By on January 8, 2015

the-jetsons-car_100322637_l

“Mm, 2000. When I was a kid, we thought 2000 was gonna be like The Jetsons or somin’. It ain’t even The Jeffersons!”-Chris Rock

Most major auto shows, barring the Geneva Auto Salon, having some substantial connection to the automotive world in some way. Detroit. New York. Los Angeles. Shanghai. Tokyo. Paris. Frankfurt. So how did Las Vegas end up with two car shows?

It used to be that the SEMA show was the only place you could catch an automotive exec pawing at a young woman one minute, introducing her as “my niece” the next. But now that the Consumer Electronics Show has morphed into a de facto auto show, you can see that twice in a row, as well as disgraced Gawker editors awkwardly trying to pick up booth babes.

CES, as the WSJ’s Holman Jenkins notes, has become the “self-driving car show”. Jenkins’ piece takes the contrarian view on the self-driving car idea, which is that while the technology may exist, it’s never going to happen. In his mind, the auto makers are merely doing it to keep pace with Google, which will likely shutter its own program as shareholders get antsy about its massive R&D spending in the face of slowing growth.

In my opinion, Occam’s razor applies here. There are just too many obstacles to getting self-driving cars on the road, en masse, in our lifetime. Autonomous cars would require a near-complete overhaul of our roads, open up massively complex questions about liability and most of all, require a substantial shift in mindset by the American public, who, despite what the affluent, childless (not to mention just as provincial as any other American) Silicon Valley set may think may think, are not enamored with the idea of piloting self-driving electric vehicles that are shared on a fractional ownership basis or a setup similar to Zipcar.

The kind of disruption they dream about does not happen in a short time frame – and if they have the magic bullet, why haven’t they gotten started developing it already? To riff on the above Chris Rock quote – we don’t even have a decent network of alternative fuel stations (EV, hydrogen, natural gas, what have you). Tesla hasn’t been able to mass produce an SUV, let alone a volume product. Ford is selling 60,000 F-150s per month. When you are placing bets against a century-old pillar of America’s economy, and the way that the majority of Americans outside New York City and the Bay Area get around, you ought to remember who your counter-party is.

Nor is CES “the most important car show” either. Like every other auto show, the vast  majority of the automotive stories generated at CES remains within the walls of the automotive media, and auto makers are using it to get some free coverage and talk about incremental improvements to infotainment systems that continue to confound and frustrate a good many customers. But that’s ok. That’s how progress really happens. It’s not sexy, nor disruptive, but it sure is effective.

 

Get the latest TTAC e-Newsletter!

145 Comments on “Editorial: You’ll Be Dead Before Autonomous Cars Are Launched...”


  • avatar
    Landcrusher

    First, the car drives itself while the driver is holding the wheel and responsible for taking over like with cruise control. Then, after a few years of people getting used to this and them noticing the lack of dangerous mistakes, you get autonomous cars.

    It’s really not magic.

    • 0 avatar
      dal20402

      Yep, this.

    • 0 avatar
      Kaosaur

      I think it’ll be even more incremental than that. Human driver 99% of the time. The network of cars will communicate and make suggestions to save the driver gas and/or time. Route changes will be suggested. Google is about to eat Garmin & TomTom’s lunch. Then it will be situations like “You have an incoming call. Road conditions are clear, would you like me to take over driving during your call?” Then the car won’t even ask. Then “You had two drinks at the bar, would you like me to take you home?” Then people will just let the car drive them everywhere.

      • 0 avatar
        wmba

        Apparently you skimmed the editorial. They can’t even fill potholes or shore up failing bridges, and magically, self- driving cars, with no more new infrastructure beyond a Google App and a few sensors, will magically whisk you home after having had a couple too many after work.

        Call, no text me, in the decade this all comes to fruition and works flawlessly..

        • 0 avatar
          stuki

          They who cannot fill potholes, are the same they that by education/indoctrination monopolization help make Google’s task of equaling and bettering a human driver, easier by the day…

          • 0 avatar
            TheyBeRollin

            Ultimately, they’ll also automate so much that people will be required for so few things that our economy as we know it will collapse. Maintaining 12-lane-wide freeways won’t be important because nobody will want nor need to travel them. That’s the real future.

            Today we’re just holding together the “high tech” road network that our great grandparents built because it was their form of our automation push. Just the concept of getting around in a personal automobile that generally started when you wanted to go somewhere and would get you there at 25 mph was the stuff of science fiction.

          • 0 avatar
            hybridkiller

            People were making these doomsday predictions decades ago vis-a-vis the decline of manufacturing, which as a share of GDP isn’t just a US phenomenon, it’s a global one (less than half of what it was 50 years ago).
            It’s what is commonly referred to as the transition from a manufacturing economy to a service-based one, and it’s been going on since the post world war era.
            Maybe 20 years ago the prospect of automation increasingly displacing human workers was a real concern, but we’ve seen massive workforce growth in other sectors – like healthcare – more than take up the slack.

            With regard to not needing “12-lane-wide freeways” anymore, that might be true if we were talking about some huge popular shift to public mass transit, but the introduction of self-driving cars does nothing to suggest fewer cars on the road.

          • 0 avatar
            Kruser

            Hybridkiller, ten years ago I’d say you were spot on, but the nature of this is very different. I am worried because in my son’s lifetime, there will be AI coupled with robotics that will do 90% of what humans are capable of for next to nothing… and they’ll do it 24/7. Here’s a nice video that wraps it up. http://www.youtube.com/watch?v=7Pq-S557XQU
            This will be a transformation akin to the agricultural revolution, but it will happen in a relative blink of the eye. Ultimately, mankind will be better off without scarcity, but the transition will be absolutely wrenching.

          • 0 avatar
            hybridkiller

            That’s a nice video. I especially enjoyed the little bit about horses. Unfortunately I stopped watching at that point – all conjecture and no data makes for a not very compelling argument. Kudos to the bright college student who made that video, I’d give him an A for effort, but he needs to go back to his economics class and start over.

            At the risk of repeating myself, every generation has people predicting the end of life as we know it courtesy of technology, and time after time they have all been proven wrong. The economics of human productivity is not a zero-sum game. Not only has our society adjusted to the ever-shrinking need for manual human labor since the start of the industrial revolution, economic growth and prosperity have in fact thrived.

            But you go ahead and worry about that if you want, it’s not like we have any real threats like climate change, water resource issues, terrorism, etc.

          • 0 avatar
            Kruser

            Well, it is a shame you didn’t watch more. I think he boils it all down nicely. BTW, the creator is an expat science teacher who has since quit his day job to make explanatory videos.
            I think there is quite a lot of data with regard to technological unemployment that has many prominent economists (e.g., Summers, Krugman) worried. Others (e.g., Mankiw) blame technology for increased income inequality, which I’d argue is a leading indicator of things to come.
            Ultimately, I think the rise of a general AI at human-level intelligence will be the way we will solve the issues you bring up, but the complete disruption of our ways of life will be the tough part.
            I’d be happy to back up my assertions with a $100 Simon–Ehrlich type wager that ten years from now, in the US, unemployment will be higher, workforce participation will be lower, and income inequality will be higher. I don’t, however, have a quantitative means for judging whether or not people will agree that tech is the root cause.

          • 0 avatar
            hybridkiller

            If you can tell me that he provided even one shred of hard economic data to back up his/your premise then I will go back and watch the whole video. I’m not holding my breath on that though.

            Who knows, perhaps you will be vindicated someday, but for now it’s all pure conjecture and speculation, with essentially no long term historical data to support it. As I’ve already said, if anything, history actually contradicts such predictions.

          • 0 avatar
            Kruser

            If you are keen to see the data, I think The Second Machine Age and Race Against the Machine by Brynjolfsson and McAfee are both good sources. They are also both good reads IMHO. I sent the video because it says many of the same things, but more quickly and entertainingly.
            BTW, my wife also thinks I’m overstating things, but in her case it isn’t based on the Luddite Fallacy… she just thinks that automated systems are inherently unable to progress beyond a certain level of complication.
            Kevin Drum has an interesting take on it here that explains why I think this will all happen rather suddenly. http://goo.gl/4Bsqh4
            Here’s another recap from the Economist.
            http://goo.gl/q2bkdz

          • 0 avatar
            hybridkiller

            I’ll just say this: I don’t so much need to see data as I need to see a compelling argument. So far I haven’t see one – just some vague predictions based on a lot of unsubstantiated assumptions.

            The Economist article you linked to does contain a lot of data, but nothing that points to some clear economic long term downtrend. Quite the opposite, they seem to be saying essentially – ignore the past, this time it’s different.

            Despite the number of unemployed and impoverished at the bottom of the economic food chain, even they have a remarkably better standard of living than the working class of a century ago.

          • 0 avatar
            Kruser

            I’ll take one last shot at overcoming your doubt and then I’ll leave it be. :) Here’s an Atlantic article (3rd of a series, excerpted from Race Against the Machines) that has both some good arguments and some data to back it up. http://goo.gl/v9z4JK
            The crux of it is that given competition from automated substitutes, wages will tend to go down to remain competitive. But at some point, no human will be willing to work for the incremental cost of a computer. To tie it back into the horse argument – horses didn’t get less good at manual labor, the competition just got much better and cheaper.
            As Kevin Drum says, “The Luddites weren’t wrong. They were just 200 years too early.”

          • 0 avatar
            WheelMcCoy

            Like @hybridkiller, I also wanted to stop watching the video shortly after the horses segment. But I forced myself to watch the whole thing and found the tone of the video rather smug.

            The video also missed a point about the stock market being run by robots. Lauded as efficient, it was those very robots that were responsible for the May 6th 2010 Flash Crash where Proctor and Gamble stock went down to a penny and then shot back up. Humans were needed to stabilize the market.

            I found the CGP Grey video on Coffee more interesting:
            https://www.youtube.com/watch?v=OTVE5iPMKLg

            Ironically, he cited that coffee beans on the same plant ripen at different rates and therefore must be picked by *hand*. Is it economical to develop specialized robots to pick coffee?

            And then there’s irony. Will computers be able to understand it? Will they be able to write comedy? At best, I think robots can be unintentionally funny like GM’s Melody Lee.

            Here on TTAC, I read reviews about cars and lumbar support. Can robots even review products meant for humans to use? Robots don’t even have a lumbar, and feel no pain.

            It’s also a misconception that computers consume little energy, require no sleep and don’t complain. Google has an estimated 1 million servers which consume 220 megawatts of power a year. Small when compared to world consumption, but it does take a share of energy. Hard drives fail regularly and need to be swapped out. Viruses need to be defended against, servers go down for maintenance and reboots.

            If a computer or computer network became as complex as a human brain, I’d wager it would respond just about as fast a human and have occasional lapses, like a human. And if it crosses the boundary to sentience, I’d be pleased to make its acquaintance.

        • 0 avatar
          mx5ta

          I live in an area in upstate NY where the village/city does an excellent job of snow removal, so there’s no question about their competence, but still, sometimes the potholes are bad. The local newspaper did a good story explaining the situation: basically, lots of thawing and freezing creates cracks, allowing water to further creep under the top layer; that freezes, causing further damage. Then, to fix it, there are two different kinds of material: the one can be put in when it’s still winter, but the drawback is that it quickly gets torn back up. The more permanent stuff can be put in in the spring, and it tends to last a lot longer.

          • 0 avatar
            WheelMcCoy

            Last winter in NYC, potholes were created as fast as they were patched. The constant plowing of snow scraped away the lane markers on the FDR drive. So we drivers were skittering around potholes and had no lane markers for guidance.

            This would be a good test for a self-driving car. But for the tech to really work, all cars would need to be able to talk to one another — “you go left around the pothole, I’ll go right… SUV, you go thru it.”

            The technology exists, but wide-spread adoption is needed for it to work. In the meantime, limiting self-driving cars to campuses, airports, and dedicated highways seems to be the most realistic approach.

      • 0 avatar
        TorontoSkeptic

        I used to work in the “location intelligence” business and it’s definitely true that Google is crushing the traditional GPS/mapping/GIS providers. Navteq, TomTom, Garmin, Nokia, Mapinfo, ESRI are all having trouble because Google is a) giving maps away largely for free b) integrating into mobile way better than most could have been imagined c) providing genuinely new and innovative products (Google Earth, Android built-in nav).

        HOWEVER it is a long leap from that to self-driving cars. Technically I don’t doubt it’s possible, especially with what’s possible with computer vision/AI/signal processing. But it’s more of a political and financial problem. Think about the issues around liability (who’s responsible for crashes?), prioritization (who gets to go first) and privacy. Those are all legal or political problems and our system isn’t too great at sorting those out quickly and in the public interest – see net neutrality, NSA spying, the marvel of the TSA, etc.

        • 0 avatar
          Kaosaur

          I guess i have to explain myself more because the general response to my comments stems from misunderstanding. We’re unquestionably “not there yet”.

          What I was trying to point out was that all of real hurdles preventing us from having self-driving cars are social/political ones. Attitudes of car users are what’s going to drive progress. It is ultimately a function of resource allocation (time and money). Things that are thought to be impossible or incredibly difficult are provably doable with the political will to put the resources behind them.

          Getting to space was impossible but that will allowed insanely expensive projects to achieve incremental gains in a very short scale of time.

          If we really really want self-driving cars, we could have a reasonable approximation in as little as 15 years with the full resources of a large state behind it. It’s clearly just not that huge of a problem.

  • avatar
    Menar Fromarz

    ah..self driving cars. despite assertions that it may not, should not or never will happen, the whole thing reminds me of the “safety” autos of the early ’70s. Bloated, slow crash cages that rational driving types thought as a tech show with limited practical usefulness. Well guess what? Most of the ideas and thinking are now part and parcel of our autos today. Crush zones, High Strength Steel, the use of alloys and composites, antilock braking, collision avoidance etc. All were Jetsonville at one time. Personally, semi-autonomous is ok, with all its current crop of on board tech that is NOW, but who knows, if the selfii crowd want Hal2000 to pilot their “car pod” to Starbucks so they can spend the time online instead, maybe the future will be here sooner. In the meantime, not for me, nor i am sure, for the insurer of my car, who probably will be the final determiner of viability of the rollout of the technology.

    • 0 avatar
      carguy67

      Incremental improvements–ABS, crumple zones, etc.–happen by evolution. For an example of what would be required for ‘self-driving’ cars check out the FAA’s ‘NextGen’ project:

      http://en.wikipedia.org/wiki/Next_Generation_Air_Transportation_System

      Note that almost all newer aircraft can already ‘self-fly,’ and many can ‘self-land’–with CAT III equipment and pilots–and think about the boondoggle self-driving infrastructure would create.

      We can’t even get decent public transportation built; you really think we’ll redo the entire road grid any time soon? Notgonnahappen.

      • 0 avatar
        dal20402

        Self-driving cars can happen through evolution. No redoing of the entire road grid required.

        We’ll only need to redo the road grid when/if we want to use self-driving cars to increase speed or road capacity beyond what we have today.

      • 0 avatar
        bunkie

        I was waiting for someone to bring up NexGen and it’s a good example of the difficulty involved. The skies are about as “crowded” as some backroad in Nevada 50 miles from anywhere, the data-gathering and reporting required for cars would be a couple of orders of magnitude larger. Aircraft operate in 3 dimensions, not two making separation easier (a car can’t climb to avoid traffic). The equipment required to implement NexGen is staggeringly expensive. The FAA has mandated ADS-B (a core component of NexGen) by 2020 for all aircraft operating in controlled airspace. It’s a real problem because the required equipment is worth more than many of the small piston-engine airplanes now flying (older Cessnas and Pipers like the airplane I fly).

        What we are more likely to see is limited automated driving on some roads in designated lanes, that’s a lot more achievable.

  • avatar

    I would hesitate to bet against most technologies (nuclear fusion is a major exception), and I have powerfully conflicted feelings about self driving cars. Driving is one of my second favorite physical activities, and I don’t want to see it become obsolete, or confined to special tracks, but self-driving cars would be a godsend to those who can no longer drive. I will read Holman Jenkins’ piece with great interest (I worked on the same publication with him about 30 years ago). Thanks for bringing this to our attention, and for the hilarious Chris Rock quote.

  • avatar
    pkov

    Can’t come soon enough. Americans are only becoming more numerous, impulsive and uneducated while our roads crumble. So naturally they’re given 300+ HP cars.

  • avatar
    Joebaldheadedgranny

    Also a great point was made here about infrastructure. Before driverless cars are fully adopted we’ll need to be sure that new highway/road construction imbeds the technology to facilitate the sensors/lidars/cameras required to make it work. That is a trillion dollar project, an obscene amount of money, but it has to be invested whether we end up using driverless cars or not.

    • 0 avatar
      carguy67

      Bingo. Technology has advanced in aviation lately at a blistering pace, pilots are, by and large, much more capable than any drivers, NextGen has advantages for everybody yet it’s turned into a massive boondoggle.

      As a nation, we don’t seem willing or able to tackle these kinds of challenges any more; maybe the high speed rail project in California will get things moving, maybe not.

      • 0 avatar
        Landcrusher

        Not so much. The past 15 years have seen some fairly nice improvements made possible by the tech used in autos for ABS and other traction control features being combined with better and cheaper GPS tech. Your 2015 plane isn’t really that much different from your 1975 plane. Self landing was achieved and then put on hold like 30 years ago. NextGen is primarily a risk and cost shift from the FAA to the aircraft owner which is being accepted because it does allow for denser use of airline and bizjet traffic in some congested areas. The auto analogy for NG is forcing all car owners to put GPS based tracking and broadcast in their cars so that the government can save money on road stripes and signs and enforcement while pulling in more fines.

    • 0 avatar
      stuki

      A Silicon brain doesn’t strictly need sensors different than a biological one to make the same decisions and take the same actions. Barely cellphone brained dimwits somehow do manage to make it home from Walmart even without lidar.

  • avatar
    John R

    Well, apparently Google still needs to figure out how to get the things to work in the rain, snow and parking garages. Sounds like by the time they crack it’ll be just in time for my dementia to settle in.

  • avatar
    panzerfaust

    I daresay that the biggest hurdle in this is not technological, nor is it infrastructure, its legal. You can bet your ass that any manufacturer who is seriously considering autonomous vehicles is also considering what it will be like to be on the receiving end of the mother of all class action suits when stupid people figure out how to do stupid and deadly things with their products.

  • avatar
    panzerfaust

    The biggest obstacle is not technology or infrastructure but legality. Any manufacturer who is considering autonomous vehicles must also consider the possibility of the mother of all class action lawsuits when their system has its version of defective airbags, or faulty ignition switches. This will be a new field of endeavor for attorneys who will pick it up quickly.

  • avatar
    dal20402

    Derek, I think you’re imagining a different kind of self-driving car from the one that’s almost here. Your headline will only be true for readers in their 70s.

    “Autonomous cars would require a near-complete overhaul of our roads,”

    Not if they evolve in the way they already are. Each year, luxury cars get a bit more autonomy. First there was adaptive cruise control. Then there was automatic obstacle recognition and panic braking. Then there was lane keeping assist. Next up (and already working on prototypes) is intelligent hazard recognition, now intended to help drivers see hazards through augmented-reality views, but also pointing the way toward better intelligence when the computer is driving. Now, engineers in the lab are working on the most difficult problems — like construction zones, damaged roads, and emergency situations. It’s just a matter of time before the computer can handle all of those substantially more reliably than a human. And they will be introduced piecemeal, just like they have been to date. The computer will slowly take over more and more of driving — when it’s able to be fully autonomous, it will feel like an anticlimax.

    “open up massively complex questions about liability,”

    Google and the automakers already are paying lots of money to very intelligent lawyers working on this. They will reserve in accordance to the risk, and the legal system will figure things out a decade or two after the technology starts to be in widespread use, just as it does with any other potentially deadly technology. Liability worries haven’t once stopped society from making a technological leap.

    “are not enamored with the idea of piloting self-driving electric vehicles that are shared on a fractional ownership basis or a setup similar to Zipcar.”

    That’s not how the technology will get into the public domain at first. It will be on private vehicles similar to the ones we have now. The proliferation of autonomous-driving technology is a separate question from alternative fuel, and I’d bet a lot more money on autonomous driving within our lifetimes than I would on alternative fuel.

    In the very long term, Car2Go-like models are inevitable for more and more of the continent, as it grows and needs to get more urban. But that’s a 100-year time horizon, and autonomous driving is coming much sooner.

    When there starts to be a critical mass of vehicles with the capability, then a tipping point will come where people will want the faster speeds, higher road capacity, and lower congestion of autonomous-car-only roads. That’s when the biggest changes will happen — and that will take place after the technology is already largely in place.

    • 0 avatar
      sportyaccordy

      I agree with a lot of this. Ghosn said it best- we will see more and more driving “assists” rather than full on automation. I think we will see some of this on the commercial side as well to help long haul truckers too. And as you hinted to, a new S550 can pretty much drive itself, at least on the highway… all that’s keeping it from doing so indefinitely is liability/legal hurdles. It’s coming.

  • avatar
    orenwolf

    Sorry Derek, You’re wrong here, IMHO.

    People thought that about aoutpilots in Aircraft (and then fly-by-wire aircraft without flight sticks!). Aerospace-grade navigation is becoming cheaper due to commoditization, and that tech will eventually trickle down over the years (like apollo-era tech did).

    Computing power is insane. My phone has more capability and RAM than my (gaming!) desktop did less than ten years ago, facial, image, and text recognition are now simple enough to do on your desktop, and robotics has taken massive strides (seen those segway-like wheelchairs that can climb stairs?)

    Then look at new cars today. Radar cruise control and lane departure warnings, are the first step towards autonomous cruise control. That IMHO will be step 1. Then automatic driving while the driver remains available to react. Cars can *already* tell if the driver is paying attention or not (see Mercedes). Then, finally, full automation. Critical mass isn’t that far away.

    The liability factor? I think many people forget that automobile collisions are the single greatest cause of fatalities today. they are *already* a liability.

    Be wary of what can be done “in a lifetime”. My grandmother started her career working in a one room schoolhouse without plumbing or electricity, and now she has an ipad.

    • 0 avatar
      JMII

      Agreed we are not that FAR away. Saw an article yesterday about Audi’s car automatically driving itself from LA to Vegas during the long highway sections. Between radar cruise, auto braking and lane departure along with some logical software (move over faster traffic ahead/behind) its pretty close. Cars already park themselves so its just the middle ground (aka city driving). Granted that part seems impossible due to some many factors, but as mentioned my iPad is more powerful then a 5 year old laptop. In addition the data gathering (aka mapping) has become so advanced that all the car has to do is avoid hitting things, the routing part is nearly done as Google can now tell which lane to be for an upcoming turn.

    • 0 avatar
      carguy67

      Avionics used in instrument flight regimens are required to be carefully calibrated, updated, inspected and certified on a regular basis (IIRC, the GPS database HAS to be updated every 90 days, or it’s not legal for flight in controlled airspace).

      Are we going to trust Joe Prius driver to maintain those standards, or will we need an army of new government employees to govern and enforce those requirements?

      • 0 avatar
        beastpilot

        Nope.

        Aircraft are allowed to fly in the USA with only a single check once every two years (altimeter), and even then that’s only some planes. No periodic calibration required by law for all the other stuff.

        Databases are almost always updated every 28 days for REGULATORY reasons, not safety. They move an airspace, not an airport. Anyway, yep, you can’t use an old database, but it pops up and says so. Grab a USB stick and do it in 5 minutes, or wait a year and it will do it by itself off the 4G network.

        And that GPS? Not legally needed at all. Believe it or not, it’s legal to fly a plane in the USA with NO INSTRUMENTS AT ALL. Seriously. A totally blank instrument panel. Legal, and has been done many times.

        Aviation is not a great example of amazing regulations leading to perfection. Unless you are a pilot, I get that the world thinks aviation in the USA is this insanely regulated thing, but for small aircraft it is basically the wild west. I’ve flown from Washington to Kansas before without talking to a single person in flight nor letting anyone know where I was going before I took off. It’s a car. In the sky.

        Source: Pilot, and oh yeah, I design avionics for a living.

        • 0 avatar
          Landcrusher

          As a former aircraft broker I have to strongly disagree. It’s only the Wild West if you actually only want a plane that flies in the Wild West. If you want to use a plane to go places and keep it at an airport near where you live, you are very likely to be forced into a very regulated world of expensive annuals, ramp checks, bumpy rides and delays at low altitudes, etc. The world has changed, and the aviation industry has been constantly evolved to the point of only retaining two types of owners. The unusually regulation compliant types and those who simply ignore the regs.

          Because of this, most people who actually choose to buy a plane are either very wealthy or don’t live in big city centers anymore or both. The cities keep growing, but their pilot populations are stagnant. The ability to fight back gets reduced as the pilot population becomes an ever shrinking percentage.

          Eventually, they will come for everybody if everybody keeps getting suckered into all this group politics nonsense.

          • 0 avatar
            28-Cars-Later

            I would add the cost of becoming a pilot -outside of the military- is quite prohibitive for all but the most wealthy and/or determined.

          • 0 avatar
            beastpilot

            Sorry. I live in Seattle, fly out of an airport under the Class B ring for Sea-Tac, and the only thing I need to do so is a transponder. I routinely take tours around Seattle, right over the Space Needle, and talk to nobody. I need no permission to take off. The lower I stay, the more places I can go without any limits.

            I regularly fly from Seattle to Reno, NV, Tucson, AZ, Oshkosh, WI, and a bunch of other places without talking to someone.

            Flying for 15 years, never been ramp checked. My dad has been flying for 40 years, 4,000 hours, never been ramp checked.

            This is not to say there is a not a lot of regulation. What I was saying is using aviation as an example of what automated cars may become is probably not the example the poster desired. The generic idea that aviation ALWAYS has this immense set of rules holding it down is not true. The US is actually very open in terms of aviation, and a daily, routine flight requires very little extra work than operating a car.

            The original poster was talking about Avionics and their re-certification requirements, not the vehicle as a whole. Aircraft do have annuals, but those hardly look at the avionics. Plus, I built my own plane, so, no annuals. Yea!

          • 0 avatar
            Landcrusher

            The cost to get a certificate actually isn’t that bad if you take into account that most of that cost is plane rental. If you don’t enjoy flying, then it certainly is going to seem expensive.

            Beast,
            Experimentals get a bit of a break, but your plane will soon require ADSB to leave your airport. You also haven’t gotten caught in the changes by the IRS. Also, you clearly are flying for fun. People who need to be places have to fly Instrument, file, chart up, procedure up, practice, etc. This requires a lot of dealing with regs and every change is time and/or money.

            Working on airport tarmacs gives you a much different perspective on ramp checks. I would see an inspector working an airport at least once every couple months. Planes parked on ramps were the primary targets because I assumed the FAA politics is such that they prefer to catch other offices’ violators.

            My DA40 was grounded twice for certification issues because the local guys believed the certification guys should not have allowed the POH to pass muster or some other nonsense. I was strongly advised to comply by the older instructors and business owners at the field.

            You can add a power outlet yourself. I had to have mine disabled at annual, and then, if I want it back, have to buy a similar part, have it checked by and A&P, and all the paper work processed and kept. All this after my original one had passed very rigorous certification, was serial numbered at the factory, and never found to have a fault.

            That’s just one example.

            You are basically flying in the Wild West both figuratively and geographically. You are likely living outside of the inner ring, or you are living a good ways from the plane. You are approved as the builder to work on your own plane.

            Now that I have given my argument, what model is your little baby?

          • 0 avatar
            Landcrusher

            Reply eaten.

            Short version, learning isn’t expensive if you subtract the cost of the flying.

            Beastie,
            No instruments, no LLC, suburban airport equals Wild West.

            What did you build?

    • 0 avatar
      danio3834

      >People thought that about aoutpilots in Aircraft (and then fly-by-wire aircraft without flight sticks!).

      Indeed. Especially after a so-equipped plane landed itself into a forest.

      • 0 avatar
        beastpilot

        You referring to the crash where a human flying the plane flew it too low, manually overrode the computer systems, and he applied power too late?

        This is like blaming stability control for letting you understeer off the side of the road when you go into a 25 MPH corner at 100 MPH. And you had turned it off 10 minutes before.

        • 0 avatar
          danio3834

          That’s the story Airbus wanted everyone to believe as it would have sunk them if it were anything to do with the systems on their latest aircraft at the time. There was a good deal of dispute over whether the aircraft failed to respond/overrode pilot inputs.

          Either way, overconfidence in the aircraft’s abilities to manage itself, lack of feedback and resulting pilot inattention/disorientation has been noted as the cause of other incidents as I’m sure you’re aware of as an aviator. My point isn’t that automation is inherently bad, but that it can foster laziness, inattentiveness and a lack of skill on the part of the operator when human intervention really is needed.

  • avatar
    Volt 230

    Yeah and in the late 60’s they predicted we would have moon bases set up already and we don’t even have a space program any longer, futurists and financiers don’t see eye to eye, I guess?

    • 0 avatar
      jhefner

      Yeah, that was the problem with “2001 A Space Odessey”, Star Trek, and other science fiction movies; someone has to pay for the spaceships and starships; and the United States is finding out it can’t do it alone. A joint program like the ISS is close; but still billions and billions away from what the movies told us the future was.

      It has not helped that neither nuclear fission nor even the prospects of fusion has made electricity “too cheap to meter.” A lot of stuff is getting cheaper; but not cheap enough to pull of the future as seen in science fiction.

      SSTs ran into the same OPEC fuel cost barrier that cars did; along with environmental issues; space planes are still too prohibitively expensive to happen; even with the military standing behind them. Much of the rest of the future (warp travel, artificial gravity, cloud cities, etc.) would require unthinkable amounts of energy in reality in addition to the technology.

      I think like others that self-driving cars will happen gradually and slowly, just like it has in aerospace, as the cost of the technology comes down, and it is designed in increments.

  • avatar
    jmo

    “heavier-than-air flying machines are impossible”

    Lord Kelvin – 1895.

    • 0 avatar
      Lie2me

      That was about the same time there was a call to close the patent office, because everything that could be invented had already been invented

    • 0 avatar
      Mathias

      “heavier-than-air flying machines are impossible”

      I thought it was Aristotle.

      Either way, that’s a poor statement from any scientist.
      They must have known it was a dumb statement.

      Consider the pigeon.
      Heavier then air, flies just fine.

      • 0 avatar
        Lie2me

        He didn’t actually say that, he was a brilliant man, but a bit opinionated. What he said was

        “No balloon and no aeroplane will ever be practically successful.”

        Which is an entirely different mistake

  • avatar
    Richard Chen

    Google likely will be dogfooding their 25mph autonomous car on campus in the coming months. I wouldn’t be surprised to see them in other low-speed situations soon afterwards – think The Villages in FL where only golf carts are allowed.

    Once they get up to speed, we’ll get to the trial attorneys suing to stop autonomous vehicles from destroying their livelihood. And if you’re a fractional as opposed to a sole owner: surge pricing!

  • avatar

    The WSJ article is well worth reading.

  • avatar
    Pch101

    “Autonomous cars would require a near-complete overhaul of our roads, open up massively complex questions about liability and most of all, require a substantial shift in mindset by the American public, who, despite what the affluent, childless (not to mention just as provincial as any other American) Silicon Valley set may think may think, are not enamored with the idea of piloting self-driving electric vehicles that are shared on a fractional ownership basis or a setup similar to Zipcar.”

    I don’t see the roads or the mindset being issues. If anything, I suspect that the public would embrace driverless cars — they would be the coolest thing since the iPhone.

    I do foresee a technology problem. It’s tough to operate in a two-dimensional environment when there are so many factors operating independently. A plane operating on autopilot has a much easier time because there are fewer causes for surprises and a lot fewer opportunities for unpredictable events.

    Even if the driverless technology works reasonably well, it probably won’t be good enough; some of the exceptions will be significant enough that human intervention will need to be available as a backup. I think that we’ll end up with semi-autonomous cars that do more of the work but that require a driver behind the wheel to respond to the surprises.

    • 0 avatar
      dal20402

      The surprises really won’t end up being that hard to manage, IMO. Google is well suited for figuring them out, because it is the best in the world at accumulating and analyzing big data. All it will take is lots of hours of driving, during which lots of surprises will happen, and then analysis of the surprises. Very few situations when driving are so unique that they won’t be captured by this method, and if the computer gets confused only in those few remaining situations, it will still be far more reliable and less crash-prone than a human.

      It will be awhile before we can entirely get rid of the controls that allow a human to drive, but it won’t be long before the human only has to drive once in a blue moon.

      • 0 avatar
        Pch101

        “Very few situations when driving are so unique that they won’t be captured by this method, and if the computer gets confused only in those few remaining situations, it will still be far more reliable and less crash-prone than a human.”

        That may be true, but that isn’t the proper way to assess risk.

        It’s a marginal cost question. There will be some incidents that could have been avoided by having a human behind the controls.

        It’s those incidents that necessitate having the people stay behind the wheel. If you end up with fewer incidents with a semi-autonomous car than with a fully autonomous one or if some of those incidents are less severe, then semi-autonomous is the way to go.

        • 0 avatar
          dal20402

          In the long term, you have to balance the cost of those few incidents (which will be very few indeed after a substantial amount of experience) with the costs of still having humans controlling cars. There will be huge benefits to autonomous-only roadways. Faster speeds, more lanes in the same space, much closer following distances, and the abolition of time-wasting traffic signals will all be possible. But to make any of that happen you can’t have any humans directly controlling vehicles on those roadways.

          I agree with you that we’ll have human control for a long time, but I don’t think it’s where we’ll end up.

          • 0 avatar
            Pch101

            It isn’t about marginal benefit, either; it’s about marginal cost.

            And it doesn’t include all of the costs. Granny not being able to use the autonomous car because she is unlicensed is a cost to Granny, but it isn’t a social cost that needs to be amortized back to the rest of us. It’s her price to pay, and we don’t worry about it.

            It’s the same reason that we don’t have fully autonomous planes. Even if the automatic plane can do the job well 99% of the time, it’s the 1% that we worry about. The cost of the 1% is unacceptably high, even though it is a pretty low figure in comparison to the whole.

          • 0 avatar
            dal20402

            “It really isn’t.”

            I can only assume you’ve never had an elderly relative stuck in the suburbs, even those with paratransit (which is unreliable and *maybe* good for a few trips a week), or had (or been) a kid who participates in after-school programs.

            It really is.

          • 0 avatar
            Pch101

            “Lack of mobility for elderly people and children who can’t drive is a real social problem”

            It really isn’t. We can give a subsidized Dial-a-Ride to the former and school buses for the latter. We really don’t care much about that otherwise.

            The need to prevent the incidents that the machine couldn’t prevent is the primary consideration for risk management.

          • 0 avatar
            Pch101

            “Lack of mobility for elderly people and children who can’t drive is a real social problem”

            It really isn’t. We can give a subs1dized Dial-a-Ride to the former and school buses for the latter. We really don’t care much about that otherwise.

            The need to prevent the incidents that the machine couldn’t prevent is the primary cons1deration for risk management.

        • 0 avatar
          Landcrusher

          The main benefit of full autonomy for consumers is the highway. If you have semi auto on surface streets and reliable auto on highways, don’t you get almost all the benefit with much less of the risks?

          I’m not nearly as interested in the Internet or movie on the way to the store as I am on the way to the next town or commuting.

          I think the exurbs will get a real boost from this stuff.

          • 0 avatar
            dal20402

            I disagree strongly with this. In my city, the highways actually work reasonably well, while there is gridlock on many city streets during most rush hours. Fully autonomous vehicles would open up those city streets considerably.

      • 0 avatar
        jhefner

        “It will be awhile before we can entirely get rid of the controls that allow a human to drive, but it won’t be long before the human only has to drive once in a blue moon”

        I have a serious problem with this, on two fronts:

        1. The computer may be driving along with the driver half asleep, drunk, or otherwise distracted, and hand off control too late for the driver to take appropiate action

        2. Turning driving into a “once in a blue moon” situation will cause drivers to loose their skill sets; when they are called on to drive; they will simply be “out of practice”, and most likely called upon to do so when the situation is already tough.

        My best sample case for this is Air France Flight 447. From what I understand, the plane was flying itself in the middle of the night through severe weather; with the pilots basically dozing at the controls. When the pitot tubes froze over, the flight computer could not figure what to do and basically shut down, leaving the plane to the pilots. Two half-asleep pilots now found themselves in total darkness, in a tunderstorm, with no instrumentation; they likely took the wrong corrective action that put the plane into a high speed stall, and caused it to crash in the Atlantic with all aboard.

        I could see a similiar possible situation with an automomous car; perhaps finding itself traveling at high speed into a severe traffic (chain reaction collosion say) or weather condition (sudden snowstorm or black ice conditions), becoming overwelmed and handing control to the driver; who may then need to be fully alert and possese professional skills to get themselves out of it. They may have neither and end up crashing; where if they were driving and paying attention, they may have slowed down and prepared themselves much earlier. (And yes, they not have, either.)

        • 0 avatar
          WheelMcCoy

          My thoughts exactly. A once-in-a-blue-moon driver won’t have the skills to be backup driver to the computer. That would be a switch though… humans assisting computers instead of computers assisting humans. :)

          Popular Mechanics did a write-up Air France Flight 447. The junior pilot was on duty and the 2 senior pilots were roused from their rest. While they tried to sort out what was happening, the junior pilot kept the yoke back in an attempt to keep the nose up, but that really stalled the engine. In hindsight, they should have gone into a dive (push the yoke forward) to restart the engines.

        • 0 avatar
          dal20402

          I think in that sort of situation the car won’t end up handing control back to the driver at high speed. It will pull off to the shoulder or other safe place and then say “You have to get me around this and I’ll tell you when I can take control again.”

    • 0 avatar
      danio3834

      This right here. Many in aviation have found out the hard way that automation can’t counter all variables either. Many an incident of “pilot error” was cause by learned laziness and lack of training for such situations because the systems were presumed too intelligent.

      • 0 avatar
        hybridkiller

        I don’t see the aircraft analogy as a good parallel. Flight is MUCH more complex and involves a lot more variables.
        Worst case scenario with the car in the event of either a software or a hardware failure would be something akin to “limp mode” that safely brings the car to a halt as quickly as possible. That’s simply not an option when you’re airborne – if something goes wrong in the air you can’t just “stop” the plane.

        In any case I’d have more confidence in a well-programmed (and well-tested) self-driver to NOT hit me than I have in most human drivers.

        • 0 avatar
          Glenn Mercer

          I see it somewhat differently. In an emergency situation the average airliner has several minutes, typically, before hitting the ground. And the people in the cockpit are a) highly trained and b) being paid to pay attention. In a car on the highway, when an impending accident situation emerges that the AI cannot handle, there are typically only going to be 1 or 2 or maybe 3 seconds before the crash occurs, and the driver is a) not necessarily highly trained (look how easy it is to renew your driver’s license in the USA) and b) not being paid to pay attention. In fact, the person may have purchased an autonomous vehicle specifically so he or she DIDN’T have to pay attention. I can imagine my elderly mother being told by some synthesized voice from the dashboard: “Handing control back to you, unmanageable problem detected.” This is why Google itself switched from high-speed test cars to one limited to 25 mph: it found the failure mode at higher speeds much too frightening. Notice how Google’s own press about the car has shifted from emphasizing safety to emphasizing mobility and convenience, showing elderly or handicapped persons tootling around in low-speed situations. So I agree that the aviation analogy is wrong, but I come to a different conclusion than you do as to why it is. (grin)

          • 0 avatar
            hybridkiller

            Point taken, it’s just that I’m assuming a level of software sophistication that wouldn’t require human intervention at all. In fact I think this notion of the car having the option of saying “sorry Dave, I don’t got this” isn’t realistic. We already have distracted drivers crashing cars, I can’t imagine requiring someone to pay attention “just in case”. The average person has to be actively engaged in order to maintain any useful level of alertness – just ask any law enforcement or military personnel.
            And don’t forget there are some, shall we say, fairly simple-minded and/or inexperienced and/or impaired people behind the wheel of giant projectiles hurtling down the highway @ 70+ mph as we speak.
            The computing power to keep a stealth fighter in the air (because they are so unstable that it’s impossible for a human to manually fly one) has been around for many years, so I’m thinking getting a computer to maneuver a car reliably without hitting anything isn’t a big stretch.
            But I agree we ain’t there yet.

        • 0 avatar
          Glenn Mercer

          (To your latest post, AM January 10: Agreed!)

    • 0 avatar
      PandaBear

      The problem is not mathematical modeling of the 3d road condition, the problem is sensing what imperfection is on the road (like a fallen piece of building material, icy road, animal jumping out, supersize pot hole, etc), as well as human drivers nervousness and trying to take control of the automated car in a panic (i.e. the Toyota unintended acceleration).

      Let’s put it this way, human are superb at seeing in all sorts of light condition as well as feeling unusual vehicle / road condition due to 16+ years of motor skill training and millions of years of evolution result in our DNA. It will take a while and a lot of money to get electronics to do just as well.

  • avatar
    APaGttH

    I was told we would have jet packs by now.

    www dot adultswim dot com / videos / robot-chicken / jet-pack-testing

  • avatar
    harshciygar

    Nothing needs to be changed in regards to our infrastructure, as cars are being developed to drive themselves on today’s roads. Volvo and other automakers are working on a “smart highway” concept that embeds small magnets in the roads for cars to follow, but the technology being developed is wholly independent of our infrastructure.

    Now that doesn’t mean it’ll be ready for primetime in the next decade or two, but paraphrasing a grey-haired WSJ writer who is obviously biased against self-driving vehicles hardly seems like a good pedestal to stand on. Google seems pretty determined to make it into the auto industry, and I don’t think their autonomous car experiment is just going to disappear overnight, especially since they’re supposedly placed an order for over 100 to test around SF.

    The only people who aren’t into the idea of autonomous cars are Luddties and Baby Boomers. Remember when PopSci was filled with dreams of autonomous cars? People would much rather read their Kindle behind the wheel than be forced to stare at some idiots taillights all the way to work and back. But besides “purists” and Luddites, once the legal complexities are out of the way, I don’t see many barriers standing in the way of autonomous car adoption.

    I think you’re off point here…but at least you’re writing about cars, not writers :P

    • 0 avatar
      Kruser

      Agreed with regard to the infrastructure changes. I’m constantly amazed by statements like the authors statement, “Autonomous cars would require a near-complete overhaul of our roads”. Huh? This is being done today, on existing roads with the traffic we have.
      I don’t agree with your timelines though. I think we’ll have this commercially available within the next 10 years… even if it isn’t in the US. There will be some country that will be itching to be the first and will find a way to push through the legal hurdles and hand wringing.

  • avatar

    Well, your last ride will need to be autonomous. No view from inside that oak casket.

  • avatar

    Again my posting is not erh… posted.

    Your last ride will need to be autonomous. No view whatsoever from inside that oak casket.

  • avatar
    Glenn Mercer

    One somewhat subtle point about the autonomous car debate. I have heard at least two ADAS supplier executives make this point, including one from Continental. And that is (and forgive me, I have trouble explaining this one clearly): when we move from autonomous level 2 or 3 to level 4 or 5, the challenge of managing the car “inverts,” and becomes exponentially more complex. That is, assume I have a nice L3 Mercedes, with active cruise control, lane departure warning, electronic brake assist, etc. Each one of those ADAS features has to monitor some very well defined parameters (e.g. where is the lane marker, and am I approaching it?), and then react accordingly. Complex software of course (a typical car has more lines of code than a Boeing Dreamliner), but eminently feasible. Now jump the car to L4 (fully autonomous some of the time) or L5 (autonomous all of the time). The challenge “inverts:” from monitoring a specific subset of parameters, the car AI must graduate to monitoring EVERYTHING the car does. As the Conti exec put it, now it has to know traffic rules, weather, traffic density and flow, whether the parking garage barrier is down or not, if that lump in the road is a branch I can run over or a brick I cannot, whether the driver on the cross street is waving me into the intersection, whether the stoplight has a burnt-out amber light or not, etc. The challenge is probably still doable, but it is quantitatively and qualitatively vastly harder.

    Thus my own view is we will get to 95% of the safety benefits without going to 100% driverless. Because it is much easier to pick off, one by one, each road hazard, than to tell the car to take over all of driving’s tasks (including hazard management). And paradoxically, this will slow the arrival of L4 and L5 cars: if L2 and L3 ADAS continue to drive down fatality rates (yes, most fatalities are caused by human error, but remember that humans are actually VERY GOOD at driving: we’re at about 1 fatality per 100,000,000 miles right now in the US (no typo)), the safety benefits of going all the way to L4 and L5 become less compelling. (Now there is a whole other discussion to do with convenience benefits — admittedly I am focusing on safety since the industry so far has focused on that also.) So the future might be EFFECTIVELY autonomous without being FULLY autonomous. (And as I say to people arguing for L5 ASAP, I ask them if they would board an airplane that had no pilots on board, but I know that is a bit of a cheap shot.)

    I think autonomous vehicles are inevitable, but wonder if they will get here as quickly as the enthusiasts expect. From a safety point of view, will they be like Concorde: it flew faster than anything else, but that performance edge was not worth the added cost. Will 95% autonomous vehicles be the “good enough” technology that delays the arrival of the 100% autonomous “best” technology?

    • 0 avatar
      wmba

      Great comment. I think you may well be correct.

      We have the non-technical types who live in good weather areas who think fully autonomous is just sround the corner, and doubters like me who can envision all sorts of events these robots couldn’t handle.

      Clouding everything is Google, which is much less smart than it thinks it is – cannot even get the correct responses to queries half the time. Friends are amazed how I can frame questions to get decent links out of Google. So if this level of outright dumbness to a mere query is present to this day (and it is getting slowly better), sorry, I don’t want this outfit driving my car for a while.

      Incremental is the way to go, with systems designed by people not so convinced of their utter brilliance as Google, which is a provincially small-minded outfit stuck in the infrastructure and norms they operate out of.

    • 0 avatar
      Nedmundo

      Excellent post. I watched a PBS show about drones last night, and some analysts made essentially the same point when analyzing whether drones could become fully autonomous. It’s very difficult to surmount those final hurdles when trying to remove human judgment from the equation.

    • 0 avatar
      burgersandbeer

      Good post and I agree with you and everyone else saying autonomous cars are an incremental effort.

      I do take issue with crediting the 1 in 100,000,000 mile fatality rate to human skill though. I’m giving that to improvements in cars safety and emergency response/medicine. If we could find statistics showing a downward trend in property damage claims per mile, that would be more compelling evidence for me.

      • 0 avatar
        Glenn Mercer

        Oh I entirely agree, the 1 in a 100,000,000 rate is NOT at all due to sheer human ability. Car tech, road tech, so much more is involved. I only brought it up because the more enthusiastic autonomous vehicle proponents (and I call myself a “pragmatic proponent”) cite the 35,000 traffic fatalities a year often, but don’t put it in context. 35,000 deaths sounds horrible (and it is!), 1 in 100,000,000 miles sounds wonderful (and it is!), but only by taking the two together do we get a complete picture. Which is, “Autonomous vehicles will be able to reduce the 35,000, but the magnitude of the challenge they face in doing that is made clear by the fact that on average we drive 100 million miles between fatalities.” When people cite the 35,000 number without providing context, it gives an unbalanced picture (IMHO).

        (And just to natter on a bit more, given that about 5,000 of those deaths are motorcycle deaths, and we are not expecting autonomous technology to touch motorcycles for a long time, the CAR death rate might be closer to 1 in 110,000,000 miles!)

  • avatar
    Crosley

    It’s 99% a litigation issue. I really don’t think the technology obstacles are all that great. For the most part, it seems they’ve already figured it out.

    I have ZERO doubt that self-driving cars’ track record vs “people” driven will not even be close. 99% of accidents are caused by driver error. But just like Toyota got roiled in a ridiculous lawsuit with “unintended acceleration” expect companies with deep pockets to have all sorts of litigation issues that have nothing to do with any real negligence and are about shakedowns and companies wanting to avoid damaging PR.

    Where I really see the benefit to this technology would be the trucking industry. The costs would quickly be recouped.

    • 0 avatar

      I was just going to bring this up. Self driving cars take us from driver error to manufacturer error. One has to believe launching a retail version would be a legal gamble.

      • 0 avatar
        Kruser

        Google has argued that the liability should rest with the manufacturer. This makes sense to me and should not be as big of an issue as people are assuming. First off, these cars will be safer (likely by an order of magnitude) than human drivers. Second, when there is an accident, there will be an excellent data trail to assign liability and, more importantly, make improvements.
        So, let’s assume Google puts out a car and it kills someone. They pay a big settlement, which car manufacturers already do, and then they improve the system to make the chance of that error much less than it was. One doesn’t have to have too many errors before the system is exceedingly safe. The scale of liability over the long run will be much lower than manufacturers have today.

    • 0 avatar
      dal20402

      Professional risk management staff, working with lawyers, can figure this stuff out. They do for other risky manufacturing ventures, and they do for insurance companies. It’s not unprecedented and not that hard conceptually, although it requires a lot of analysis.

      • 0 avatar
        Crosley

        Too many people die every day on public roads, it’s just a unique endeavor. Just the nature of millions of drivers making mistakes.

        There would be even more deaths with non-automated cars, but the first “Google car” that kills someone is going to be an epic PR and legal nightmare.

    • 0 avatar
      e30gator

      Think about the commercial airline industry where the lives of millions of pepole depend on the automated systems of these vehicles. There are liability issues here for the companies and manufacturers as well, but this has mostly been worked out.

    • 0 avatar
      PandaBear

      This is the reason they will first roll it out in long haul interstate highway by tailgating another human driven car first, for non human transport, before roll it out for human transport.

      When they have enough accidents to test out different ideas, they’ll know how much risks there are.

  • avatar
    Volt 230

    And who. pray tell, is gonna pay for this modern, updated roads? as it is, we don’t have enough money to maintain or repair our collapsing infrastructure, even fixing potholes, please stop dreaming.

  • avatar
    Master Baiter

    “You’ll Be Dead Before Autonomous Cars Are Launched”

    I’ll go farther: your children and grandchildren will be dead too.

    I watched Blade Runner last night. A 1982 film set in 2019 Los Angeles. It featured silent, flying cars and robots so similar to people that it took special skills to identify them.

    • 0 avatar
      Lie2me

      Well, that’s a good 4 years away, so you know, could happen

      I’m still waiting on the Transporter

      http://www.outsidethebeltway.com/wp-content/uploads/2014/06/Star-Trek-Transporter.jpg

      “Beam me up, Scottie”

      • 0 avatar
        WheelMcCoy

        “I’m still waiting on the Transporter”

        I thought about this… a transporter pad will work with maybe a landing party of 6?

        When this initially goes commercial, it will be very expensive and a means a travel only for a certain clientele (think Concorde SST). When the price comes down and we can teleport in bulk, imagine the long lines waiting for the transporter room. Then the redshirts will slow things down even further as they search you and your bags. Yeah yeah, Scotty can check the pattern buffer for explosives, but that still takes time.

        Welcome to the 23rd century.

    • 0 avatar
      Lie2me

      Well, that’s a good 4 years away, so you know, could happen

      I’m still waiting on the Transporter

      http://i.telegraph.co.uk/multimedia/archive/01729/star_1729224c.jpg

      “Beam me up, Scottie”

  • avatar

    This is all bringing us, step by step, closer to the TTAC:Fiction piece from 2011 by Jack, The Blockers: https://www.thetruthaboutcars.com/2011/05/fiction-the-blockers/

  • avatar
    JohnnyFirebird

    I think that smaller nations with a more compact infrastructure will see them first. Urban zones may have EV / autodrive only areas even in North America.

    Japan autodrive only by… what… 2030?

    • 0 avatar
      PandaBear

      They would only need subway and buses with good drivers. It is a lot more flexible and speedy to have a good professional driver than automation (except when they go on strike).

  • avatar
    OneAlpha

    Autonomous cars?

    Thanks, but until it’s possible to make a real-world version of KITT, I’m not interested.

  • avatar
    e30gator

    I see this becoming a real thing, but only on certain roads (major highways, etc) before the entirely autonomous car is a thing. This is doable and no, you wouldn’t need to create an entirely new infrastructure to do it, just imporove vehicle monitoring/tracking/driving systems that are already available to some extent.
    Personally, I would trust a computer driving the car beside me on the interstate a lot more than I trust the 20 year old girl texting on her phone not to make a mistake.

  • avatar
    PandaBear

    As someone who has been working on automation equipment for a few years and cars for a few years (back in school as student projects), and a few years on consumer electronics. I think that autonomous car will be here after a long time. The reason is that it is much easier to change the environment than to train a bunch of computer with sensors for the perfect skill. Highway will be adding emitters that autonomous car latched on first so they can do cruise control without human involvement, then it will be the mandatory features like auto following of the cars ahead of you, and finally road building standard for all but gravel roads.

    And you can bet that when accidents happen, there will be lawyers all over trying to bleed automakers to death, so it will take a long long time before enough people are killed on the beta testing that we will finally toss out the steering wheel.

  • avatar
    ihatetrees

    Autonomous cars will first appear in Asian (and maybe Euro) cities, where there are differing views on liability. (I bet Singapore is 1st).
    Combining autonomous cars with an Uber-like app should result in huge improvements in traffic flow and efficient land use.
    Once demonstrated in one urban area, the obvious benefits will sell themselves.
    Of course, in many statist areas (like France or NYC), rent seeking public transit, taxi, and trail attorney lobbies will resist ferociously and may successfully kneecap any change.

  • avatar
    toxicroach

    Hah. Nobody will every buy anything on the internet when they can drive to the store and see it in person!

    In the future, computers will shrink until they can fit in a single room!

    30000 people die each year in the US because of drunk driving. Hundreds of millions of people spend money they can’t really afford on a car to get stuck in traffic for an hour or two a day. Some people are really awful drivers.

    The liability issue is a non-starter really. The car comes with an all encompassing liability policy for computer screw ups that protects the owner and the manufacturer. Should be cheap, since it won’t be causing many accidents, especially compared to the burrito-eating, make-uping, texting, drunk, old, and stupid drivers we all deal with every day.

    The advantages of self driving cars are huge. Lives saved. Traffic reduced. Hours of rush hour traffic gone, or at least turned into productive time. Elderly people able to keep themselves out nursing homes for longer.

    It really depends on whose lifetime you mean, but it’ll happen and it will happen fast. There’s a reason why boring cars sell, and it’s because most people think driving is a chore. Once this is real device, and people have had a few years of them not driving off the nearest cliff, the switch will happen at about the same speed smartphones got adopted.

    • 0 avatar
      beastpilot

      You mean 10,000, right? 30K is the total deaths in cars, about 1/3 are alcohol related.

      http://www-nrd.nhtsa.dot.gov/Pubs/812102.pdf

      I don’t agree on liability. For some reason, people expect airplanes (even little ones) to be perfect. Any crash must be the aircraft’s fault. Aviation companies get sued into bankruptcy over and over, and the jury keeps making huge payouts. Even when the part wasn’t involved in the accident!

      When it literally becomes true that it wasn’t your fault, then the companies are going to eat serious lawsuit losses every accident. We already know what that looks like- airplanes cost $300,000 each new, and about 50% of that is liability insurance.

      In fact, between 1985 and 2000, most companies did not sell small aircraft. It took a literal act of congress to limit aircraft liability to 18 years after it was sold to get them to get back into the business.

      Come to think of it- doesn’t the car company need to pre-load the car with all the liability insurance at the sale? Meaning if a car lasts 15 years on average, you need 15 years (say $10K) of insurance all in the sales price?

      I’m not against autonomous cars at all, and I think they are coming. But I think the liability issue will be one of the larger ones to solve.

      • 0 avatar
        Glenn Mercer

        I completely agree with beastpilot. (Despite his or her odd handle!) (grin)

      • 0 avatar
        dal20402

        A large proportion of aircraft accidents result in death, which you can assume will lead to a claim for multiple millions in damages per decedent. Although there are way more car accidents, way fewer of them proportionally result in death. Most car accidents result in claims for a few *thousand* dollars in damages. It will be way easier to insure autonomous cars than airplanes once risk managers figure out the actual risk involved.

      • 0 avatar
        Kruser

        Yes, the manufacturers will be assuming much more of the liability, but there are ways to fail gracefully with a car that aren’t necessarily available to an aircraft (e.g., slow down, pull over and shutdown). Also, if this is a service, they can just offer it in certain low-risk conditions. For instance, Google is limiting their first gen cars to 25mph.

        • 0 avatar
          Glenn Mercer

          On your first point, at the last AUVSI conference a GM speaker made the reverse argument: at 30,000 feet an airliner has minutes to deal with a bad situation… cars on the highway may only have seconds. On your second point, I agree entirely: Google pivoted away from highway-speed vehicles to low-speed vehicles because (as Google itself said) they couldn’t figure out how to handle the car-to-driver transition at high speed: when the car’s AI can’t handle something it has to give control back to the driver, and at 60 mph few humans have the reaction times to be alerted, understand the situation, and take action. Notice how Google’s own statements have shifted from the safety argument (avoid crashes) to the convenience and mobility argument (empower the elderly, the blind, etc., to move about).

          • 0 avatar
            Pch101

            “at 30,000 feet an airliner has minutes to deal with a bad situation… cars on the highway may only have seconds”

            The autopilot’s job is also easier because anything that is detected nearby can be considered some sort of threat and worth monitoring.

            In urban traffic, there are all kinds of things that could theoretically pose potential risks, yet don’t. Too many false positives make the driverless technology less useful.

            Ultimately, this may end up being some fancy variation of cruise control and lane control technology, which gets switched on and off and always requires a driver behind the wheel who can intervene when needed. You can forget having those three martinis and reading your tablet while driving.

          • 0 avatar
            Pch101

            “at 30,000 feet an airliner has minutes to deal with a bad situation… cars on the highway may only have seconds”

            The autopilot’s job is also easier because anything that is detected nearby can be cons1dered to be some sort of threat and therefore worth monitoring.

            In urban traffic, there are all kinds of things that could theoretically pose potential risks, yet don’t, such as nearby pedestrians who could bolt into your path but probably won’t. Too many false positives make the driverless technology less useful.

            Ultimately, this may end up being some fancy variation of cruise control and lane control technology, which gets switched on and off and always requires a driver behind the wheel who can intervene when needed. You can forget having those three martinis and reading your tablet while driving.

          • 0 avatar
            Landcrusher

            PCH,
            What model AP can do that? I’m only aware of AP that will shut off under threat, and unaware of any being networked to traffic alerts anyway. Terrain, sure, but traffic? I’m not ATP or even commercial so I just don’t know.

            I suspect Airbus might have a solution, but their stuff is so problematic I prefer anything flying over an Airbus. If we let the Airbus guys design autonomous vehicles the pedals would get switched around to prevent drivers from attempting to go manual.

          • 0 avatar
            Pch101

            What I am saying is that anything in airspace that shouldn’t be there can be considered to be a threat that requires attention. Pilots are alerted and can act accordingly; planes fly with a certain cushion of empty space around them and pilots know to be aware of any encroachments into that zone.

            That cushion isn’t available on the street. The electronics for the driverless car have to be more discerning.

          • 0 avatar
            Pch101

            What I am saying is that anything in airspace that shouldn’t be there can be cons1dered to be a threat that requires attention. Pilots are alerted and can act accordingly; planes fly with a certain cushion of empty space around them and pilots know to be aware of any encroachments into that zone.

            That cushion isn’t available on the street. The electronics for the driverless car have to be more discerning. If the equipment goes into alert mode each time that there is a moving object or body heat nearby, it is never going to work.

          • 0 avatar
            Landcrusher

            Well, that’s the generally accepted common sense view, but it has a few issues.

            Big sky, little plane is fine until it isn’t. The math on closing rates is really ugly. Most planes don’t have any air to air radar at all. Instead of a center line, we mostly depend on altitude rules to separate. Incorrect settings can cause one plane and its crew to think they’re on the wrong side of the center line so to speak.

            There is a lot more stuff to avoid on a street, but it’s all moving pretty darn slow. The sensors on the cars being tested are likely years past anything on the newest airliners.

            Big Auto seems to think they will crack it, and now even MCS is optimistic.

            We can come up with hypotheticals all week, but they are essentially all the same.

          • 0 avatar
            Kruser

            Perhaps, but generally cars have a limited repertoire when there is a problem (i.e., brake & steer). So, the event may take seconds, but then it is ususally over.
            Certainly, there are gotcha situations, but even with those, once a computer has mastered them, it can be incorporated into the system. So, over the long term, the computer dominates.
            It was only in 2004 that no vehicle could finish the DARPA Grand Challenge – with time and effort these problems are being solved.

      • 0 avatar
        Landcrusher

        One big problem with aviation though is that the plaintiffs have created an art for deceiving ignorant jurors. Most jurors aren’t so easily fooled in analogous car issues. Furthermore, plane crashes tend to destroy evidence and the NTSB gets first dibs on the evidence and then their findings are inadmissible.

        We could construct a bunch of analogies, but I don’t think car manufacturers are going to see the same level of issues that aviation companies do.

    • 0 avatar
      Glenn Mercer

      I think liability will actually be a huge issue. Look at Toyota, who has paid out over $1 billion in legal fees, fines, recall costs, all because of “sudden unintended acceleration” that was proven to have been mostly the fault of misplaced floor mats. And yes, the autonomous vehicle will theoretically record everything so that blame cannot be misplaced — but note that Toyota paid a lot of the costs before it was finally exonerated by investigators. It is hard to over-estimate how hard lawyers will work to find a lawsuit if the pockets are deep enough.

      Put it this way, imagine the headline the first time an autonomous vehicle kills someone (and it will happen, given we drive some 3 trillion miles a year in the USA): “FrankenCar Kills Family of Four!” The lawsuits will erupt. Especially, as one automotive supplier has said, that we give leeway to human drivers. That is, if there is a horrible accident caused by a human driver who makes the wrong decision under immense time pressure (when that semi rolls over in front of him), the worst charge is manslaughter. If the autonomous vehicle makes the wrong call, well the programmer who vetted the software a few months earlier, safe and sound in his cubicle, is liable to a murder charge. (That is not my opinion, I am just reporting what I have heard at autonomous vehicle conferences.)

      Litigation and liability will be huge issues, IMHO, although Silicon Valley likes to wave them away (see AirBnB for example). But if a smartphone app misfires, all that happens is we lose our Angry Birds high score. If the car glitches (and remember, us dumb humans only kill someone once in every 100 MILLION miles), someone dies.

      • 0 avatar
        dal20402

        Liability is not as big a deal as everyone makes it out to be.

        Look at it this way: there is a fixed (and huge!) amount of loss that occurs as a result of driving today. Vehicle owners collectively cover that amount of loss through insurance, because owners are held liable when the factfinder determines they are partly or completely at fault (or when they settle in anticipation of that). Some or all of the liability may transfer to manufacturers as a result of automation, resulting in lower insurance premiums for owners but higher car purchase costs. But the total amount of loss is going to go way, way down, because the automated cars will make fewer errors. Owners will still be able to afford to cover the loss (although now in their car payment rather than their insurance premium), and they will.

        • 0 avatar
          mcs

          What about traffic offenses? There are plenty of scenarios that might result in a traffic offense. What if for whatever reason a car misses a speed limit change by a few hundred feet? Let’s say there’s a sudden sensor overload – too many things happening at once in a scenario not anticipated by the designers and the system delays dropping it’s speed and triggers a speed cam. Who gets the ticket? What if you can’t prove the overload and can’t blame the manufacturer? They’ll claim it wasn’t their fault and direct you to their legal team. What if it was caused by a solar flare or other weather incident? Do we make laws that automatically place the blame on the manufacturer? That’s not going to happen.

          Other scenarios – like what if it spooks a horse by driving too close and too fast for that particular horse? The horse throws the rider, the rider is killed, and the whole thing gets captured by a GoPro. So your car gets charged for manslaughter and driving to endanger and leaving the scene of an accident? Okay, so blame the manufacturer, but the too big to fail manufacturer went though bankruptcy and can’t be charged. Who do the prosecutors go after – the twelve year old being driven to soccer practice?

          BTW, I’m an experienced designer in this field. Aviation collision avoidance in the past, autonomous robots now and later versions will be (hopefully) be able to operate a modified car. I hope to be using IBM’s Synaptic chips and NVIDIA’s X1. I’m taking the safe route and focused on developing a patent portfolio rather than deploying a product – at least for now.

      • 0 avatar
        Pch101

        Liability won’t be much of an issue. Governments will pass laws shielding OEMs from excessive liability, car owners, lessees and renters will sign waivers, and the insurance industry will sort out the rest by setting premiums that account for the risk.

        But I would presume that insurers will want a driver behind the wheel who can take responsibility for a crash. They may require that a driver be behind the wheel in order for their policies to remain binding.

        • 0 avatar
          WheelMcCoy

          A slightly annoying video, the interviewer does ask BMW who is responsible in the event of crash in a driverless car:

          http://www.cnn.com/videos/cnnmoney/2015/01/07/cnn-bmw-self-driving-car-cant-crash.cnnmoney

          CNN: If it crashes, who is responsible? Me, my insurance? You? It’s your fault if it crashes, not mine. I’m not driving.

          BMW: Since there is no driver involved in the driving task, he can’t be blamed for any failures. That means… we will… uh… take care for all this consequences.

          It’s early in the game and the BMW rep seemed more like a tech so I doubt he can speak for corporate legal. I think Pch101’s scenario is the most likely: insurers will require a driver behind the wheel.

        • 0 avatar
          Glenn Mercer

          I agree completely with your statement: “In urban traffic, there are all kinds of things that could theoretically pose potential risks, yet don’t, such as nearby pedestrians who could bolt into your path but probably won’t. Too many false positives make the driverless technology less useful.” As an illustration, I was at a Stanford conference on autonomous vehicles, and a presenter was going on about all the situations they were programming into a car, and a lawyer (sic) stood up and said “Halloween.” The room went quiet, then he explained: “Can your car detect a small child in a white-sheet ghost costume, and tell if it is a piece of trash blowing across the road, or a child?” Fascinating point. Of course people leapt on him for coming up with an example that is so out-of-range as to be ludicrous, but his reply was “We have 310 million people driving 250 million cars 3 trillion miles a year, and who here wants to predict what happens after a newspaper prints the headline ‘Frankencar Kills Trick-or-Treating Toddler!” Another speaker at the same conference cited the pedestrian false positive problem as currently well out of reach (of course, given enough time….). if you’re really into the false positive issue, google research by Dollár, Wojek, Bernt, and Perona, really great (wonky) stuff.

    • 0 avatar
      toxicroach

      Odd that Boeing and Airbus are still around then.

  • avatar
    Landcrusher

    How do the naysayers explain all the big companies claims for when they will have products available? Surely things slip, but it seems Big Auto thinks this will be done by 2025 with really serious semi auto by 2020.

    People are making plans now, and if I were a big car company, I wouldn’t want my projections to be too far off unless my competitors were already ahead of me. Then you lie all you want to stall purchases of their products.

  • avatar
    mx5ta

    A young woman I knew once rolled her car when she swerved to avoid… get ready… a chipmunk. Obviously, this was a poor decision, and you assume the self-driving car’s algorithm would be set to keep going. Okay so far, but what about a squirrel? A cat? A small dog? A big dog? A deer? A pedestrian? Would there be a User Preferences screen, where an animal lover could choose to crush chipmunks, but to brake sharply for anything larger? The overall point is that these preferences, whether hard-coded or set by the user, had better be spot on.

    • 0 avatar
      Master Baiter

      Good points. In other words, driving requires INTELLIGENCE. While computers have gotten much faster and sensors have gotten cheaper and smaller, computers are not intelligent and they won’t be any time soon.

      • 0 avatar
        dal20402

        Driving really doesn’t require that much intelligence. It requires situational awareness and good judgment. If you want proof, just look at all the professional truckers and bus drivers out there with excellent safety records who really aren’t that bright. We can program computers to have much better situational awareness than humans and very consistent judgment. The hardest part is figuring out all of the situations we need to teach the computer. That will take lots and lots of testing time, but it’s not insurmountable at all.

    • 0 avatar
      toxicroach

      Brake. Not oversteer.

      With the reaction time of a computer, an animal or person will have to be thrusting themselves in front of the vehicle to get hit.

      The vast majority of accidents are due to people not paying attention or breaking the rules. The benefits of 100% of drivers obeying the rules and always paying attention will make the occasional potential FUBAR about as relevant as that one in 10000000 time when wearing your seatbelt gets you killed.

  • avatar
    jimbob457

    Well said.

  • avatar
    ttacgreg

    Google, an advertising and surveillance if not outright spy operation, has tons of bucks and needs to look sexy to Wall Street, so they throw all manner of crap at the walls (see Google Glass) and see what sticks. My understanding is they have very little to show for all the effort, and their core desktop advertising (they are taking a relative dump on mobile advertising) and data gathering does not have the growth potential it once did, if there is any growth at all anymore.
    Android? Sorry, it does not generate significant profits for Google. I have read accounts that Google loses money on Android, and I personally wonder what sorts of surveillance/data gathering they build onto the operating system that only they know about. They are a shady company (who am I kidding, what monster company these days is not?). They operate on and old Burning Man (one of their big retreats every year) mode, “it is better to apologize than to ask permission”. Reference the Google maps cam cars gathering Wi-FI networks data scandal a few years ago.

    Sorry officer I was not speeding, my car was.

    • 0 avatar
      dal20402

      We are so cynical these days. What you label “throwing all manner of crap at the walls” used to be called “basic research,” and it was the biggest reason why the U.S. became the premier technological power in the world during the ’50s and ’60s, a role which it still plays (although others are catching up). Most basic research is going to prove useless. But the companies that do it, including Google, are national heroes, and should be given a lot of slack for that part of their budget. Even if Google is producing lots of ideas that don’t go anywhere I really hope analysts don’t manage to succeed in shutting their R&D machine down.

      And I don’t think they will. Google sometimes needs direction, but it’s full of extremely capable and original engineers, and it knows that it won’t have any future at all if it’s reduced to a simple ad seller.

      • 0 avatar
        Nessuno

        It was all those German Scientists in fake moustaches actually… WWII was a massive transfer of technology.

        • 0 avatar
          dal20402

          Those guys wouldn’t have been much use without the massive, largely unaccountable R&D budgets of Bell, IBM, and the like. For the sake of the country, we need more budgets like that in the private sector today. We have the talent.

  • avatar
    JaySeis

    Your making the wrong statement. It should be “AI will take my job in your lifetime”. A web surfing robot will (soon) find and assemble words indistinguishable from human writing by anybody but a cryyptologic linguist (likely using another robot). The speed at which AI is developing is the limiting factor that needs the complimentary demonstration of cost savings through traffic/accident reduction. Once autonomus vehicles demonstrate effectiveness in industrial parks, factories/farms (already using/testing robots), gated communities, etc. Game Over.

    Then there’s this http://in.mobile.reuters.com/article/idINKBN0KH2CS20150108?irpc=932

    What more proof is needed?

  • avatar
    JaySeis

    Your making the wrong statement. It should be “AI will take my job in your lifetime”. A web surfing robot will (soon) find and assemble words indistinguishable from human writing by anybody but a cryptologic linguist (likely using another robot). The speed at which AI is developing is the limiting factor that needs the complimentary demonstration of cost savings through traffic/accident reduction. Once autonomus vehicles demonstrate effectiveness in industrial parks, factories/farms (already using/testing robots), gated communities, etc. Game Over.

    Then there’s this development which indicates somebody at Google is playing on the future chessboard of…auto insurance bhwahahahaha.

    http://in.mobile.reuters.com/article/idINKBN0KH2CS20150108?irpc=932

  • avatar
    Charliej

    In the US, it may take a long time, but in countries not ruled by lawyers, it will be sooner. It may surprise some from the US, but in other countries, there is not even an occupation of personal injury lawyer. The US is such a litigious country that introducing anything new is a risk. Honda was asked about their anti lock brakes on motorcycles sold outside the US. Why they weren’t introduced in the US. The answer, if they were brought in immediately, personal injury lawyers would claim that the previous years bikes were unsafe because they did not have anti lock brakes. It will be the same with self driving cars. Other countries will get them and the US will not.

  • avatar
    Nessuno

    Our infrastructure was built with CHEAP energy, the cost alone to redo the infrastructure is prohibitive enough.

    Although there is something to be noted about Baudrillards observation in System of Objects that “the car achieves an extraordinary compromise, for it makes it possible to be simultaneously at home and further and further away from home”

  • avatar
    CapVandal

    Stuff happens incrementally or it doesn’t happen at all.

    1. Self driving tractors are effectively here. It is much easier than a self driving car, but the tolerances are impressive. They can cover a field with close to zero missed areas or overlap. And these are huge machines.

    2. There are already convoy like electric linkage between trucks in Europe. The trucks have a lane, and they driven electronically very close together and fast.

    I see this as related to the impossibility of high speed rail. The only possible alternative is dedicated, high speed lanes with cars electronically controlled. We will never have high speed passenger rail in the US. Why? Because. Or if more is needed, look at the history of the 2nd Avenue Subway. It might be finished 100 years after breaking ground.

    But, a 100 mph lane would make 300 mile trips as fast as a commercial jet, counting the drive to airport, rental car, need to get to airport early, security, etc. Let the system evolve more, and driving will be viable for even longer distances.

    One caveat — commercial aviation in the US is incredibly safe. Most years have no serious crashes.

Read all comments

Back to TopLeave a Reply

You must be logged in to post a comment.

Recent Comments

  • Lou_BC: “Build thousands of homes in a forest that you know is going to burn.” My understanding of...
  • Lou_BC: @Carlson Fan – thanks for the rebuttal. I’m obviously not a sled guy. I looked at a bunch of...
  • Lie2me: I would be interested in an AWD hybrid, hopefully that’s coming next
  • ponchoman49: I wouldn’t say the Hemi’s are stone reliable at this point. Several repair shops including...
  • eng_alvarado90: HWY MPGs are a bit low for a Hybrid but keeping in mind the boxy shape and weight this is not bad....

New Car Research

Get a Free Dealer Quote

Who We Are

  • Adam Tonge
  • Bozi Tatarevic
  • Corey Lewis
  • Mark Baruth
  • Ronnie Schreiber