Street Smarts

Bad news: with all due respect to Terrafugia, unless you’re a fan of Futurama it’s probably going to be awhile before you see a flying car. But cars that drive themselves are coming, probably within most of our lifetimes and possibly sooner than you might think. They will drastically cut traffic congestion, improve safety, and be a terrific boon to those like the young and the old who are deprived of mobility. The ability to take our hands off the wheel will also undoubtedly send sales of Big Macs and mascara skyrocketing. But do we have the drive to make robot cars a reality?

Technologically, we’re closer than you might think. All of the elements needed to make driverless cars – radar, automatic pilot software, computing power, wireless communications and, of course, navigation systems that know where you’re going (okay, usually know where you’re going) — are technologically feasible, and in many cases are even available commercially.

Way back in 1997, California’s Partners for Advanced Transit and Highways (PATH) ran a platoon of self-driving autos, separated by mere single car lengths, down a highway at 60 mph.

In the 2007 DARPA challenge, several computer-controlled vehicles successfully navigated a simulated urban course that included moving cars, traffic signs, buildings, and pedestrians, this time without any direct aid from a smart street system.

And all this progress has come without any overarching entity directing the development of the technology. But if we’re going to make driverless cars a reality, that may soon have to change.

First, the cars in a driverless system will ideally be able to talk to each other. This may not be totally necessary; after all, the current system basically works with very little explicit communication between drivers except for the occasional horn honk or gesture whose meaning cannot be reproduced here. But assuming technical hurdles (like stopping hurdles) can be surmounted, it would be nice if cars could communicate and cooperate with each other, to improve reaction times and smooth maneuvers like lane changes.

Just as importantly, cars may need to talk to the road itself. Technology which keeps you from veering out of your lane will probably (at least at first) require a string of magnets or transmitters in or next to the road (as in the PATH experiment), or at the very least very clear lane marker stripes (which may be as hard to maintain as the magnets).

It might also be more efficient and safer to coordinate the actions of cars from a central computing system, rather than simply letting the cars act as free agents or negotiate with each other (though central control would also bring problems, like the complexity involved in managing thousands and even millions of vehicles).

Thus we face a chicken-and-the-egg problem: why should the public go to the expense of building smart roads when there are few smart cars, and why should people buy smart cars when there are no smart roads?

How to get past this problem? Option one: we could choose inaction, which has the advantage of being a policy course which our leadership has often shown itself to be quite skilled at implementing. Fortunately, this path is not as hopeless as it sounds.

Even without collective action, driverless technology will advance. The reason is that decades of experience have shown that we will pay for car safety. Technologies like adaptive cruise control (which slows you down when an object in front of you is getting too close for comfort) and lane keep assist (which keeps you from drifting on the road) are already being installed in cars, though as a backup for drivers, not a replacement for them. But it is not hard to envision a gradual evolution as computers stealthily take on more and more of the driving chores.

One major issue is that technical standards will need to be set. If we’re going to have the roads talking to the cars and the cars talking to each other, they’ll need a common language.

But this need not be a major impediment, even without central planning. Standards often grow up organically because they aid both consumers (who can buy from a wide range of manufacturers) and producers (because each company doesn’t have to undertake the expense of building a platform from the ground up).

For example, the computer and electronics industries have, without government direction, developed numerous standards through active cooperation between consortia of firms (e.g. the standardized configuration of the USB port), or cooperation combined with bare knuckled survival of the fittest (VHS and the DVD).

However, in many cases a government role in setting standards is required. This is especially true when a publically-owned resource is involved. For example, the FCC was a motive force behind HDTV, because TV involves the use of precious public broadcasting spectrum.

The public obviously owns the vast majority of the roads in the country, so at some point we will probably need government involvement, particularly if we are considering having a central intelligence directing the traffic.

Also, even when the technological hurdles are surmounted there are serious legal and regulatory issues. At what point should you actually be allowed to take your hands off the wheel? Who will have the liability in crashes, the car maker, the owner, or the public entity which runs the road system? Should computer driving be made mandatory (more efficient), or should those who indulge in “Fahrvergnügen” always have their love of driving somehow accommodated? All of these issues will eventually fall in government’s lap.

Will we get the kind of leadership we need? Past experience shows that government involvement in the promotion of technology can be maddening; for example, thanks to politics, lobbying, and bureaucracy, HDTV was far longer in gestation (the FCC started pushing for it in 1987) and rockier in implementation (seven years behind schedule) than was originally planned. As economists often point out, government may not have any stellar expertise in picking winning technologies.

Yet we have also seen that colossal government action in transportation is possible; witness the Interstate system, which, for all its many imperfections, was a stunning achievement and the world’s biggest public works project. (FYI, along with professors Brian D. Taylor and Jeffrey Brown, I’m working on a book on Interstate history.)

Driverless cars have the potential to match and indeed surpass that accomplishment. Thus one would think this issue would loom large on the public radar screen. Instead, at the moment government seems more interested in a 19th century transportation technology than a 21st century one. More on that next time.

HAT TIP: Randal O’Toole, whose recent provocative book Gridlock nicely lays out the argument summarized here. More on O’Toole, the archnemesis of many transportation planners, coming up.

Mike L

Google "No Hands Across America". Their car drove itself cross-country in '95.

Ben Sauer

I think the politics are more complicated than you are thinking. Ideas such as raising speed limits as vehicle safety has improved meet harsh emotional rejection often without being considered on the merits. States would lose revenue if vehicles became driverless and some states just have inane policies (rural freeways in Oregon that are only 65 for example).
Further, driverless cars will incentivize driving and there's a significant political faction who believes driving is irresponsible earth destroying behavior.


I think the FCC was pushing digital television, not high definition television.


If we ever get smart cars, teenagers will stay at home.


I always thought the first robot-car driven roads would be divided like HOV lanes, and eventually, become a requirement for a vehicle to use interstates.

Although some do in some circumstances, states shouldn't be able to keep scooters, golf carts, mopeds, bicycles and Amish horse buggies from local roads, but these generally street-legal vehicles are prohibited from Interstates. Giving car owners say a 10-15 year notice that at that point in the future, interstates will be under robot operation only (with an increasing set of lanes going to the robots over time, like existing HOV lanes or the center car lanes of the NJ Turnpike) is probably the way to flush the technology through the national fleet of vehicles through attrition and turnover.


Ben, could you please elaborate on "States would lose revenue if vehicles became driverless"?


Rick in Raleigh

Lots of complications with a driver-less car-with-passengers.

But if we change the model to a passenger-less, driver-less vehicle, perhaps we could have auto-delivery trucks, attractive if there is a fleet of trucks, like say Wal-Mart.

This takes aim at inattentive/impaired drivers, and would skirt driver hours regulation. I suppose I might optimize the fleet for driving when the traffic is lowest, a kind of driving-all-night autopilot truck fleet. Just a thought.

Or personal pickup/delivery vehicles. I could order my groceries online, then send my auto-wagon over to the store to pick it up. They would load it and send it 'home'.

I think auto-drive cars (auto-auto's?) will not be what we think they will be, but be used in new ways, to do things differently or better. (eg: did the Segway people start with the knowledge that their market was going to be security guards? Probably thought we'd all be using/buying them.)



Would these vehicles have a manual override in the event of a computer malfunction?


"Ben, could you please elaborate on 'States would lose revenue if vehicles became driver-less?'"

I can think of two quickly, and am willing to be shot down for not thinking them through:

1. More efficient driving means less gas usage means less gas tax revenue. It is no accident that congestion does not get solved until it unbearable. States have to spend money to fix, and the result is less revenue.

2. No more tickets. Presumably, automated autos will obey traffic laws, which will reduce or eliminate that lucrative source of revenue.


Sally, he probably means the punitive fines police hand out for speeding.


"..And all this progress has come without any overarching entity directing the development of the technology.."

perhaps i misunderstood the wording or intent, because few singular groups are responsible for the development of a set of new technologies or the adoption of them into a market segment.

but relatively speaking, yes, there is one group that has been a deciding influence and driver for the adoption of autonomous systems navigation of vehicles - the set of darpa challenges has done this. if you re-read the charter of their series you will see thats spelled out as an intent, and there are research reports available on how the challenges have progressed things.

bit of a lack of due research there, so i guess another helpful example is to suggest following up on jaus/sae as-4 standards. and i dont see how this is all that much of a deviation from most high tech development economics.



I don't think the liability will be the biggest issue for society to tackle. Pretty much everywhere in the country today we use the standard common law system that we've been using for centuries and have adapted for the modern world.

The crux of the tort system is the concept of negligence. Whenever anyone takes an action in public they have the duty of reasonable care. That is they have to act as would a reasonably prudent person acting in the same scenario. When they do not, and that causes injury or property damage that person is liable. If for some reason you are driving today and you've taken all reasonably prudent steps and did nothing wrong but somehow a crash occurs you would not be liable-even if you or your car caused it. In this modern world with self driven cars it would be the same way-did the owner take the reasonably prudent steps. These include maintenance, safety features, and possibly any manual override that could occur.

The tort system also has in place a strict liability tort for manufacturers of products when they make a design defect (a safer design could have reasonably and cost effectively been made), manufacturing defect (the product that was purchased was designed to be safe but the product itself was different from the design in an unsafe way), and improper warnings (where a warning could have been used to inform a user about dangerous foreseeable uses).

All in all, I think the system for liability currently used would adapt quite easily into an age with driver-less vehicles.


Eric M. Jones

@7- Rick in Raleigh:

I think Rick is onto something. Just making what we currently do automatic....driving crazy miles to and fro work is nuts. But there are applications that make sense.

I opine that other models of transportation will be integrated with what we do now. But the wholesale robotizing of what we do now is not going to happen and would serve no point.

Rather than have a robot do some odious task like driving 750 miles. Give me a 250 mph maglev that transports my car and feeds me lunch and has easy chairs and bathrooms while we travel.

hawk in NC

I can feel drool from the lawyers jaws now.


Government has a rocky track record at picking winning technologies, and as the Air Force aerial refueling tanker debacle is capably showing, it can't insulate itself from the interests of lobbyists if its very life depended on it. Instead of sagely picking best of breed, the one thing you can count on is that their pick will be sure to enhance the largest number of legislators, at significant taxpayer expense.


We already have driverless cars, also know as Personal Rapid Transit or PodCars. There are now three companies that have market-ready systems, One is ULTra PRT from the U.K, another is Vectus PRT from Sweden and Korea and the third is 2getthere from the Netherlands.
All three are under construction now. Details are available at
Robocars operating on the conventional street system are a long ways off - PodCars are here now and ready to be deployed.


Oh wonderful, the great success of the interstates and government transportation systems... at drinking oil, creating accidents, subsidizing commutes, boosting inefficient car companies (and the unions that thrived on their rotting carcasses), and generally turning cities into giant parking lots.

Robocar standards... continuing the high standards of failure that have been government transportation systems for the last 50 years.

Joel Upchurch

Actually the first part of the system to provide automated steering for automobiles is already under construction. The High Accuracy-Nationwide Differential Global Positioning System should have the first segment go online in a year. This will let a vehicle determine it's position within 4 inches, which is better than than the average human driver.

It seems to me that the most important change is to encourage the transition of automobiles from mechanical steering and braking systems to drive by wire systems so they can be retrofitted for automated control.


Related to jeryzy's comment: Driverless vehicles will reduce the value in having individual vehicle ownership. Driverless vehicles may end up being a part of integrated public transport systems.


As always, the most interesting aspect to think about is the unintended consequences, of which I can think of at least two:

First, self-driving cars will reduce the inconvenience caused by parking shortages, while (probably) increasing traffic congestion. If I want to go to a restaurant, say, in an area where parking is limited, I will have the car drop me off and then tell it to go look for a parking spot.

Second, they will greatly increase the tendency towards urban sprawl. An hour-long commute is no big deal if you can work or read a book while you're in the car. And it will be a lot more worthwhile to invest in a vacation home three or four hours out of the city if you can get there by getting in your car, taking a nap, and waking up when you arrive.