Greater Greater Washington

Our car/bike/ped fights will get fiercer with driverless cars

Driverless cars sound less and less like science fiction with each passing month, and that's prompted widespread discussion about how they might change society. They will bring many changes, but when it comes to the car's role in the city, they may just intensify current tensions.

The Atlantic Cities' Emily Badger interviewed a research team of computer scientists at the University of Texas at Austin, who studied how to make intersections move far more cars than they can today. They devised algorithms that let cars flow through the intersection without need for lights that only let one direction of traffic move at a time.

But what's missing from this diagram? How about... people?

Badger writes,

[H]uman-driven cars would have to wait for a signal that would be optimized based on what everyone else is doing. And the same would be true of pedestrians and bike riders.
That certainly sounds like all other users of the road will have to act at the convenience of the driverless cars, under constraints designed to maximize vehicle movement instead of balance the needs of various users.

My background is in computer science, too, and computer scientists love figuring out how to make complex systems perform efficiently. Driverless cars provide an opportunity to optimize the real-world traffic system, if you can get most people driving computer-controlled cars and can get all of those computers to cooperate.

But you can't optimize people so easily. Already, cities host ongoing and raucous debates over the role of cars versus people on their streets. For over 50 years, traffic engineers with the same dreams about optimizing whizzing cars have designed and redesigned intersections to move more and more vehicles.

These changes frequently pushed other users aside with longer waits for crosswalks, the need to push buttons to get a walk signal, awkward bridges over wider and wider arterials, or simply omitting bike or pedestrian facilities entirely and then blaming those users when careless drivers hit and kill them.

Some pro-automotive advocacy groups like to push the theme of a "war on cars," but bicyclists and pedestrians feel like there's been a war against them since the early 20th century. This Texas team's video just perpetuates that impression.

The video even depicts an intersection with a whopping 12 lanes for each roadway, at a time when most transportation professionals have come to believe that grids of smaller roads, not mega-arterials, are the best approach to mobility in metropolitan areas.

Driverless cars, therefore, are poised to trigger a whole new round of pressure to further redesign intersections for the throughput of vehicles above all else. It won't only happen in the cities, either. Suburban areas are often ground zero for these debates, where the majority of people drive, but a significant and often growing number are either unable to drive due to age or disabilities, or are unable to afford cars. (Driverless cars probably won't be cheaper.)

Suburbs, therefore, often develop a greater tyranny of the majority, where county and state departments of transportation optimize their roadways for car throughput and leave bus stops in awkward and narrow roadside spots, leave crosswalks out or even remove existing ones, and set the stage for rising deaths.

If autonomous cars travel much faster than today's cars and operate closer to other vehicles and obstacles, as we see in the Texas team's simulation, then they may well kill more pedestrians. Or, perhaps the computers controlling them will respond so quickly that they can avoid hitting any pedestrian, even one who steps out in front of a car.

In that case, we might see a small number of people taking advantage of that to cross through traffic, knowing the cars can't kill him. That will slow the cars down, and their drivers will start lobbying for even greater restrictions on pedestrians, like fences preventing midblock crossings.

Our metropolitan areas could then look, more and more, like zoos for humans interlaced with pathways for the dominant species, the robot car. Maybe the machines really are on the way to taking over, but instead of Skynet declaring war on humans, we'll be the ones passing laws and reshaping our communities for their convenience.

I'm not suggesting we avoid research into driverless cars. Like any technology, they can bring good or evil, depending how society handles them. Driverless cars can allow buses to become on-demand jitneys and virtually eliminate the need to own a personal car in a city, or to build huge amounts of parking under office buildings. Instead of storing cars during the day, they can just drive around and transport people like taxis.

But we do need researchers excited about driverless cars not to forget the human element. The goal of our built environment is not to move cars as fast as possible everywhere, but to create a better quality of life. The computer science researchers need to also talk to their colleagues in other disciplines, set appropriate goals that consider all users of the roads, and think about what algorithms can actually make life better.

Cross-posted at The Atlantic Cities.

Support us: Monthly   Yearly   One time
Greatest supporter—$250/year
Greater supporter—$100/year
Great supporter—$50/year
Or pick your own amount: $/year
Greatest supporter—$250
Greater supporter—$100
Great supporter—$50
Supporter—$20
Or pick your own amount: $
Want to contribute by mail or another way? Instructions are here.
Contributions to Greater Greater Washington are not tax deductible.

David Alpert is the founder and editor-in-chief of Greater Greater Washington. He worked as a Product Manager for Google for six years and has lived in the Boston, San Francisco, and New York metro areas in addition to Washington, DC. He now lives with his wife and daughter in Dupont Circle. 

Comments

Add a comment »

Nice comment: "Great. No provisions at all for pedestrians. Patriotic Americans travel by fossil-fuel powered personal transportation. As soon as we can find a way to amputate the population's legs and replace them with bionic segways, the walking menace will finally be defeated."

Now we just need to hear from the usual suspects about how fusion-powered, self-driving robocars are the future, so we can finally bulldoze our big, walkable cities and all decamp to the suburbs/exurbs so we can experience true freedom and safety, or something...

by Ricky on Mar 5, 2012 10:24 am • linkreport

It's an interesting concept. Two thoughts. First, it would probably be applied only to limited access roads. There, it might lead to more separation of modes, liek pedestrins/bikes bridges. Despite the theoretical advantages of using it for high-volume intersections in urban settings, I suspect it would not get used there. Second, any use would probably involve locking out human intervention. Any vehicle that did not share the optimizing algorithm would throw the whole thing off. So basically, it's all or nothing.

by Crickey7 on Mar 5, 2012 10:31 am • linkreport

The oft decried civil and traffic engineers are at least immersed in transportation issues, and have (often) heard about complete streets, multimodalism, etc.

Computer scientists, dabbling in transportation as an application, but really dedicated, not to moving people OR to moving cars, but to their technologies, are likely oblivious to all the discussions of complete streets, etc, etc.

BTW, wouldnt the best solution for driverless cars and peds (in a well designed multimodal intersection, not the programmers' abstraction) be something like a Barnes dance?

by AWalkerInTheCity on Mar 5, 2012 10:32 am • linkreport

Sure, this system is great on paper — until the first major computer system failure or malicious hacking by some guy who lives in his mom's basement and goes by a name by "Ph1b3rPh4c3" or "KaptainKrunk" (or by our North Korean or Iranian friends) when we get the country's first 8 million car pile-up on 1-95 or wherever. Imagine the news stories on that one!

I can see this system marginalizing non-vehicular traffic even more, with pedestrians and cyclists being outright banned from most streets "for their own safety". The auto/oil/highway lobby will basically tell us to "get a car" and stop living in the past.

Not to mention that self-driving or not, these vehicles will need to run on something. Where are we going to get the energy to run something like this? Back to the Future's 'Mr. Fusion' is still science fiction.

Pathetic. Just an attempt to reinforce the autocentric model that's been forced down our throats since the 50's.

by Helen on Mar 5, 2012 10:35 am • linkreport

@Helen

[Deleted for violating the comment policy.] Driverless cars will eventually have the computing power to react many times more quickly to peds and cyclists. In fact, they already do.

Driverless cars remove the potential for human error - not signaling, simply not seeing the bike in your peripheral vision.

We already have an enormous infrastructure built for cars - or in your words, "forced down our throats since the 50's". Computing, and driverless cars, will make this infrastructure actually safe.

Read this: http://www.wired.com/magazine/2012/01/ff_autonomouscars/all/1

by Nick on Mar 5, 2012 10:40 am • linkreport

Too many people love driving too much to ever go for the self-driving car idea. Once the tech is ready for the road, I can't see it ever being more than a niche market. In the cities, yes, but rural America?

by LoLo on Mar 5, 2012 10:42 am • linkreport

I would think we will want to run with "human back up" for a long time - IE an alert, sober, adult, licensed driver at the wheel ready to deal with the vehicle during any system failure - the benefits of the automated systems would be, I think, in fewer crashes and possibly slightly improved roadway capacity - the suggestions of massive capacity improvements AND massive safety improvements, and true driverlessness, seem overblown to me.

We still have pilots (AND copilots) to back up autopilots in aviation, and we still have human backups for automation in all but the simplest transit operations - and AFAIK we havent even gotten that far in freight rail operations.

by AWalkerInTheCity on Mar 5, 2012 10:42 am • linkreport

I agree with Helen. If driverless cars are implemented, say goodbye to walkable/bikable streets forever, at least in the sububs and exurbs. Insurance companies would use their power to ensure that driverless vehicles are *never* at fault if someone gets killed by one, thus, there will be no push to make sure these vehicles are actually safe for interaction with pedestrians and cyclists. If this were the Netherlands or Denmark, there would be concern about this, but Americans generally don't give a damn about pedestrian/cyclist deaths, the rationale being that the people struck and killed should have been driving a car.

Whether or not driverless cars would kill interest living in real cities is debatable, since these have so many draws over the dull life of the exurbs.

The fuel issue is separate and more insurmountable. When gasoline becomes unaffordable to the average person as a vehicle power source (which it will at some point), the idea of a 2 to 8 ton hunk of metal, plastic and glass hauling a 150 lb. human being around will seem ridiculously wasteful. Electricity and natural gas would only make the transition off of cheap oil less painful. Operating a vehicle would still be more expensive, as would its initial manufacture.

by Thomas on Mar 5, 2012 10:52 am • linkreport

I thought this was a well reasoned argument -- and coming from an ex-googler, very welcome.

An intersting comparion is cabbies. Now, they are not as dumb as comptuers -- although there is a lot of hate for them out there. But generally, their method of driving really screws up with everyone around them. They tend to be slow, very deferential to other drivers (again, when they don't have a fare they aren't in a rush) and perform some sudden moves (sudden u turn to grab a fare) which don't make sense to other users.

I'd say -- their obvious benefits aside -- streets would be better off without cabbies.

I doubt you'll ever be able to run google style cars in the city. The data and bandwidth needs are just too high. A more probable result is further exurbization -- a self driving car becomes your private bus out to Warrenton. Your commute becomes an extension of work time.

by charlie on Mar 5, 2012 10:54 am • linkreport

"I would think we will want to run with "human back up" for a long time - IE an alert, sober, adult, licensed driver at the wheel ready to deal with the vehicle during any system failure"

The cars would be operating too close together and traveling too quickly for this to matter. By the time a human operator noticed a problem, their car would have already had the impending collision.

by geraldRX on Mar 5, 2012 10:56 am • linkreport

There might be a few "demonstration" projects, probably on stretches like the future Beltway HOT lanes where there are also regular lanes, but it's just not gonna happen on a widespread basis. Crikey7 and Helen made alot of good points, but to elaborate on "it's all or nothing", used cars will not be able to operate on driverless roads. It may be technically possible to retrofit newer cars to driverless technology, but it would probably be cheaper to buy a new one. So you're mandating that people buy a new, expensive car if they want to use any driverless road. I doubt many Amercians would like that.

by kinverson on Mar 5, 2012 10:56 am • linkreport

I find it surprising we're talking about driverless cars, but I don't hear any proposals for driverless trains. Wasn't our own Metro even designed to be driverless? Surely the challenges for trains are vastly fewer, plus, it would be a great way to cut operational costs for Metro.

by Mystery, Inc on Mar 5, 2012 10:58 am • linkreport

@DA: but a significant and often growing number are either unable to drive due to age or disabilities, or are unable to afford cars.

1. But they will be riding in autonomous cars, which will drop them off at their destination and park somewhere else far away. 2. Cars are cheap and have always been getting cheaper. You can put one on the road for $1000. This trend will continue, and I think the driverless cars will soon be inexpensive enough -- say within a generation -- to be within reach of just about everybody. This will particularly improve mobility for the very poor, enabling them to get to better jobs in the suburbs and improve their life. By dropping one person off and then going home to fetch another, driverless cars will better enable family members to share a vehicle. 3. I think you are confusing urban form with the new technology. The existing city streets will change very slowly, and is bounded by planning and zoning. I don't think there will be very many 12-lane roads.

by goldfish on Mar 5, 2012 10:59 am • linkreport

I totally agree about the comment about how this eases the marginalization of non-motorists. At a time when we should be thinking about ways to reduce the number of miles we drive these folks are finding ways to reduce the psychological burden of driving and thus increase the number of miles we drive.

How about we rethink the low-density, auto-centric development model we foolishly invested in years ago? There are lost of ways to retrofit the suburbs, especially the inner suburbs, to facilitate car-free or "car-light" lifestyles.

by Koe on Mar 5, 2012 11:02 am • linkreport

"The cars would be operating too close together and traveling too quickly for this to matter. By the time a human operator noticed a problem, their car would have already had the impending collision."

Only if you accept the technohype that this enables dramatically lower spacing between vehicles. What if we use the technology to improve safety and only do VERY modest reduction of vehicle spacing? Then human back up could work - but that eliminates the technohype that this dramaticly multiplies lane capacity.

by AWalkerInTheCity on Mar 5, 2012 11:12 am • linkreport

@LoLo:

Too many people love driving too much to ever go for the self-driving car idea. Once the tech is ready for the road, I can't see it ever being more than a niche market. In the cities, yes, but rural America?

I agree. Just to flesh-out the argument, I can see driverless cars operating in an urban environment, but at very low speeds. The fact that they're automated would make it quite easy to force them to operate at speeds at or below the legal speed limit. Given that they'll be mixed in with non-driverless traffic, they'll have the effect of calming non-automated vehicles as well. Obviously, this will be great for non-drivers, as reduced vehicle speeds will greatly improve quality of life on the streets.

Now, as far as the benefits on limited access highways, those will never come to pass for the reason LoLo pointed out: in order to realize the benefits of high-speed travel with smaller "gaps" between vehicles, less of an "accordian" effect you get with manual braking, etc... *all* the cars on the limited access highway will have to be automated. In other words, "normal" vehicles will have to be either banned, or their use will have to naturally become extinct.

That might happen in a couple of hundred years, but in 21st century America, that will never happen. It's less realistic than expecting everyone to start driving a Prius.

by oboe on Mar 5, 2012 11:12 am • linkreport

Great,
Now there's the potential that I or my kid or wife can get flattened by a self driving car traveling down a residential street at 50 mph (the insurance companies, the local Chamber of Commerce, and auto industry all will tell us this an acceptable speed) because of a blocked sensor or faulty circuit or bad WiFi connection or whatever on one of these gizmos. But hey, if someone needs to get to work or Burger King in 8.2 minutes instead of 11.7, who am I to deny them their need by speed? Talk about the death of safe and complete streets. This would start the beginning of the "motor only" era of streets...at least until gas is $30 or $40 a gallon.

by Dave Riley on Mar 5, 2012 11:17 am • linkreport

@Dave Riley: Autonomous cars will undoubtedly be fail-safe.

by goldfish on Mar 5, 2012 11:23 am • linkreport

We still have pilots (AND copilots) to back up autopilots in aviation, and we still have human backups for automation in all but the simplest transit operations - and AFAIK we havent even gotten that far in freight rail operations.
Autopilots don't really do very much. They take care of the most predictable, boring, non-intensive work. That's all.

Besides, the sky is a big place. Except around runways, airborne collisions are surprisingly improbable.

by David R. on Mar 5, 2012 11:24 am • linkreport

In terms of automated driving through the city, I'm not sure I understand why we can't continue to use pedestrian crossing signals. When a pedestrian pushes the crosswalk button, it sends out a signal to all driverless cars within XX blocks about an upcoming stoplight signal change from green to red. As the pedestrians are crossing (crossing like in Gallery Place could expedite pedestrian crossings), all automobile traffic would be halted, but it would be done so automatically to minimize traffic delays. If all cars come to a simultaneous stop and accelerate together, you minimize congestion.

In terms of cyclists and pedestrians sharing the roads with automated driving, I think the technology exists (http://www.youtube.com/watch?v=9fVWB1I9a08) for sensors to detect objects sharing the road, and the car's "behavior" can adapt to avoid a collision.

by Scott M on Mar 5, 2012 11:26 am • linkreport

Sure...fail-safe. I don't know of any real life thing that fits the description.

by Thomas on Mar 5, 2012 11:26 am • linkreport

You could mix automated and manual vehicles, in theory. But you would get very few of the benefits that automation is capable of generating. For example, in an automated setting, a light turning green will result in every vehicle stopped for the light starting to move at the exact same time. Even one manual driver in that lane would eliminate the time savings. Same for roundabouts--optimization will flow vehicles through at a far greater rate, unless some vehicles are, well, suboptimally operated.

I can also foresee that manual drivers would game the system for their benefit. Given the accident avoidance software, manual drivers would count on automated vehicles yielding to them in every instance. The increase in last-second merges and other behavior now somewhat risky might actually result in a degardation in capacity.

by Crickey7 on Mar 5, 2012 11:28 am • linkreport

I'm not sure this will really fly if the time-to-market software guys are programming these cars. Who's going to get driverless car 1.0 what with all the crashes and having to restart it again and again, and downloading patches and upgrades, don't forget the privacy policy, because they want to know where you go so that they can advertise to you better, and then there is that company that went bankrupt so now you need another software, or this software doesn't work on that platform....aggggghhhhhh!
Yeah, you have it on your laptop and on your phone (you know, the one that you have to get a new one of every year). Now it's going to drive you!

by dc denizen on Mar 5, 2012 11:30 am • linkreport

Thomas: there are so many fail-safe devices that you don't really appreciate them. Cars are already fail-safe with regards to their emission equipment and computer-controlled fuel injectors. Circuit breakers of your house. This has been the standard way of designing something for decades.

Obviously, it won't prevent every disastrous consequence of a circuit fault. What matters for a driverless car is that its safety performance is better than a human driver -- which is not that good, considering how many people I see yakking on their cell phone as they drive.

by goldfish on Mar 5, 2012 11:33 am • linkreport

"What matters for a driverless car is that its safety performance is better than a human driver "

once again, thats true IF the only benefit we are attributing to the system is a safety benefit. Same vehicle following distances, same lane capacity, BUT 10% or 20% fewer accidents -sure.

But IF we are suggesting much shorter following distances (to get higher capacities and fundamentally change the mode vs mode economics) than the automated vehicles need to be substantially better than humans. A fortiori if we expect BOTH shorter following distances AND improved safety

by AWalkerInTheCity on Mar 5, 2012 11:41 am • linkreport

People missing from the plan? This goes all the way back to Le Corbusier and his ideas of a dehumanized, sterile, vehicle-centered city Ville Radiuse, and Frank Lloyd Wright and his proto-exurban Broadacre City with its low-density development, once again based on the idea of the car being king. The only new thing is the technology, the underlying engineering is the same Johnny one-note thing we've heard since WW2; cars, cars, cars!

by Allen on Mar 5, 2012 11:46 am • linkreport

@AWITC: A fortiori if we expect BOTH shorter following distances AND improved safety

The main commercial and economic advantage goes much farther than that. Autonomous cars will relieve parking pressure and costs, because the driver will not need to park the car nearby. It will enable far more car sharing. It will allow people to work in their cars (not sure if that is a good thing). They will enable those that cannot drive to get around. They may lower taxi fares because they forgo the expense of the human driver. In short, they will clearly improve mobility in many ways, both foreseen and unforeseen.

by goldfish on Mar 5, 2012 11:48 am • linkreport

Driverless computer controlled cars might be just the thing for long, interurban trips on limited access roads. But inside urban areas? Not so much.

Do I dare make the obvious jokes about new meaning to computer crashes and blue screen of death?

by Ken Firestone on Mar 5, 2012 11:50 am • linkreport

many of those benefits rely on operation with no human backup, which I am skeptical of. However given that, I was referring more to the "line haul" benefits. If you get more car sharing or remote parking, but not roadway capacity benefits, than this tech isnt built form reshaping in the way some people think.

by AWalkerInTheCity on Mar 5, 2012 11:54 am • linkreport

@Ken Firestone: Please do. Start with this one.

by goldfish on Mar 5, 2012 11:57 am • linkreport

@charlie. "Private bus.... extension of work time." Yes, exactly. 100%. You just nailed it. The rest is details.

by Bill on Mar 5, 2012 11:57 am • linkreport

Autonomous cars will undoubtedly be fail-safe.

Right, just like metro trains are fail-safe and it's impossible for there to be an accident on the Red Line. Seriously, we don't even have automated controls for metro these days, much less driver-less trains, so how realistic is driver-less cars? What you're forgetting is that fail-safe systems still rely on error-prone humans in the long run because everything needs maintenance and correct installation/manufacturing.

"What matters for a driverless car is that its safety performance is better than a human driver "

Right, that's why no one fears flying because it's safer than driving.

What you're forgetting is that actual safety doesn't matter as much as perceived safety (from the standpoint of acceptance). And, perceived safety is much higher when you're in control.

by Falls Church on Mar 5, 2012 12:21 pm • linkreport

Personally, I'm of the opposite opinion -- driverless cars have less applicability in urban areas and are better suited for instances where cruise control is already used (highways). Basically, rather than introducing the scary concept of driverless cars, the next stage is marketing cruise control that automatically brakes if you get too close to the car in front of you, and then cruise control that also keeps your car in lane.

by Falls Church on Mar 5, 2012 12:26 pm • linkreport

@Falls Church: obviously this will be introduced gradually in the ways you suggest, so people can get comfortable with it. Nevertheless, the advantages are so plain that I think it is inevitable.

is marketing cruise control that automatically brakes if you get too close to the car in front of you

Has been available for several years now (see here).

by goldfish on Mar 5, 2012 12:50 pm • linkreport

I think the problem wit ha human fail safe (at least in the video above) is that a human can't really realize in time if something is wrong when you are approaching another car perpendicularly at 30 MPH (or faster) and you are only supposed to miss them by a foot or two.

Kind of on a tangent, but are there any places in the country where two 12 lane roads intersect? How about the world?

by Steven Yates on Mar 5, 2012 12:58 pm • linkreport

The solution for pedestrians and cyclists seems obvious - in the future, we will simply have microprocessor chips implanted in the back of our heads, enabling us to interact with all the robo-cars.
As long as nobody wears a tin-foil hat (which would interfere with the radio waves), everything will work smoothly. :-)

by Mike on Mar 5, 2012 1:04 pm • linkreport

Sorry to say, most cyclists I travel with already have an algorithm that lets them run red lights if nobody's looking or morph into pedestrians and cross at the zebra stripe.

Can we not assume that the cyclist's special rule-free force field is stronger than any technology road engineers come up with?

by Kevin C on Mar 5, 2012 1:32 pm • linkreport

Yeah, Kevin C. I imagine some of the bellyaching here has to do with the fact that walkers and bikers don't think that the rules apply to them and cross where and when they wish, ignore signals, walk along sides of roads meant for car traffic, and for not walkers/bikers, "take the lane" and other irritating stunts, etc. If they try doing that with automated cars...well, KerrRRR-SPLATTTT!!! :)

And not a judge in the world will side with them!

by Krimber on Mar 5, 2012 1:39 pm • linkreport

Wow a lot of car hate here. Relax guys. That video is a 12 lane road. Obviously not an urban intersection.

With driverless cars, the same rules will apply for ped and bikes that we have now. Faster roads will have less integration. Urban environments will not loose the walkability that they have now as long as the pedestrian is still a factor in the design process. I fail to see how having driverless cars will change that.

by Frank on Mar 5, 2012 1:41 pm • linkreport

'People' aren't missing from the diagram. They're in the cars. Cars are how people get around today, except for a few cities in the Northeast and some academic towns. Or if you're dirt poor.

Most Americans today live in the burbs, and not in the Northeast, but in the Sun Belt, the West, and the Midwest, and they get around using personal transportation because we like the suburbs, freedom, space, and having control over our lives, not to mention the lower taxes and less regulation, which encourages business and prosperity.

I personally think self driving cars won't take over any time soon, but there's more of a chance that they'll become a reality than the liberals idea that we'll all decide to move back into the crowded, decaying, dangerous, dirty cities of the past. People do like driving. When you drive your own car you're the one in control.

And what Krimber said above is right on target.

by JasonM on Mar 5, 2012 1:48 pm • linkreport

. If they try doing that with automated cars...well, KerrRRR-SPLATTTT!!! :)

And not a judge in the world will side with them!

Incidentally, this is why VA should have passed the "due care" motorist law. People questioned why it was necessary...it's so the judge doesn't have to side with the above nonsense regardless of whether the car is driverless.

The Virginia House Transportation subcommittee 2 today voted to kill, a bill that would have drivers of motor vehicles exercising due care to avoid crashing into a pedestrian or bicyclist.

When assessing liability, a “due care” provision makes it clear that a driver cannot avoid liability simply because he or she had the right of way

http://fabb-bikes.blogspot.com/2012/01/bill-to-see-drivers-to-exercise-due.html

by Falls Church on Mar 5, 2012 1:58 pm • linkreport

Well if the rules are programmed to ignore pedestrian and bike needs then it should be natural to expect an increase in "rule-breaking". I know that if i'm even on a bike going through an automated intersection that isn't programmed to know what to do with me then I'm going to do whats safest, which would be to get the intersection ASAP.

by Canaan on Mar 5, 2012 2:01 pm • linkreport

I actually think the threat from these vehicles to promoting walkable and/or denser living environments will be indirect rather than direct.

These vehicles will make point-to-point travel even easier, faster, and cheaper than it is now, meaning there will be even less demand for walkable cities than we have now. Self-driving, hybrid electric vehicles will probably be the death of the city, public transportation, and the minimal concern that we have today for non-motorized transportation.

If that's ultimately the case, I think it's unfortunate, but that's human progress and freedom for you.

by tomlinson on Mar 5, 2012 2:04 pm • linkreport

@JasonM: When you drive your own car you're the one in control.

I'm tickled by this statement, made as a comment on a post about designing intersections for cars that drive themselves.

I do wonder, however, whether it also applies to other forms of locomotion -- for example, are you also the one in control when you walk your own feet? How about when you ride your own bike?

by Miriam on Mar 5, 2012 2:06 pm • linkreport

#MotoristDueCare
I'm all for this, as long as it doesn't mean that cyclists get to run red lights and then plead that the car with the green (and the right of way) should have exercised due care.

I'm in favor of transit, traffic calming, walkable streets, more bikes, and fewer cars. But a future of more bikes mean that cyclists will have to stop at lights like cars, not run them like bike messengers or switch to the sidewalk/zebra like they were on foot.

And improve bike visibility with mandates for lights and reflectors.

by Kevin C on Mar 5, 2012 2:13 pm • linkreport

You're in control in a way, but biking and walking don't give you the freedom of options that driving a car/motorcycle/truck does. This is America, and not a European country or Japan or some other mini-nation where all your destinations are in a 3 mile radius of one another. In most of this country your house, job, your kids' schools, and stores you need to visit are NOWHERE near one another. My work commute is 36 miles in each direction. Taking a bike (unless it's a Harley) is quite impossible.

My point still stands, we need mobility for liberty and prosperity, and we decided long ago that cars are how we're going to have it. I personally, am tired of the war on cars and think it will ultimately fail because it involves coercion and goes against the will of the people.

Self-driving cars may not happen soon, or ever, but they're a few shades more plausible than expecting Americans to become urbanites again.

by JasonM on Mar 5, 2012 2:14 pm • linkreport

I do wonder, however, whether it also applies to other forms of locomotion...

Not when riding a horse -- what the city was originally designed for, after all.

by goldfish on Mar 5, 2012 2:16 pm • linkreport

This is like the flying car idea, never going to fly. I just saw this lecture on that TED series, http://www.ted.com/talks/avi_rubin_all_your_devices_can_be_hacked.html showing how every computer can be hacked, from your mobile device to your pace maker to all the computers in your car. Now the driving will be computerized?

People like driving until driving isn't fun anymore and that is where all the trend lines are going. To each his own, but when people talk about freedom, the Foo Fighters "Walk" videois closer to reality. http://www.youtube.com/watch?v=4PkcfQtibmU

by Thayer-D on Mar 5, 2012 2:17 pm • linkreport

Meh, this seems like alarmist claptrap to me. I think driverless cars are the solution that best satisfies the needs of all parties.

I most agree with oboe in some of the general principles he laid out -- namely, that urban travel will primarily be at low speed limits for driverless cars. Where I differ (I imagine) is that I still think high speed highways will persist. You'll have several main arteries in city that will give cars a priority. It is on these few roads that pedestrians will find their journeys retarded by a wait -- but it will only be a few roads in the city. There will be a number of minor arterials that balance the needs of peds and cars, but for the most part, streets will be inhabited by low speed cars and dominated by peds or bikes.

I don't think people will run into traffic, knowing that driverless cars will stop for them. There's simply no way that driverless cars will reduce the risk of death to pedestrians to zero, allowing for them to run amok.

I agree with Koe, that this will increase the number of miles "driven" per person, but I don't see that as a bad thing. If you do some work on your laptop while commuting in from the suburbs, you can afford a longer, slower commute. Our concept of the workday starting when you show up at the office will be obsolete. People will have more flexibility to arrive at the office at 10, 11, or 12. Rush hour will become less and less severe.

This, I think, is the main driver of adoption of driverless cars, initially. Would I pay more for a car which I could work out of, could pick me up from a bar/restaurant, and could take my child to school? Absolutely -- I'd pay a lot more. Being able to attach wages to the time now spent commuting overwhelms the cost of a driverless car and/or service -- which means the initial cost of these things could be quite high and still sell in significant numbers. People are going to open their wallets to for what amounts to a vast improvement in lifestyle and potentially, income. It doesn't have anything to do, really, with shorter commutes due to increased efficiency in the use of roads.

I think once driveless cars become fully legal, it's a matter of a generation of cars (7 - 10 years) before most cars on the road are driverless. Soon thereafter, I imagine the gov't begins to curtail the issuance of driver's licenses with higher fees or greater restrictions.

by tresluxe on Mar 5, 2012 2:36 pm • linkreport

@goldfish, you are in control when you drive your own car, but you are not in control when you ride your own horse?

by Miriam on Mar 5, 2012 3:04 pm • linkreport

@Miriam -- you think you are in control when you ride a horse, due to the animal's forbearance and calm demeanor. But a horse can spook and cause injuries.

by goldfish on Mar 5, 2012 3:12 pm • linkreport

@goldfish --

you think you are in control when you ride a horse, due to the animal's forbearance and calm demeanor. But a horse can spook and cause injuries.

Yes, that's exactly the situation when I drive a car!

by Miriam on Mar 5, 2012 3:22 pm • linkreport

@Kimber:

walk along sides of roads meant for car traffic, and for not walkers/bikers, "take the lane" and other irritating stunts, etc. If they try doing that with automated cars...well, KerrRRR-SPLATTTT!!! :)

Funny how two people can read on the same subject and come to such radically different conclusions. I'm completely in favor of having automated cars in an urban environment. Given that the cities of the future will inevitably be more pedestrian and cyclist friendly than they are at the moment, we're unlikely to see speed limits go up (as high traffic speeds degrade quality of life at the pedestrian level) so we can enforce our 20 mph speed limit automatically, and the automated car will obey blindly.

Same thing with riding a bike: I can take the lane and ride as slowly as I like, and the automated car will gladly chug-a-lug along behind me.

Sounds like paradise!

(PS: as for the "irritating stunt" of taking the lane, you might take a bit of time to learn the difference between "things that are illegal" and "things that pose a minor inconvenience to me". Look into it.)

by oboe on Mar 5, 2012 3:24 pm • linkreport

In 20120, an estimated 32,788 people were killed in traffic accidents (National Highway Traffic Safety Administration). If robo cars can reduce this amount by 50% or so wouldn't that be fantastic.

by David J on Mar 5, 2012 4:44 pm • linkreport

The robocars will know all the traffic laws pertaining to cyclists. They will maintain a minimum 3 foot distance at all times. They will not honk at me "to let me know they're there", so that I will get out of their way or because they believe I am violating a traffic law they've just invented. They will not halfway pass me, then start drifting into my lane. They will not take a right on red at 20 mph while I'm entering the intersection. They will not make a u-turn into me. They will not throw bottles at me or spray washer fluid while passing me. They will not assault me on the side of the road after they've just run into me.

by Crickey7 on Mar 5, 2012 4:55 pm • linkreport

By the way, cities were never designed for horses. Horses were always extremely expensive and difficult to maintain and were almost exclusively in the ownership of businesses and transit services. At the peak in 1900, Washington, D.C. had only 4,471 horses for 278,718 inhabitants. Pre-automobile cities were dominated by pedestrians.

by Eric Fischer on Mar 5, 2012 5:05 pm • linkreport

To piggyback on Crickey...

Robocars will be able to asses a potentially hazardous cycling situation for a biker, thus giving them a gentile honk even if the danger is coming from another car.

Robocars will be equip with video cameras and can playback any altercation that may happen so it's clear exactly who's at fault. This will hopefully work to minimize pedestrian abuse of their conservative driving.

by Roccy on Mar 5, 2012 5:07 pm • linkreport

Robocars won't need horns. They might not even need exterior lights of any kind.

by Crickey7 on Mar 5, 2012 5:19 pm • linkreport

@JasonM- not to mention the lower taxes and less regulation

If you think bigger/wider roads and/or the suburban form cost you LESS in taxes, you're wrong. I'll direct you to a couple of articles that might change your mind on that score somewhat.
http://www.strongtowns.org/journal/2012/1/2/the-cost-of-auto-orientation.html
http://www.postindependent.com/article/20110712/VALLEYNEWS/110719986/1083&ParentProfile=1074
http://www.strongtowns.org/journal/2011/8/15/measuring-productivity.html

by thump on Mar 5, 2012 5:19 pm • linkreport

Every time one of these threads pops up I am shocked and amazed at the new and ever more complicated things this nonexistent technology "will" be able to do!

The computing and networking power required to implement something like what is depicted in that video is astounding. These cars will NEVER do that on regular streets. I could foresee highways with the infrastructure to support cars at high speeds close together, but this video is ridiculous. One dropped packet and you've got yourself a 15-car pileup.

Cars that can detect bicyclists and follow all the rules around them and have spidey-sense that allows them to alert cyclists of impending doom? Please.

by MLD on Mar 5, 2012 5:22 pm • linkreport

I look forward to cheap teleporter technology. It will be so great! No one will ever enter a city again! We'll all live in isolated bliss in a suburban cul-de-sac. Heck, with telecommuting we may never have to leave our bedroom at all once we attain adulthood.

FREEDOM!

by oboe on Mar 5, 2012 5:39 pm • linkreport

Interesting article David.

Obviously, the intersection technology is really just a tech fantasy that won't come to pass for a very long time.

Driverless Cars don't need magical traffic lights to work. They don't need to have magical transponders that can communicate with all cars around them.

In fact, today's cars can be adapted - have been adapted - to be fitted with Driverless Car technology. Driverless Cars are currently being tested in live environments in not only the United States but in Germany and also China.

The benefits are enormous:

- According to the NHTSA, over 30,000 people die annually on American roads with over 1m are injured, 90% of these deaths being result of human error.
- Road accidents cost the US economy approx $230bn annually, close to 2% of the economy every year.
- Traffic accidents are actually the number 10 cause of death worldwide (note that the figure includes all deaths, not just premature death).
- This is not to mention the indirect threat posed by cars in causing untold numbers of deaths from respiratory infection and possibly global warming.

Matthew Newton
Driverless Car HQ

by Matthew Newton on Mar 5, 2012 6:28 pm • linkreport

I doubt you'll ever be able to run google style cars in the city. The data and bandwidth needs are just too high.

Google is already running its self-driving cars in the city. Dense, congested Californian cities. I don't know why you think data and bandwidth are an issue. Self-driving cars are not drones controlled by a central computer that needs to transfer massive amounts of data over a communications network. They're autonomous agents making their own decisions. Almost all the sensing and processing is done on board each vehicle. They're like regular cars, but with an on-board computer doing the driving instead of a human being. Future self-driving cars may talk to neighboring vehicles to negotiate cooperative local maneuvers (passing through intersections, etc.), but the only kind of distant external communication needed is for things like GPS navigation and traffic conditions. There's no reason to think data and bandwidth limits are going to be a problem.

Not to mention that self-driving or not, these vehicles will need to run on something. Where are we going to get the energy to run something like this?

The same sources we get it from now. But the mix will probably involve less petroleum products and more natural gas and biofuels, and much more electricity, both centrally-generated (coal, nuclear, hydro, wind, solar) and locally-generated (mostly rooftop solar installations).

used cars will not be able to operate on driverless roads.

There won't be "driverless roads." Or, at least, there won't be many of them. There might be some "driverless-only lanes" on some freeways, but self-driving cars will mostly operate on ordinary roads alongside human-driven vehicles are other robotic vehicles. This is how the driverless cars from Google and all the major automakers work. This is how they have to work as we transition away from human-driven vehicles. It would be far too expensive to build large-scale road infrastructure just for driverless vehicles.

by Bertie on Mar 5, 2012 6:41 pm • linkreport

@mystery, inc. See this

@goldfish, Fail-safe?

@Krimber, there is one thing driverless cars won't do - and that is KerrRRR-SPLATTTT. See Asimov's Rules.

@oboe and Crickey7 are dead on. Driverless cars (if they ever hit the road, and I'm highly skeptical they will) will make the roads much, much safer. But they won't make driving much faster, because those two goals are in direct contrast. In fact vehicle speeds will likely go down and following distances will go up. Still driving might be just as timely thanks to savings that result from very few crashes and efficient movement.

by David C on Mar 5, 2012 7:30 pm • linkreport

@David C: Dr Strangelove is the far better movie. Fail-Safe was earnest, humorless, and worst of all, boring.

by goldfish on Mar 5, 2012 7:39 pm • linkreport

Correction: Dr. Strangelove.

by goldfish on Mar 5, 2012 7:45 pm • linkreport

The computing and networking power required to implement something like what is depicted in that video is astounding. These cars will NEVER do that on regular streets. I could foresee highways with the infrastructure to support cars at high speeds close together, but this video is ridiculous. One dropped packet and you've got yourself a 15-car pileup.

Huh? The location, speed and direction of each vehicle would require no more than a few hundred bytes of data. Communicating and processing that amount of data is trivial. How do you think a tiny chip in your phone or video player is able to process HD video, at 50 MBytes per second or more? And no one is going to design a communication and control protocol so fragile that a single lost data packet would result in catastrophic failure.

by Bertie on Mar 5, 2012 7:45 pm • linkreport

@bertie; what google is doing is doing a massive laser 3d scan of the area, comparing it to a previous 3d scan, and then figuring out the differences and calculating the car's route.

For one car, you might find the bandwidth. You'd certainly be flooding that cellur tower internet connection. 5 or more cars driving around -- no.

I know storage costs are always falling, but we're talking about a trunk full of hard drives to download one slice of a metro area.

by charlie on Mar 5, 2012 7:59 pm • linkreport

@bertie; what google is doing is doing a massive laser 3d scan of the area, comparing it to a previous 3d scan, and then figuring out the differences and calculating the car's route. For one car, you might find the bandwidth. You'd certainly be flooding that cellur tower internet connection. 5 or more cars driving around -- no. I know storage costs are always falling, but we're talking about a trunk full of hard drives to download one slice of a metro area.

I've seen the videos and technical descriptions Google has released, and none of them indicate a networking requirement that would prevent multiple vehicles from operating simultaneously in the same area. That wouldn't make any sense. They're not designing a system that would allow only a single self-driving vehicle at a time. They're designing a general-purpose replacement for human-driven cars, for all driving environments. That's why they've been testing their prototypes all over the state, from busy streets in downtown San Francisco to isolated country roads.

Driverless cars (if they ever hit the road, and I'm highly skeptical they will) will make the roads much, much safer. But they won't make driving much faster, because those two goals are in direct contrast.

No, they're not in conflict. Driverless cars will reduce travel times in a number of ways. They'll reduce congestion, by increasing the effective capacity of roads, by more efficient routing, and by reducing the number of accidents. And they'll reduce wait times at intersections. Ultimately, as the animation illustrates, vehicles may not need to wait at intersections at all (or may need to wait only rarely, when pedestrians or cyclists need to cross). And on freeways, self-driving vehicles will be able to travel at much higher speeds than human-driven vehicles without increasing the risk of accidents.

by Bertie on Mar 5, 2012 8:57 pm • linkreport

Bertie,

Slower = Safer. So fast is in direct conflict with safe. If you want cars to follow each other by 3 feet at 100 mph, that will be less safe. It takes just one glitch to cause a lot of damage. It makes more sense to keep the 2 second rule in place, which limits top speeds. But most people would be willing to give up some speed for safety.

by David C on Mar 5, 2012 9:11 pm • linkreport

It would be nice to see this simulation alongside simulations of normal cars using standard intersections.

I'm not convinced that this configuration is any more efficient than standard operation. Sure there's constant flow through the intersection, but as people are pointing out, the intersection is _really_ big, especially in comparison to the number of cars on the other portions of the road. What roads worth easing congestion have midblock sections that are 90% empty?

by Paul on Mar 5, 2012 10:04 pm • linkreport

Slower = Safer. So fast is in direct conflict with safe.

That may be true when the comparison is human driving at different speeds. But you're comparing human driving to self-driving cars. Unlike humans, computers don't get tired or drunk or distracted. They don't make mistakes judging the speed and distance of other vehicles. They have much faster reaction times than any human. For these reasons and others, it is almost certain that self-driving cars will be able to drive faster than current cars and still be safer than current cars.

That doesn't mean we necessarily will allow self-driving cars to drive faster, of course. We may choose to keep current speed limits in the interest of saving fuel, for example. But self-driving cars will still save time compared to current cars by reducing delays caused by congestion, accidents, waiting at intersections, etc.

by Bertie on Mar 5, 2012 10:07 pm • linkreport

But self-driving cars will still save time compared to current cars by reducing delays caused by congestion, accidents, waiting at intersections, etc.

Yes, things will be wonderful in 2145.

by David C on Mar 5, 2012 10:29 pm • linkreport

@David C

Not to be glib, but how long did it take us to transition from horse and buggy to the automobile? The technology exists today. The biggest hold up is going to be passing new laws -- a formal adoption of driverless cars by our culture. Long before that happens, Google's going to have partnered with Ford (or whomever), and be ready to mass produce a line of "chauffeured" Fords vehicles.

by tresluxe on Mar 6, 2012 1:27 am • linkreport

Bertie is right. These cars are just around the corner and it's pointless to try to stop the march of progress. It will be fun for future historians to answer kids' questions like "What was public transportation?", "Why would people use a train at all?" and "Why did so many people used to live so close together?" :)

by John on Mar 6, 2012 8:01 am • linkreport

@David C: Slower = Safer

That's just plain silly! Fast will not be in direct conflict with safe, because Newton's laws of motion will not apply to driverless cars, of course! It's another one of the amazing benefits of the technology.

by Miriam on Mar 6, 2012 8:03 am • linkreport

Bertie is right. These cars are just around the corner and it's pointless to try to stop the march of progress.

Exactly! Remember back in the 50s and 60s before people had jetpacks and personal aircraft to fly themselves to work. There were petty naysayers back then as well. I bet those who are alive to see how our culture has been transformed are pretty embarrassed!

Here's a bit of context for those poor unfortunates who were born without the capacity for skepticism:

http://www.popularmechanics.com/technology/engineering/future-that-never-was-next-gen-tech-concepts

by oboe on Mar 6, 2012 9:25 am • linkreport

More spot-on predictions from the archives:

http://blog.modernmechanix.com/2006/10/05/miracles-youll-see-in-the-next-fifty-years/

by oboe on Mar 6, 2012 9:26 am • linkreport

@tresluxe,

how long did it take us to transition from horse and buggy to the automobile?

The first automoile was invented in 1769, and you can still find buggies on the street today. So you tell me.

The technology exists today.

No. It really doesn't. A car can do much of the driving, but it still needs the driver to take over from time to time. And it can't do the things that people are talking about in this thread.

The biggest hold up is going to be passing new laws -- a formal adoption of driverless cars by our culture.

No it will be making AI work at an acceptable level. I can't even get my laptop to work more than 95% of the time. And these cars will need to work 99.9999999% of the time (if not more).

by David C on Mar 6, 2012 9:52 am • linkreport

John

These cars are just around the corner and it's pointless to try to stop the march of progress.

They aren't around the corner, as others have pointed out above, unless it is a very long corner. But even if it were, no one is trying to stop the march of progress. They're just trying to make sure that this progressive future includes a place for people who want to walk and bike. We don't want to repeat the mistakes of the past. I for one welcome safer, more efficient vehicles. And I'm sure you do too, which is why you'd be content with limiting cars to no faster than 75 mph and doubling CAFE standards.

by David C on Mar 6, 2012 10:07 am • linkreport

DavidC,

I don't know why you keep insisting that self-driving cars are far off in the distant future when prototype vehicles already exist and have already logged hundreds of thousands of miles of accident-free autonomous driving in all kinds of environments and conditions. Every year, automakers introduce new features that bring production models closer to fully autonomous operation. Mass-market production cars already have active cruise control, automated lane departure warning systems, automated parallel parking systems, automated pedestrian detection and collision avoidance systems, and other automated features. Next year, Mercedes will add a new feature to its S-class models that allows the cars to drive themselves within their lane in slow-moving traffic. The car will steer itself and maintain a constant distance from the car in front. GM and other automakers plan to introduce the same feature in a couple of years. Around 2016, Nissan expects to introduce an automated valet parking system. The car will drop you off, and then go and park itself in a special valet parking area. Auto industry pundits are predicting that fully-autonomous vehicles will be on sale within 15 years. Last month, Nevada became the first state to add specific provisions for the registration and testing of driverless cars to its road laws. A bill was introduced in California a few days ago to do the same thing there.

As Will Handsfield pointed out, it took only 30 years for motorized vehicles to almost completely displace horses and horse-drawn vehicles in America's urban areas.

by Bertie on Mar 6, 2012 10:34 am • linkreport

I think to a certain extent people are talking about two different forms of automation. Safety automation on cars will become more commonplace, though as an overlay to the human controller. Those systems can exist comfortably with vehicles that do not the technology, and really are just an extension of the trend toward ever-safer cars. Nothing revolutionary there.

The study was about automated control. That's in place of human control and in theory is revolutionary. But it cannot exist on the same system as uncontrolled vehicles without losing nearly all of the benefits. That kind of system is at least a generation away, and is likely to remain so.

by Crickey7 on Mar 6, 2012 10:36 am • linkreport

I don't know why you keep insisting that self-driving cars are far off in the distant future when prototype vehicles already exist and have already logged hundreds of thousands of miles of accident-free autonomous driving in all kinds of environments and conditions.

Because those vehicles require a driver to take over a few times every trip. They can only be used on pre-selected and programmed courses and they don't yet work with the hive technology one needs to make the video above work. Then there are the data problems that relate to all of these cars trying to communicate massive amounts of data with one another and how that overwhelms current communication systems. There are probably another 2 dozen serious technical issues that need to be overcome.

When the lead engineer will buckle his new-born child into the car by themselves and send them between two places chosen at random (around 100 miles apart), then I'll buy that we're 30 years away.

Getting to the next level is significantly more difficult then putting the cars into production.

by David C on Mar 6, 2012 11:23 am • linkreport

@oboe

Fun article, but that's not really an argument against driverless cars themselves. Notably, the article does predict the invention of online shopping. Some technologies that are predicted do come to fruition. You should read, "The Machine Stops", a short story by E.M. Forester, which over 100 years ago makes a number of remarkably prescient predictions about modern technology and lifestyle. It's a good read.

@David C

We're going to have to differ on this. I could point to the fact that my phone has as much processing power as the tower you or I owned 15 years ago. Technology is advancing at a breakneck pace. I honestly don't think the simulation in the video is around the corner, but clearly, cars that self drive are. The beta version of these cars exist today. How long do you really think it will take to get to 1.0? I'd be amazed if in 5 years, there wasn't a prototype that was 99.999999 accurate, based on redundant systems or something similar.

by tresluxe on Mar 6, 2012 11:32 am • linkreport

@David C:

The technology exists today.

No. It really doesn't. A car can do much of the driving, but it still needs the driver to take over from time to time. And it can't do the things that people are talking about in this thread.

I think the mismatch here is between people with any sort of engineering experience, and those without. As @Bertie says, "We're *almost* there!"

Of course, you can always tell the engineers from the non-engineers in any meeting, because when someone inevitably says, "We're 85% there!" the non-engineers clap their hands and plan an early vacation, and the engineers roll their eyes and get ready for a slog of sleepless nights.

by oboe on Mar 6, 2012 11:34 am • linkreport

The study was about automated control. That's in place of human control and in theory is revolutionary. But it cannot exist on the same system as uncontrolled vehicles without losing nearly all of the benefits.

Huh? The single biggest benefit of a self-driving car is that you don't have to drive it. You'll be able to get in the car, tell it where you want to go, then sit back and read a book or watch a video. You'll get that benefit whether other cars on the road are self-driving or not. Even the limited forms of automation already available provide substantial benefits. That's why people are willing to pay thousands of dollars for them. Automated parallel parking systems make parallel parking much easier. Active cruise control relieves the driver from the chore of constantly monitoring and adjusting speed to match the vehicle in front. Lane departure prevention systems make it easier to stay in your lane. And so on.

by Bertie on Mar 6, 2012 11:35 am • linkreport

Even the limited forms of automation already available provide substantial benefits. That's why people are willing to pay thousands of dollars for them.

Yes. Because people never pay tons of money for frivolous bells and whistles - like racing stripes.

But on a serious note, all of those systems you mention are trivial compared to self-driving or even more complex cooperative self-driving.

by David C on Mar 6, 2012 11:49 am • linkreport

Here's Kevin Drum's take on it (one with which I disagree, btw):

Sure, cars won't need signals at intersections, but neither will people. If you want to cross a road, you'll just cross. The cars will slow down and avoid you. You could cross blindfolded and be perfectly safe. You'll be able to cross freeways. You'll be able to walk diagonally across intersections. You'll be able to do anything you want, and the cars will be responsible for avoiding you. Your biggest danger will come from cyclists and other pedestrians, not cars.

Hopefully he'll be right, and we'll set our city speed limits to something like 15-20 mph, and drivers will be incapable of driving faster than that. Hopefully crosswalks (and "jaywalking") will be a thing of the past, and we can put the onus for avoiding collisions back on vehicle operators where it belongs.

It would be a much saner environment than the "Bully Rules" we currently operate under, where we start from the position that non-driver humans should GTFOOMY, and derive our laws from that principle.

by oboe on Mar 6, 2012 11:49 am • linkreport

But on a serious note, all of those systems you mention are trivial compared to self-driving or even more complex cooperative self-driving.

Bah! We've got coverage of over 85% of the functional requirements, David. Why so pessimistic?

:)

by oboe on Mar 6, 2012 11:51 am • linkreport

Somehow I imagine a world of silent electric cars, speed limited to 25 MPH, but able to pass within six inches of a bike or pedestrain thanks to milimeter wave radar isn't going to be very fun.

by charlie on Mar 6, 2012 11:54 am • linkreport

@charlie,

Agreed. And of course, the competing model where all motor vehicles are speed-limited to 15 mph, and required to yield ROW to pedestrians and cyclists at all times (with a 3 foot buffer, no less!) doesn't seem like it would appeal to many of our current drivers.

"Ummm. Now remind me why in the fnck I'm giving up my non-autonomous 911 Turbo S again?"

by oboe on Mar 6, 2012 12:02 pm • linkreport

Because those vehicles require a driver to take over a few times every trip. They can only be used on pre-selected and programmed courses and they don't yet work with the hive technology one needs to make the video above work.

Again, this just isn't true. Last October, Google announced that its driverless cars have already made trips of more than 1,000 miles without any kind of driver intervention at all. It's now working on a million miles. And even when the driver did intervene, it wasn't for a safety-critical need. As for the video above, it's not about driverless operation. It's about eliminating the need to stop at intersections.

Then there are the data problems that relate to all of these cars trying to communicate massive amounts of data with one another and how that overwhelms current communication systems.

What data problems? There is no need for the cars to communicate massive amounts of data with one another, any more than human-driven cars need to do that. As I said in an earlier comment, self-driving cars may eventually talk to neighboring vehicles to facililate cooperative maneuvers like passing through intersections, but that's certainly not a requirement for driverless operation, and it would involve only very limited amounts of data (location, speed, direction, etc.).

When the lead engineer will buckle his new-born child into the car by themselves and send them between two places chosen at random (around 100 miles apart), then I'll buy that we're 30 years away.

Alan Taub, Vice President of R&D at GM, predicts that the first fully autonomous vehicles will be commercially available by 2020. That's only 8 years away. Other auto industry experts have made similar predictions. Perhaps that's too optimistic. Perhaps it'll be 2030. The point is that all the evidence suggests that it's coming soon. Again, it took only about 30 years for motorized vehicles to almost completely replace horses and horse-drawn vehicles in American cities.

by Bertie on Mar 6, 2012 12:32 pm • linkreport

I guess my question to the naysayers, is why would one value the opinion of an anonymous blog commented over that of industry experts? Fair question for this fierce debate.

by tresluxe on Mar 6, 2012 12:46 pm • linkreport

You'll probably want to Google "vaporware".

by oboe on Mar 6, 2012 12:47 pm • linkreport

A skeptical perspective to balance some of the techno-utopianism:

"There are ways for self-driving car technology to have an impact and save lives without delivering completely autonomous vehicles," [MIT's John Leonard] says, adding that bits and pieces of the technology will likely deliver new and ingenious safety and environmental benefits. "But the question of how to deliver fully autonomous vehicles remains unanswered to me."

http://www.scientificamerican.com/article.cfm?id=google-driverless-robot-car&page=2

by oboe on Mar 6, 2012 1:04 pm • linkreport

Wierd not to have heard from the "car driving equals freedom" crowd. What happens if you are checking out an old city neighborhood and some cool looking brewery building catches your eye? Can you jump in and out of control as your mind wanders? Too many permutations and therefore vunrabilities. Kind of like young'uns prefering urbanism, some need to mate (especially in spring time) and to do that you need to be around eachother, not in programmed metal and glass boxes. I'm sure there's an app for that, but not quite the same.

Technutopia will destroia.

by Thayer-D on Mar 6, 2012 1:12 pm • linkreport

"Again, it took only about 30 years for motorized vehicles to almost completely replace horses and horse-drawn vehicles in American cities."

the cutting edge pre auto techs for urban passenger transport in 1890 was NOT the horse drawn vehicle, they were electric railways (street cars and elevateds), and bicycles.

by AWalkerInTheCity on Mar 6, 2012 1:21 pm • linkreport

"Last October, Google announced that its driverless cars have already made trips of more than 1,000 miles without any kind of driver intervention at all. "

have the made public the routes? how the routes were selected? if there were any route changes made on the way? The extent of immediate pretrip bug checking? Etc, etc?

by AWalkerInTheCity on Mar 6, 2012 1:24 pm • linkreport

driverless cars have already made trips of more than 1,000 miles without any kind of driver intervention at all

Under what conditions?

Again, it took only about 30 years for motorized vehicles to almost completely replace horses and horse-drawn vehicles in American cities.

What is your start point. Certainly not the first automobile. Certainly not the first modern auto. Certainly not the first commercially available auto. And when did this replacement finish exactly?

Even when the first car was available for sale, we were still 50-70 years away from complete conversion. The first driverless car isn't even available for sale.

why would one value the opinion of an anonymous blog commented over that of industry experts?

You tell me. You're the one arguing with me. Still, I have worked in the car industry, have a masters in engineering and have spent some time automating vehicles, so it's not like I'm talking out of my kiester here.

by David C on Mar 6, 2012 1:31 pm • linkreport

Last I heard, the google cars can't drive in the snow or the rain. And have trouble in fog. So that seems like a problem.

by David C on Mar 6, 2012 1:34 pm • linkreport

@David C:
That's okay. Human drivers can't drive in snow or fog either.

by Matt Johnson on Mar 6, 2012 1:37 pm • linkreport

@Matt Johnson; nope, just southerners.

by charlie on Mar 6, 2012 1:47 pm • linkreport

Robots already (mostly) fly most of our planes for thousands of flights a day and millions of miles a year. Rules are used to allow them to share the skies with balloonists, gliders, visual-navigating hobbyists, rocket launches, supersonic military flights, and fully instrumented planes. It works ok.

Sometimes some modes are excluded from some paths. Fully GPS flights have to fly higher or use exclusive runways.

I don't see any reason to suppose the robots are coming to take your bike away or kneecap your stroll.

by Kevin C on Mar 6, 2012 2:07 pm • linkreport

Even when the first car was available for sale, we were still 50-70 years away from complete conversion. The first driverless car isn't even available for sale.

Part of the difficulty here was the development of gasoline and its distribution, a process that took 20 years at least. This hindered the adoption of automobiles. You know, the "standard oil" thing (a marketing campaign, btw) and John D. Rockefeller monopoly and all that. I think that compared to horses and bicycles, everybody recognized the advantages that gasoline-burning transportation offered, and eagerly awaited its arrival.

Much like autonomous vehicles today. Here there are no economic and commercial roadblocks (such as the development of fuel distribution); the primary obstacles are legal.

by goldfish on Mar 6, 2012 2:09 pm • linkreport

"Robots already (mostly) fly most of our planes"

Flying in three dimensional space with no incumberances is hardly the equivalent of driving in a two dimensional city street with pedestrians and bikes on their own illogical paths. How'm a gonna splain diss!
Where's the King of Spain when you need him?

by Thayer-D on Mar 6, 2012 2:16 pm • linkreport

No, robots do not fly most of our planes. Patrick Smith, Salon columnist and commercial pilot, writes:
Essentially, high-tech cockpit equipment assists pilots in the way that high-tech medical equipment assists physicians and surgeons. It has vastly improved their capabilities, but it by no means diminishes the experience and skill required to perform at that level, and has not come remotely close to rendering them redundant. A plane can fly itself about as much as the modern operating room can perform a surgical procedure by itself.
And:
You’d be surprised how busy a cockpit can become — with the autopilot on. For all of its scripted protocols, checklists and SOP, hundreds if not thousands of subjective inputs are made by the crew, from deviating around a cumulus buildup (how far, how high, how long), to troubleshooting a mechanical issue to performing the takeoff and landing.
The task of avoiding traffic falls to the air traffic control system, not to the autopilot or - except under VFR - primarily to the crew. Even without ATC, the sky is a big place. Except around runways, you would be hard-pressed to hit another plane, even intentionally.

by David R. on Mar 6, 2012 3:38 pm • linkreport

I have put a commentary on this article at http://ideas.4brad.com/bicycles-robocar-world in my robocars blog

by Brad Templeton on Mar 6, 2012 5:12 pm • linkreport

A skeptical perspective ...

That article is from May of last year. Google has announced several key milestones since then, including the trips of over a thousand miles without any human intervention. The automakers have also provided more information and results about their own driverless vehicle projects. The field is advancing very rapidly.

the cutting edge pre auto techs for urban passenger transport in 1890 was NOT the horse drawn vehicle, they were electric railways (street cars and elevateds), and bicycles.

I wasn't talking about the transition from electric railways to automobiles. I was talking about the transition from horse-drawn vehicles to motorized vehicles. That was a huge shift in transportation technology over a very short period of time. In just a few decades, we virtually eliminated from urban areas the form of transportation that had been dominant for thousands of years. Today, horses are limited to tiny niche transportation markets, like romantic horse-drawn carriage rides through Central Park in NYC.

What is your start point. Certainly not the first automobile. Certainly not the first modern auto. Certainly not the first commercially available auto. And when did this replacement finish exactly?

No, not the first auto or the first commercially available auto. When they were first invented, private automobiles were very expensive, very unreliable and very difficult to operate and maintain. It was hard to refuel them. The paved road system was limited and of poor quality. That's why the first motorized urban transportation was electric railways. Automobiles only started to take off after Henry Ford and others introduced the assembly line, mass production and practical, affordable vehicles. We had to create a whole new urban infrastructure of roads, parking lots, gas stations, stoplights, etc. But even then, automobiles took off amazingly fast. The transition to driverless vehicles will almost certainly be even faster, because they don't need any expensive new infrastructure and because computer technology is already cheap and is getting exponentially cheaper.

by Bertie on Mar 6, 2012 5:24 pm • linkreport

Automobiles only started to take off after Henry Ford and others introduced the assembly line, mass production and practical, affordable vehicles.

So then, you're talking about 60 years after the invention of the modern automobile. The driver-less auto hasn't even been invented yet. So once it is, we can start the clock on its 60 years to dominance.

by David C on Mar 6, 2012 5:34 pm • linkreport

That article is from May of last year. Google has announced several key milestones since then, including the trips of over a thousand miles without any human intervention. The automakers have also provided more information and results about their own driverless vehicle projects. The field is advancing very rapidly.

And so the race is on! Who will be first to market? Autonomous cars or jetpacks??


http://shopping.yahoo.com/articles/yshoppingarticles/436/martin-aircrafts-commercial-jetpack-looks-to-take-flight/

Seriously, though. If we see *truly* autonomous vehicles in less than 20 years, I'll eat my shoe, a la Herzog.

by oboe on Mar 6, 2012 5:50 pm • linkreport

So then, you're talking about 60 years after the invention of the modern automobile.

You appear to have read only the one sentence you quoted. There are many other sentences before and after that one. In those other sentences, I explain why the transition to driverless cars is likely to take much less time than the time between the very first automobile and mass adoption of automobiles.

by Bertie on Mar 6, 2012 6:21 pm • linkreport

OK, oboe, since you seriously seem to think that jetpacks are a realistic analogy to self-driving cars, here are a few reasons why jetpacks are unlikely to become widespread in the forseeable future:

- Expensive ($100,000)
- Dangerous
- Requires certified, physically fit pilot to operate
- Extremely noisy and uncomfortable ride
- Rider is exposed to cold, wind, rain, etc.
- Limited speed and range
- No ability to carry passengers or cargo
- Will almost certainly remain banned for widespread use in populated areas because of extreme noise.

by Bertie on Mar 6, 2012 6:59 pm • linkreport

You appear to have read only the one sentence you quoted. There are many other sentences before and after that one.

No I read it, I just wanted to clarify. Let's summarize:

You: The transition from horses to the modern automobile is totally analogous to the future transition from the modern automobile to the robocar

Me: That transition took 60 years.

You: The transition from horses to the modern automobile is in no way analogous to the future transition from the modern automobile to the robocar

So I guess now I say...touche? Excellent point? I agree, the two are in no way analogous?

by David C on Mar 6, 2012 7:14 pm • linkreport

What exactly are the existing technical issues that must be overcome? I feel like any further back and forth -- without addressing *real* issues -- is pointless. Neither side is spending a lot of time dealing with facts -- instead, we've become sidetracked into a fairly abstract discussion about *new technologies*. Why not stick to the subject at hand?

Personally, I'd love to hear more from the engineering savvy about what the current impediments are. I've been looking over some very recent reporting, and there's not a lot to go on regarding that.

by treslexue on Mar 6, 2012 7:37 pm • linkreport

David C,

You're not summarizing. You're misrepresenting. The analogy I described was the rapid transition from horse-drawn vehicles to motorized vehicles, not from horse-drawn vehicles to automobiles. For the reasons I explained at length (cost, reliability, difficulty of operation, lack of fuel facilities, lack of road infrastructure, etc.) the time between the invention of the first automobile and the mass adoption of automobiles was much longer.

by Bertie on Mar 6, 2012 7:38 pm • linkreport

The analogy I described was the rapid transition from horse-drawn vehicles to motorized vehicles, not from horse-drawn vehicles to automobiles.

Oh you meant motorized vehicles. Like cars AND buses (invented in 1830) AND motorcycles (1885). Hmm...I don't see how that changes anything.

by David C on Mar 6, 2012 8:05 pm • linkreport

Oh you meant motorized vehicles.

Yes. Strangely enough, that's why I wrote "motorized vehicles."

Hmm...I don't see how that changes anything

It doesn't change anything. As I said, the transition from horse-drawn vehicles to motorized vehicles was very rapid. It was a huge shift in transportation technology in a short period of time. It's a precedent for another shift in transportation technology, from upcoming transition from human-driven cars to self-driving cars.

by Bertie on Mar 6, 2012 8:20 pm • linkreport

It doesn't change anything.

So then why make the distinction? Why make the point that you were talking about the "transition from horse-drawn vehicles to motorized vehicles, not from horse-drawn vehicles to automobiles" if it doesn't change anything? Either way it took 60 years from invention to dominance.

by David C on Mar 6, 2012 8:39 pm • linkreport

@David C: you've spend a lot of time saying why everyone else is wrong. Let's say we 100% agree to do exactly as you tell us to. What would you have us do?

by Kevin C on Mar 6, 2012 8:57 pm • linkreport

Why make the point that you were talking about the "transition from horse-drawn vehicles to motorized vehicles, not from horse-drawn vehicles to automobiles" if it doesn't change anything?

Because you kept falsely asserting that I was talking about the transition to automobiles, when I repeatedly said "motorized vehicles," not "automobiles," and even explicitly told you that "the first motorized urban transportation was electric railways."

Transportation technology can and has changed very rapidly. It only took about 30 years for motorized vehicles to almost completely displace horse-drawn ones. The evidence suggests that the transition to self-driving cars is likely to be similarly rapid. You don't seem to have any response to this point.

by Bertie on Mar 6, 2012 9:03 pm • linkreport

Cars will be programmed by the guys from Grand Theft Auto.

by JAY on Mar 6, 2012 9:03 pm • linkreport

What would you have us do?

If we want safer, more efficient cars we can have that with existing technology. Simply raise CAFE standards and mandate cars that don't go over 75 mph. Then add in tons of speed cameras. Ta-da.

by David C on Mar 6, 2012 9:51 pm • linkreport

Because you kept falsely asserting that I was talking about the transition to automobiles, when I repeatedly said "motorized vehicles," not "automobiles,"

That's because automobiles are motorized vehicles. And then you conceded that saying "motorized vehicles," not "automobiles" doesn't change anything.

But now, your point is that it was the development of the electric railway that spelled the end of the horse. Which is interesting, since the electric railway was invented around the same time as the modern automobile (1888). So, yet again it was another 60 years before the horse-drawn cart was "almost completely displaced" by the electric railway (which was oddly in it's own decline by then).

Of course if we're talking about true "motorized vehicles" it was nearly 250 years from the debut of the "motorized vehicle" before the horse-drawn vehicle was "almost completely displaced."

But why not quit this guessing game? Why not define the dates you're talking about? What specific event are you talking about that started this 30 year transition (and what year did it happen), and at what point are you considering the horse-drawn vehicle displaced?

What is the actual analogy you see as fitting for the future we're heading towards.

by David C on Mar 6, 2012 10:06 pm • linkreport

If we want safer, more efficient cars we can have that with existing technology. Simply raise CAFE standards and mandate cars that don't go over 75 mph. Then add in tons of speed cameras.

We will have even safer and even more efficient cars with new technology.

That's because automobiles are motorized vehicles.

But they're not the only motorized vehicles. That's why pretending I wrote "automobiles" when I wrote "motorized vehicles," and explicitly told you that the early motorized vehicles were electric railways, not automobiles, is a misrepresentation of what I wrote.

But now, your point is that it was the development of the electric railway that spelled the end of the horse.

No, it was the development of motorized vehicles that spelled the end of the horse. Electric railways were just the earliest kind of motorized vehicle that became dominant over horses.

Which is interesting, since the electric railway was invented around the same time as the modern automobile (1888). So, yet again it was another 60 years before the horse-drawn cart was "almost completely displaced" by the electric railway (which was oddly in it's own decline by then).

No it wasn't. I don't know where you keep getting this "60 years" from. We're not talking about the time from the date cars (or electric railways) were invented. As I have already explained to you, early cars were expensive, unreliable, hard to drive and required constant maintenance. There was little infrastructure for refueling them. Paved roads were limited and of poor quality. These and other problems slowed the spread of cars in the early years. That's why electric rail and then motor buses were the dominant early forms of motorized urban transportation. But even with these problems, car sales grew dramatically after the turn of the century. And none of these problems apply to self-driving cars. They'll use the same infrastructure as existing cars, and use computer technology that is already cheap and is getting exponentially cheaper over time. Which means they will probably displace conventional cars even faster than motorized vehicles displaced horses.

by Bertie on Mar 6, 2012 11:08 pm • linkreport

@Dave C. Is that what we want?

Taxpayers and the car-bound (the vast majority of people) might want to get more throughput from the same or less infrastructure. Sometimes by different road rules and geometries (the point of the video), but also by following more closely, drafting, or making virtual trains of cars.

Others will want to read email, eat, drink, shave, make-up, or phone or text on their commute. The same things that transit users enjoy, only in the privacy of their car.

Others will want to stay in their familiar-but-car-dependent suburban home into their old(er) age. The government may want to save on hospitalization by encouraging the independent lifestyles of those who'd move to a nursing home otherwise.

Google, of course, wants you to spend 2 more hours per day interacting with media featuring their ads.

And some may chose to solve the problems you note (economy and safety) through exactly the sort of multi-car coordination that the video envisions (and others not mentioned, such as when TCAS instructs near-colliding airplanes to take coordinated-but-opposite evasive maneuvers).

TCAS, autopilot, GPS/WAAS takeoff/landing, pulse-doppler radar etc, are the sort of enabling technologies used widely in airplanes today that could, when adapted to cars, address many of the desires above. Why be "either/or" when you could be "both/and" in looking for solutions?

The airlines have long ardently sought better fuel economy, higher throughput, and higher safety, and have taken a both/and approach and so have adopted sensors, robot helpers, and higher speeds (eg. Regional Jets), right alongside lighter materials and better engines. Ta-da?

by Kevin C on Mar 7, 2012 12:03 am • linkreport

@Bertie @David C

I'm very disappointed in this thread. Both sides devolved into a worthless debates about prior technologies (aided heavily by oboe's non sequiturs regarding technological advances as portrayed in pop culture). Either a debate is had based at least *somewhat* on facts, or it's worthless endeavor. We might as well admit that no one here really knows what they're talking about. Even the one guy who has a background in this specific subject (David C) isn't willing to argue the merits of an emerging technology based on where it actually is today -- as opposed to at some point in time in the past. He's talking about horses. Other people are talking about jetpacks. I'm waiting for someone to bring up Pegasuses...

My own guess is that the only reason that driverless technology didn't emerge until now (really, about 15 years ago), is that the computing power didn't exist to process real time data of the magnitude necessary. Can someone tell me that's not true? Because if so, I'd love to hear it. I'd love for someone to bring some real meat to this discussion.

If, on the other hand, this is a hardware issue not related to computing power, then I'll concede to the skeptics that driverless cars may indeed be the work of a generation to bring to fruition. If it's a software issue, there's no reason to assume the beta phase will last 20+ years.

I'll add -- apparently, the Google car has been tested in rain. Other driverless cars have a well. All it takes is a little bit of internet research to find this out. It seems that Google is bullish on their new toy, as evidenced by how they trot it out to the media. If this is "vaporware", then hopefully it is vaporware as much so as the Volt was, when it was decried in 2008 by critics as such.

by tresluxe on Mar 7, 2012 12:05 am • linkreport

@tresluxe: nicely said. The two good things about this thread can be 1) embracing the falling cost of computing power and 2) integrating it into our systems (including bikes and transit).

The aerospace world has implemented pilot-assist technologies because they are already cheap vs an aircraft and its economic uses. As costs fall, we'll see them in cars. GPS, lane-keeping and self-braking to name a recent three.

And the last frontier in airline safety is the part that hasn't been automated yet: the part where the pilots take wrong turns and taxi onto active runways. The solution? not lower speeds, but more sensors and automation.

Today, the robots can cut off the ignition of drunk drivers (its not a cheap or popular option today, but it is real and getting cheaper). Next they'll call OnStar for a cab. Eventually, they'll just drive you home. We'll get happier and safer each time.

by Kevin C on Mar 7, 2012 12:30 am • linkreport

Either a debate is had based at least *somewhat* on facts, or it's worthless endeavor.

I've presented numerous facts about the current state of commercially available automobile automation, about new automation features the carmakers expect to introduce over the next few years, about Google's driverless cars, and so on. I can provide citations for any of them. As you said, it's not exactly hard to find reliable information on this topic. There are plenty of reputable sources of information, including Google's own website and the websites of the automakers themselves.

The best and most comprehensive overview I've found of the current state and likely future developments of self-driving cars is Brad Templeton's robocars articles. I think Templeton does a great job exploring all important angles of the topic, including potential downsides and obstacles. Overall, he is extremely bullish on the prospects for self-driving vehicles. Most people just don't seem to realize how fast the technology is advancing.

by Bertie on Mar 7, 2012 12:42 am • linkreport

Kevin C, I'm not arguing that robocars are bad or should be opposed. Again, I'm all for them. I'm just far less confident that we will see them as the dominant vehicle on the road in my lifetiime. If we want the benefits they promise, we don't have to wait that long.

If we want more throughput, we can institute congestion pricing and use the revenue to improve transit. And transit will allow people to read, sleep etc...with current technology. It will also improve safety.

I'm not being either/or. I'm being now/later.

by David C on Mar 7, 2012 8:44 am • linkreport

...aided heavily by oboe's non sequiturs regarding technological advances as portrayed in pop culture...

Hmm. Obviously I'm biased, but previous examples of breathless techno-utopianism seem pretty relevant here. Very much an id quod sequitur.

As @Bertie pointed out, there are numerous minor kinks that need to be ironed out in order for personal jetpack travel to be cheap, safe, and quotidian...but the underlying technologies are all there. For example:

- Expensive ($100,000)

How much would the Google car cost if you were to try to buy it today?

- Dangerous.

Not if we integrated many of these same automatic safety systems you see in commercial aircraft, just as we're assuming will occur with autonomous personal vehicles.

- Requires certified, physically fit pilot to operate

Currently automobiles require a licensed, skilled operator as well. Integrating our jetpacks with off-the-shelf safety and navigation systems is a "no-brainer".

- Extremely noisy and uncomfortable ride

It would be trivial to implement some sort of enclosed passenger compartment, and obviously this first generation prototype will be noisier than the production version. Noise damping technologies have come a long way.

- Rider is exposed to cold, wind, rain, etc.
See previous.

- Limited speed and range

Speed and range are relevant for some applications, but not for others. A range of, say, 4 miles and top speed of 25 mph would be invaluable for someone commuting from Alexandria to downtown.

- No ability to carry passengers or cargo

Obviously this is a commuter solution, much like a Smart car, or motorcycle, so this is largely irrelevant.

- Will almost certainly remain banned for widespread use in populated areas because of extreme noise.

Reducing jet noise is a huge area of research in mechanical engineering departments across the country. And we've made incredible strides in the field over the last 20 years. We're about "85%" of the way there.

You should look at the history of the automobile sometime--for many, many decades they were prone to mechanical failure, were quite slow, and very very expensive. By any objective measure (or at least the metrics we're hearing from "google car" enthusiasts) jetpacks are juuust around the corner.

Til then:

http://youtu.be/73Onygnmltg

by oboe on Mar 7, 2012 9:11 am • linkreport

@tresluxe

A partial list of technical hurdles left to overcome:

1. Sensor perception and decision-making gets iffy in bad weather. These can include times when sunlight shines directly into a camera or in fog or mist. The cars have been driven in the rain. But all reports are that they need more interventions per mile. The same is true about night, snow, etc...

2. The vehicle can be thrown off by small changes in the environment. Sometimes the cars will treat a piece of blowing trash as it would a person and will suddenly stop.

3. Once something leaves the field of view it is forgotten. So, if a cyclist disappears around a bend in the road, the car will speed up.

4. A 2010 Volvo demonstration of auto-follow technology resulted in the car smashing into the truck it was supposed to stop behind. I bet they were pretty bullish on the technology since they trotted it out to the public. Also, the fact that ONLY Google is doing this speaks to the confidence automakers have in the effort - which would be low.

5. Currently Google estimates that it's driverless car would cost $150,000. So getting the price down is a technological hurdle. It may be that it is more like solar power - totally doable, but always cost prohibitive.

6. The solution may be environmentally unsustainable if it leads to more VMT. Electric cars hold promise, but they are also not "there" yet.

7. Robocars require human intervention on average every 1000 miles. That means if you read the paper on the way to and from work, you'll be dead by the end of the month. This needs to go way way down.

8. It is difficult to design a warning system for when intervention is needed. If the car knew, then it would intervene.

9. future autonomous vehicles will rely heavily on global positioning satellite data and other systems, which are vulnerable to jamming by malicious computer hackers.

10. Some trivial tasks for human drivers — like recognizing an officer or safety worker motioning a driver to proceed in an alternate direction — await a breakthrough in artificial intelligence that may not come soon.

11. How do robocars interact with rude drivers and pedestrians? Will they be held captive at intersections because the car is so polite and no one else is coming to a stop.? Still unknown or dealt with.

And there are surely more things that Google's engineering team isn't telling us. In general, technical issues aren't advertised.

by David C on Mar 7, 2012 9:17 am • linkreport

@Bertie, I apologize, but I've been instructed to no longer argue with you.

by David C on Mar 7, 2012 9:19 am • linkreport

Jetpacks, schmetpacks.

http://www.youtube.com/watch?v=TkyLnWm1iCs

The future!

by Alex B. on Mar 7, 2012 9:41 am • linkreport

David, thanks for the insight. It'll certainly be interesting to watch how the technology evolves. I'm beginning to think that some lawmaking may precede the advent of a consumer grade robocar -- as more states pass friendly driverless laws, more dollars/effort will be devoted to robocar R/D.

by tresluxe on Mar 7, 2012 10:10 am • linkreport

The Martin jetpack is commercially available, $100,000.

by goldfish on Mar 7, 2012 10:13 am • linkreport

@goldfish:

Cost of an automobile in 1900 was about $70k in 2012 dollars.

[Just for the record, I think the odds of ubiquitous personal jetpacks versus ubiquitous and fully-autonomous (as opposed to driver-assisted) automobiles in the next 30 years is about the same. Could it happen? Sure. Will it? I'm extremely skeptical. I think we tend to look back at massive technological advances in IT (and medical tech which is somewhat related) over the last 50 years or so, and extrapolate that growth curve to other fields. Maybe it's that stack of "Discover" and "Science 84" magazines in my parents' attic which tout interferon as the final weapon in the war against cancer, or commercially viable fusion, etc ad infinitum... Meanwhile, people are still getting chemo, we're stuck with dirty fission reactors, etc...]

by oboe on Mar 7, 2012 10:50 am • linkreport

@oboe: Profitable, practical utility will inspire a far quicker advance and widespread adoption for autonomous vehicles than it will for jetpacks.

by goldfish on Mar 7, 2012 12:27 pm • linkreport

@goldfish,

Well, you're likely right. I'm still not holding my breath for swarms of fully autonomous vehicles operating in an urban environment. Or them to depopulate the cities.

by oboe on Mar 7, 2012 1:46 pm • linkreport

@oboe: fully autonomous vehicles (will) depopulate the cities
Actually I think they will encourage the opposite effect. Part of the reason families need a lot of space is so they can park nearby. This problem is solved with robo-cars because they can be sent to and summoned from far away parking. The parking pressure and cost in the central part of the city will be substantially reduced, reducing the hassle of city living for those that need a car.

by goldfish on Mar 7, 2012 1:59 pm • linkreport

oboe, I agree with part of your point - which is that there is a certain survivor bias at play. We look at the truly revolutionary products that quickly supplanted the prior products (iphones, cell phones, DVD players, Digital cameras) etc... and we project whatever is being developed onto that path. What we ignore is the 35 products behind them that did not evolve quickly and did not change everything. Things like the roomba or the segway or the V-22 Osprey...

Which brings me to a story. In 1994 I saw a project manager from Bell helicopter give a presentation on the V-22. It was about how it was going to revolutionize transportation. [The V-22 is an airplane that can take off and land like a helicopter]. While developed for the military, they saw it having commericial applications. You would no longer drive to the airport. You would park at the mall and hop into a V-22 that would take you to the airport if you were travelling more than 300 miles or to another one of hundreds of V-22 ports within 300 miles. It would replace greyhound and Amtrak and make air travel more efficient. You could even commute by V-22. They were going to sell thousands of them. "Sure," he said "there are some technical issues to overcome like tilt-rotor harmonic flutter, but nothing that we can't over come." At that point the V-22 had already flown. He showed a video of one taking off from an aircraft carrier and flying around and then landing.

I ate the whole thing up. I was sure he was right. I mean, they had the prototype right. They just had to work out a few issues and get the price down. How hard could that be?

15 years later and they've sold just over 100 and exclusively to the US Military. I have not caught an Osprey to Ocean City.

Point being, things don't always work out. Even with a working prototype and enthusiastic design team. In fact, on average, technology usually doesn't work out. And even when it does it isn't always adopted as quickly as you think it should (like CFLs). So...I'm a skeptic of almost all technology until they actually get it working for realsies.

by David C on Mar 7, 2012 2:44 pm • linkreport

@David C: The cost (as of 2008) of the V-22 Osprey is $67M per aircraft. This excludes the development costs. That puts it in a different league from consumer items.

by goldfish on Mar 7, 2012 2:59 pm • linkreport

@goldfish, the belief was not that the V-22 would replace cars, but that it would replace most buses, trains and many other airplanes. So I'm not sure how your point is relevant.

by David C on Mar 7, 2012 3:07 pm • linkreport

@goldfish: but will it be ENOUGH utility? The thing I keep wondering is how much utility/benefit a driverless car has in an environment that is 95% driven cars + peds + bikes. What percentage of driverless cars must be on the road before it will be safe for the car's pilot to spend the commute reading Facebook and teleworking instead of monitoring for takeover situations (which will inevitably be more likely in commute situations)? I don't know, I don't think sitting there, not being in control, watching the robot car make decisions I may or may not agree with while not being able to completely tune out sounds like fun - or worth the inevitable premium car companies will stick on the tech and probably monthly user fees, too.

Then again, I think car-teleworking so I can live in a secluded SFH in Herndon but work in DC/Tysons/wherever sounds terrible. I either want the full telecommute or to have my commute for appointment-making and Angry Birds like any "sane" person... ;)

by Jen on Mar 7, 2012 3:15 pm • linkreport

I think this is a very long way off, if it ever gets implemented. What a massive project it would be! And times are tough.
I agree that people love their driving way too much to relinquish. How would you get the rush of pretending you're a skilled race car driver, or of running your big truck up the tail of the car in front of you like the big man you are?
A couple wonders: Would there be speed limits, or would the vehicles be allowed to travel as fast as the computers determine is efficient and safe? Also: Would the occupants of these vehicles be allowed to go ahead and enjoy an alcoholic beverage on their ride? I don't see why not.
I can say that human laziness and selfishness is the biggest problem on our roadways, so if this system would eliminate that, things might actually improve – overall. Hmmm.

by Tad on Mar 7, 2012 4:53 pm • linkreport

How much would the Google car cost if you were to try to buy it today?

I don't know. The question is irrelevant. Google's car is a prototype, not a commercial product. The cost of a prototype is not a meaningful guide to consumer prices, especially for computer and electronic products. The iPad prototype probably cost a million dollars. The consumer iPad costs $500.

Not if we integrated many of these same automatic safety systems you see in commercial aircraft, just as we're assuming will occur with autonomous personal vehicles.

What automatic safety systems in commercial aircraft? What safety systems do you propose could be added to the Martin Jetpack to make it as safe as a car, and how much would they add to its price? The current price ($100,000) is already far too expensive for most people.

Currently automobiles require a licensed, skilled operator as well. Integrating our jetpacks with off-the-shelf safety and navigation systems is a "no-brainer".

Getting a driver's license is easy. Most adults already have one. The required skill level is very low. And driverless cars won't require any skill or licensing to operate. You'll just tell the car where you want to go. Jetpack certification is rare and is likely to require a considerably greater skill level than driving. Qualifying would probably be very difficult for people who elderly, overweight or in poor fitness or health.

It would be trivial to implement some sort of enclosed passenger compartment, and obviously this first generation prototype will be noisier than the production version. Noise damping technologies have come a long way.

It's not a first generation prototype. It's commercial product that costs $100,000. If adding an enclosed passenger compartment were trivial, they'd already have done it. Instead, the operator has to dangle from a harness, exposed to the elements. A passenger compartment and seat would add weight and bulk. And there is little or nothing they can do about the noise. Fans and propellers are inherently noisy.

Speed and range are relevant for some applications, but not for others.

Speed and range are extremely important, especially when you're expecting people to pay $100,000.

Obviously this is a commuter solution, much like a Smart car, or motorcycle, so this is largely irrelevant.

No, it's hugely relevant. The fact that the jetpack cannot carry any passengers or cargo is yet another reason why very few people are likely to be willing to shell out $100,000 for one, even if they could afford it.

reducing jet noise is a huge area of research in mechanical engineering departments across the country. And we've made incredible strides in the field over the last 20 years. We're about "85%" of the way there.

No we're not. Jets are still extremely noisy. That's why we have strict regulations about planes taking off and landing over populated areas, and why there is so much opposition to the creation of new airports and flightpaths in populated areas. There is no technology for dramatically suppressing the noise of jet engines or turbines. Absent some technological breakthrough to massively reduce their noise, there's simply no way these jetpacks are going to be approved for widespread use in urban areas. It would be like hundreds of leaf-blowers running at the same time throughout the day.

by Bertie on Mar 7, 2012 5:00 pm • linkreport

1. Sensor perception and decision-making gets iffy in bad weather.

So does human perception and decision-making. A combination of radar, lidar and video, with 360 degree coverage, is going to provide far better sensing than any human driver. And computers are going to provide far better decision-making. They don't get drunk or tired or distracted. They don't make mistakes judging speed and distance. They have much faster reaction times. Driverless cars don't need to be perfect to save lives. They just need to be better than human drivers. That isn't really a very high hurdle.

2. The vehicle can be thrown off by small changes in the environment. Sometimes the cars will treat a piece of blowing trash as it would a person and will suddenly stop.

Please substantiate this claim. I have seen no reports of this problem for Google's car. Or for automated collision-avoidance systems that are already available in production cars, like Volvo's pedestrian detection and avoidance system.

3. Once something leaves the field of view it is forgotten. So, if a cyclist disappears around a bend in the road, the car will speed up.

Please substantiate this claim. It seems rather unlikely that engineers would design a system to "forget" objects that may involve a safety hazard, simply because they have left the current field of view.

4. A 2010 Volvo demonstration of auto-follow technology resulted in the car smashing into the truck it was supposed to stop behind.

I'm not sure what you mean by "auto-follow technology." The video you link to is a demo of a prototype crash avoidance system. A failure in a prototype doesn't mean much. That's why they're called prototypes. Volvo's crash avoidance system is now available on its production vehicles and has apparently reduced the rate of accidents by 25%

5. Currently Google estimates that it's driverless car would cost $150,000. So getting the price down is a technological hurdle. It may be that it is more like solar power - totally doable, but always cost prohibitive.

Google's driverless car is a prototype. The cost of a prototype isn't a meaningful indicator of consumer pricing, especially for electronic and computer products.

6. The solution may be environmentally unsustainable if it leads to more VMT. Electric cars hold promise, but they are also not "there" yet.

Furture cars, whether self-driving or human driven, will almost certainly be much more environmentally friendly than current models

7. Robocars require human intervention on average every 1000 miles.

Please substantiate this claim. The reports I have seen are that, as of last October, Google's fleet of driverless cars had completed trips in excess of a thousand miles without any human intervention. It's now working on completing a million miles without human intervention.

8. It is difficult to design a warning system for when intervention is needed. If the car knew, then it would intervene.

Huh? If the car is confronted with a situation in which it is not confident how to act, it can notify the passenger. I don't know why you think that would be difficult. But this would only apply to non-critical situations, like maybe deciding where to park. A self-driving car must handle any time-critical, safety-critical situation itself. It can't notify the passenger and wait for a reply when a split-second decision is needed.

9. future autonomous vehicles will rely heavily on global positioning satellite data and other systems, which are vulnerable to jamming by malicious computer hackers.

No, GPS data might be needed for navigation, but not for safety-critical control. That's done with on-board sensors.

10. Some trivial tasks for human drivers — like recognizing an officer or safety worker motioning a driver to proceed in an alternate direction — await a breakthrough in artificial intelligence that may not come soon.

I don't know why you think that. Even cheap digital cameras can now recognize human faces.

11. How do robocars interact with rude drivers and pedestrians? Will they be held captive at intersections because the car is so polite and no one else is coming to a stop.? Still unknown or dealt with.

Not true. This issue is already understood. Google even has a video showing its driverless car interacting with other (human-driven) cars at a 4-way stop and asserting its right to enter the intersection to avoid being taken advantage of by the other drivers.

by Bertie on Mar 7, 2012 6:10 pm • linkreport

Ahem...

"And despite Google’s early success, technological barriers remain. Some trivial tasks for human drivers — like recognizing an officer or safety worker motioning a driver to proceed in an alternate direction — await a breakthrough in artificial intelligence that may not come soon."

Just as one example.

by David C on Mar 7, 2012 7:15 pm • linkreport

Ahem yourself. I'm not sure why you think the editorial comments of a newspaper reporter, who apparently has no recognized expertise in either self-driving cars or artificial intelligence, should be considered a reliable statement of outstanding problems. One wonders how Markoff thinks cheap cameras can recognize faces and cheap video games can recognize arm and leg motions. I guess he's never seen the Kinect.

by Bertie on Mar 7, 2012 7:52 pm • linkreport

@David C: the belief was not that the V-22 would replace cars, but that it would replace most buses, trains and many other airplanes. So I'm not sure how your point is relevant.

[Deleted for violating the comment policy.] The V22 (wingspan 45 ft, total width with rotors 85 ft) will never fit in the current infrastructure to replace a bus, which fits in a 10-ft wide lane. It is not a comparable "drop-in" replacement for buses, like robo-cars are for regular cars. Moreover, the V22 cost is some 70 times that of a (very nice) bus. To adapt the V22 for buses would require a complete change of the built infrastructure, costing trillions.

Now could it replace other airplanes? Maybe if it used airports, because its cost and function is comparable to a normal aircraft -- but that would hardly be revolutionary. What about trains? Consider how they are implemented, connecting dense urban areas: I leave the rest of this exercise for you to complete.

So the point is relevant because, due to cost, the V22 can never replace the existing transportation modes.

by goldfish on Mar 8, 2012 8:20 am • linkreport

@Jen: but will it be ENOUGH utility?

The change can only occur gradually, to allow people to get comfortable with it. I think you are closer than you realize: for example, there are a lot of driverless trains, particularly in airports, these days.

Some early adopters will blaze the trail; most will hang back until they become good enough, and cheap enough, to make the plunge.

by goldfish on Mar 8, 2012 8:30 am • linkreport

"Driverless cars don't need to be perfect to save lives. They just need to be better than human drivers."

Except that the manufacturer would have liability in an autonomous vehicle (other than for maintenance-related accidents) and human operators have it for non-autonomous. Factoring that liability into the sales price is a threshold problem. Eventually, as the technology improves and reliable data on real-world incidences of accidents becomes available, that cost will come down to close to what regular liability insurance now costs (though you'll still need some for maintenace-related liability). But for the early adopters, there will be a big premium just for the liability insurance.

by Crickey7 on Mar 8, 2012 9:59 am • linkreport

The V22 will never fit in the current infrastructure to replace a bus, which fits in a 10-ft wide lane.

I haven't been clear perhaps. The idea being pitched at the time was not that the V22 would pull into bus stations and then drive down highways like buses. Nor am I talking about transit buses - only intercity buses.

Consider how they are implemented, connecting dense urban areas:

The idea was that the V22 would provide point to point transportation to city centers like greyhound and Amtrak do. The V22 can take off and land like a helicopter, which means that it can access city centers just like helicopters do now. so it could land in helipads and then fly point to point. In the same way it could replace Amtrak (Amtrak already competes with airplanes for traffic) and do so with far greater flexibility. For example it would much cheaper to add Phoenix to their manifest than it is for Amtrak to do so. Or to boost flights to New Orleans during Mardi Gras. Adding helipads to overly large parking lots like those at malls, which is what they proposed, would require very little infrastructure and would certainly NOT cost trillions.

I leave the rest of this exercise for you to complete.

This wasn't my proposal. This is what high-ups at Bell Helicopter thought would happen. So, there were some very smart, very educated people who thought they could make this work. Were they crazy for thinking it would happen? Maybe. But only because they overestimated how cheaply these could be made. And those who are saying that the cost of robocars will be competitive with current cars may be making the same mistake. We don't know because they haven't made a commericially available robocar nor have they priced what it would cost. Maybe it will be competitive, but maybe it will be like the Volt, which is struggling even with massive government subsidies.

If you don't think there are any hurdles left to overcome, then how come there aren't any robocars on sale right now?

by David C on Mar 8, 2012 11:43 am • linkreport

Driverless cars don't need to be perfect to save lives. They just need to be better than human drivers

True. But they have to be better at driving than I think I am for me to let it drive. And, unfortunately, everyone thinks they're a better than average driver. Which means robocars will have to be much better than average for people to buy them.

by David C on Mar 8, 2012 11:45 am • linkreport

This is what high-ups at Bell Helicopter thought would happen. So, there were some very smart, very educated people who thought they could make this work. Were they crazy for thinking it would happen?

Not crazy, just trying to sell a product. I've done it myself when pitching research or to build a prototype: you throw in everything, even the farfetched. It is the buyer's job to bring them back to reality.

If you don't think there are any hurdles left to overcome, then how come there aren't any robocars on sale right now?

I never said there were not unsolved problems. However, because of the existing driverless trains and the successful Google prototype, I think we are pretty close, much closer than 85%. Bear in mind Moore's law, which amazes me that it has kept going for 40 years. Difficult things become easy with far greater computer power.

by goldfish on Mar 8, 2012 12:13 pm • linkreport

Not crazy, just trying to sell a product.

And Google isn't? And who were they trying to sell the product to? A bunch of college undergrads?

Bear in mind Moore's law,

Moore's law doesn't make hardware cheaper. It hasn't made solar power competitive yet. Nor has it kept my laptop from crashing.

by David C on Mar 8, 2012 12:56 pm • linkreport

Bear in mind Moore's law...

Thanks for this--it got me looking for the "law" defining the phenomenon the skeptics have been flirting with:

"The first 90 percent of the code accounts for the first 90 percent of the development time. The remaining 10 percent of the code accounts for the other 90 percent of the development time."[1]
—Tom Cargill, Bell Labs
http://en.wikipedia.org/wiki/Ninety-ninety_rule

by oboe on Mar 8, 2012 12:59 pm • linkreport

David C: Moore's law doesn't make hardware cheaper.

Yes it does.

It hasn't made solar power competitive yet.

Of course not, because that is not a computing problem.

Nor has it kept my laptop from crashing.

1. Delete non-standard and outdated software. 2. Update your virus protection, and stop visiting dodgy websites. 3. Buy a new one.

by goldfish on Mar 8, 2012 1:17 pm • linkreport

1. Delete non-standard and outdated software. 2. Update your virus protection, and stop visiting dodgy websites. 3. Buy a new one.

i.e. "steer clear of potential 'edge cases'". No clear analogues in the real-world, of course. "Stop driving on those dodgy roads in dodgy neighborhoods!"

:)

by oboe on Mar 8, 2012 2:02 pm • linkreport

Goldfish, Moore's law only applies to integrated circuits and the number of transistors that can be placed inexpensively on it. It does not have to do with making LIDAR cheaper. Or any of the 10,000 other components that make up the robocar. So, No it doesn't. And linking to an article that doesn't even mention Moore's law does not prove your point.

Making the robocar cheaper is not a computer problem either.

by David C on Mar 8, 2012 2:23 pm • linkreport

True. But they have to be better at driving than I think I am for me to let it drive. And, unfortunately, everyone thinks they're a better than average driver. Which means robocars will have to be much better than average for people to buy them

It seems rather unlikely that "everyone thinks they're a better than average driver" or that most people who do think that will refuse to ride in self-driving cars. People routinely ride as passengers in cars driven by other people, even if they think they're a better driver. And self-driving cars almost certainly will be much better than the average human driver, anyway. Like most new technology, self-driving cars will probably be embraced most enthusiastically at first by younger people. Older people tend to be much more resistant to adopting new technology and new ways of doing things.

Goldfish, Moore's law only applies to integrated circuits and the number of transistors that can be placed inexpensively on it. It does not have to do with making LIDAR cheaper. Or any of the 10,000 other components that make up the robocar.

Lidar doesn't need to get much cheaper, though it almost certainly will. Lidar is already used in active cruise control systems in mass-market vehicles. The additional technology needed for self-driving cars consists mostly of computer processors and software. I see no reason to think that in mass production it will add more than a few thousand dollars to the cost of a car. Owners of self-driving cars will probably be able to recoup the entire additional cost within a few years from lower insurance premiums. But I think the vast majority of people will be willing to pay the cost of self-driving capability simply so they don't have to drive when they don't want to, even if there were no other benefits. I think there's probably a huge market just among people who want to go out and drink without having to worry about a DUI.

by Bertie on Mar 8, 2012 3:44 pm • linkreport

It seems rather unlikely that "everyone thinks they're a better than average driver"

At the risk of appearing combative and/or cynical, I think this is the most natural and likely thing in the world. In fact, the worse the driver, the more likely they are to think they're above average.

It's Dunning-Kruger all over.

by oboe on Mar 8, 2012 4:11 pm • linkreport

The biggest problem to self driving cars with be how quickly they can respond. I see a big problems with these cars and wild animals like dear in rural or non urban areas and in cities with the blind.

by kk on Mar 8, 2012 6:35 pm • linkreport

From wikipedia on Illusory Superiority:

Svenson (1981) surveyed 161 students in Sweden and the United States, asking them to compare their driving safety and skill to the other people in the experiment. For driving skill, 93% of the US sample and 69% of the Swedish sample put themselves in the top 50% (above the median). For safety, 88% of the US group and 77% of the Swedish sample put themselves in the top 50%.[25]

McCormick, Walkey and Green (1986) found similar results in their study, asking 178 participants to evaluate their position on eight different dimensions relating to driving skill (examples include the "dangerous-safe" dimension and the "considerate-inconsiderate" dimension). Only a small minority rated themselves as below average (the midpoint of the dimension scale) at any point, and when all eight dimensions were considered together it was found that almost 80% of participants had evaluated themselves as being above the average driver.[26]

by David C on Mar 8, 2012 9:23 pm • linkreport

The biggest problem to self driving cars with be how quickly they can respond. I see a big problems with these cars and wild animals like dear in rural or non urban areas and in cities with the blind.

Deer are scary, no doubt. Where there are many of them, the locals drive quite slowly. It is strange to get behind a local car in Northern Wisconsin driving 50 mph on an entirely open, clear, mostly straight road. Then a 400 lb deer with a 24 point, 4 ft-wide rack suddenly jumps out in front of you, then a doe, then another doe, then another doe, and you understand why they are driving so slow.

I think this danger will be lessened with robo-cars, which will not be lulled into complacency after driving many uneventful miles on such a road. Robo-cars use motion detectors to sense large things approaching from the tall weeds, that human cannot see at night.

by goldfish on Mar 9, 2012 9:54 am • linkreport

Here's a good article on the difficulties of getting to the point where people can sleep while the car drives. Highlights:

"John Leonard, a Massachusetts Institute of Technology mechanical and ocean engineering professor who led that university's team to a fourth-place finish in the 2007 DARPA Urban Challenge, thinks that major technological hurdles in robot perception need to be overcome before self-driving cars can be deployed on a large-scale."

"In order for the vehicles to function the route needs to be driven by a human ahead of time in one of the test cars and mapped using its array of sensors. This rich set of mapping data is then stored on a Google data center and a portion of it is loaded into the car's hard drive. The location of stoplights, school zones and anything else that is reasonably static is marked so the car will acknowledge them without having to interpret them in real-time."

Which means that these cars will need to access the internet to download these maps. That's fine (though labor intensive) with one car, but we lack the capability to do this with millions. And it relies on Google not making any errors with these maps, which any user of Google maps knows is not a guarantee. [I think they still have the Naval Observatory in SE].

by David C on Mar 9, 2012 10:10 am • linkreport

@David: I would think that you would actually like this technology, as it can be used to solve range and recharging problems that electric vehicles have.

by goldfish on Mar 9, 2012 11:10 am • linkreport

I would think that you would actually like this technology

I do like this technology. Show me where I said otherwise.

I just don't buy that robocars are going to start replacing current cars in any significant number within my lifetime (~50 years), so I'm too worried about it. No more than, let's say, voting rights of Americans living on the Moon.

by David C on Mar 9, 2012 1:18 pm • linkreport

Which means that these cars will need to access the internet to download these maps.

No it doesn't. Google's car is a prototype. Commercial self-driving vehicles may or may not use this mapping technique, and may or may not store the map data in the vehicle itself. My cheap dashboard GPS unit stores all its maps on the device. The only data it downloads over the network is traffic reports.

That's fine (though labor intensive) with one car, but we lack the capability to do this with millions.

Huh? We already have millions of wireless devices using 3G network connections. LTE ("4G") technology, which is just starting to be adopted widely, is much faster. And 10 years from now, there will be even faster wireless technologies.

by Bertie on Mar 9, 2012 2:08 pm • linkreport

Google's car is a prototype. Commercial self-driving vehicles may or may not use this mapping technique, and may or may not store the map data in the vehicle itself.

Every problem in the world can be solved with "technology that doesn't exist yet," but I'm talking about the current state-of-the-art, the one you keep holding up as an example.

by David C on Mar 9, 2012 3:00 pm • linkreport

If the google driverless car is a prototype and the actual production one doesn't use the same tech, then saying "these cars are already here!" as evidence that this tech is moments away from the market is completely false.

As for how great our cell phone networks are, have you never experienced a rally or a big event where cell phones cease to send and receive data? Because the interaction of driverless cars during peak periods presents that exact problem - tons of data being thrown around by a lot of users all in a small space at one time.

by MLD on Mar 9, 2012 3:13 pm • linkreport

If the google driverless car is a prototype and the actual production one doesn't use the same tech, then saying "these cars are already here!" as evidence that this tech is moments away from the market is completely false.

The Google car demonstrates that the tech for completely autonomous self-driving cars is already here, or very close to being here. No one is saying that commercial, mass-market driverless cars are "moments away." That will require further refinement of the technology, additional testing, legal and regulatory changes, and so on. It's probably going to be another decade or so before driverless cars are commercially available for general-purpose use on public roads. It's unlikely to be much longer than that. Limited-use driverless vehicles, for military applications or private road systems, are likely to appear first.

by Bertie on Mar 9, 2012 7:36 pm • linkreport

Add a Comment

Name: (will be displayed on the comments page)

Email: (must be your real address, but will be kept private)

URL: (optional, will be displayed)

Your comment:

By submitting a comment, you agree to abide by our comment policy.
Notify me of followup comments via email. (You can also subscribe without commenting.)
Save my name and email address on this computer so I don't have to enter it next time, and so I don't have to answer the anti-spam map challenge question in the future.

or