transit will struggle even for those trips. If the capacity of Manhattan’s streets is doubled by driverless vehicles, a huge number of additional Manhattan commuters will be able to use cars instead of having to rely on buses and trains. So demand for buses and trains will fall even at peak travel times. And at off-peak times, when there is lots of spare road capacity, demand for buses and trains will fall even more. So even if really congested places like Manhattan, mass transit is going to have a very hard time competing with driverless cars. And in most places at most times, where there is already plenty of road capacity and mass transit already has just a tiny share of the market, transit will be wiped out.

“I suspect that automation will spell the beginning of the end of mass personal vehicle ownership except for the wealthy.”

Since mass personal vehicle ownership is already affordable to the general population (that’s why it’s “mass”), and driverless cars will make it even more affordable, this seems very unlikely.

Major July 19, 2012 at 10:25 pm

“Cities will have to redesign streets to create buffers between pedestrians and autocars. Probably trees and street-side vendors will take on the calming role once played by parked cars. “

Since driverless cars will be less likely to hit pedestrians than human-driven cars are today, the need for “buffers” between cars and pedestrians will be reduced, not increased. There will also be fewer pedestrians, since everyone will be able to use a driverless car to get where they want to go instead of having to walk to a transit stop or walk the whole way.

“Suburbs will be even more radically transformed as fields of strip mall and shopping mall parking can be converted to new development, sidewalks, or open space”

Open space. Driverless cars will increase average travel speeds. That means people will be able to travel a greater distance in a given period of time. That reduces the incentive to build things close together. Density will probably decline, just as it did after motorized urban transportation first became widespread (early trams and streetcars promoted the growth of early suburbs), and then again as conventional cars became widespread (further suburbanization and sprawl).

Sigivald July 19, 2012 at 6:59 pm

All that is needed would be illuminated lane dividers that can increase the number of lanes available. Driverless cars could take advantage of the extra lane capacity to reduce congestion and delays.

Cost a fortune – and of course that would mean mandating only driverless cars of a standardized size.

One also wonders what happens when there’s a blowout or a deer in the road, with the greatly reduced space margin. (Or, if we run them really tight, what happens to the ability to exit cars during a traffic jam or breakdown…)

Frankly, I suspect it’s cheaper to add lanes than to redo everything with “shrinkable” lit lanes and get the controls right and reliable. Funny thing about paint – it doesn’t just fail because a trunk of wiring got cut or power went out or due to a software error; having seen how very simple and short-wiring traffic lights fail, and how electronic road information signs fail I am not sanguine about that idea being wise.

Steve Sailer July 19, 2012 at 7:20 pm

“Driverless cars don’t need the same wide lanes, which would allow highway authorities to reconfigure roads ”

I went to an auto show a decade ago where a father-son team of inventors was displaying their home-made two-seat car that mounted one seat behind the other. The car was only half as wide as the average subcompact. They pointed out that freeways could be reconfigured with twice as many lanes.

For some reason, this reconfiguring of roads hasn’t happened yet, even though, unlike with robot cars, the technology for narrower cars has existed for a long time.

Toby Penington August 2, 2012 at 9:57 pm

Are driverless cars the future of driving? Well I guess so but is this possible? I know it will definitely end the traffic woes and parking scarcity we experienced. Equipped with high-end techniques and probably some new auto parts and accessories installed like radar, GPS and computer vision, this autonomous vehicle can sense its environment and navigate on its own without bumping into other vehicles or obstacles. With this I don’t think cars can be necessarily driverless. It can still have a driver for as long as the car is wifi-enabled and have those technology installed.

Comments on this entry are closed.

Previous post:

Next post: