Uber’s Survival Strategy? “Tens of Thousands” of Self-Driving Volvo-Ubers Starting in 2019

Ready or not, here they come. Tough luck for drivers.

Uber is desperate. Its business model may not allow it to ever make money as long as it has to pay human drivers, spend massive resources to recruit them, and deal with the fallout when they cause problems. So it has been furiously working on self-driving technologies. And now it’s taking a small-scale experimental program to the big league – not decades down the road but starting in 2019.

Volvo Cars, which is owned by China’s Zhejiang Geely Holding Group Co., announced today that it has signed a “framework agreement” to sell Uber “tens of thousands of autonomous driving compatible base vehicles between 2019 and 2021.” The announcement added:

Our aim is to be the supplier of choice for AD [autonomous driving] ride-sharing service providers globally. Today’s agreement with Uber is a primary example of that strategic direction.

Volvo will supply the XC90, a luxury SUV, which seats up to seven passengers:

The base vehicles incorporate all necessary safety, redundancy and core autonomous driving technologies that are required for Uber to add its own self-driving technology.

The XC90 has a starting MSRP in the US of just over $45,000. Some versions are already available today with self-driving features, such as collision avoidance systems for low-speed accidents and systems that keep the vehicle in its lane and maintain the proper distance to the vehicle in front.

Financial details were not disclosed. Uber is going to get a big discount for a fleet purchase of this size, but it will also have to pay extra for whatever autonomous driving technologies that Volvo might preinstall. So this is not going to be cheap. If Uber pays $40,000 per vehicle, a fleet purchase of 24,000 vehicles – the number now being kicked around – would amount to nearly $1 billion.

Volvo also said that it will continue to develop its own technologies to pursue “its own independent autonomous car strategy, which is planned to culminate in the release of its first fully autonomous car in 2021.”

Uber and Volvo have been working together on self-driving vehicles for a while. In August 2016, Uber started testing specially-equipped XC90 models in Pittsburgh, though they still had an Uber employee in the front seat. It expanded its testing to Tempe, Arizona, and California.

But tests were briefly halted after an accident in Tempe in March this year. A vehicle driven by a human driver failed to yield, according to Police, and slammed into the test vehicle. These incidents are trotted out to show how self-driving technologies don’t work, when instead they should be trotted out to show how human drivers don’t work.

Humans are terrible drivers. Human drivers in the US caused 40,000 traffic fatalities in 2016. Self-driving technology doesn’t have to be perfect. It just has to be a lot less terrible than human drivers. That’s the promise.

The agreement with Volvo “puts us on a path towards mass produced self-driving vehicles at scale,” said Jeff Miller, Uber’s head of auto alliances, cited in Volvo’s press release.

The first batch of vehicles will be built in Volvo’s assembly plant in Sweden. But Volvo is also gearing up to build the vehicles at its planned factory near Charleston, South Carolina. Uber will then add its own systems and software to the vehicles.

This announcement of a mass-production deal for autonomous vehicles – the first of its kind – still leaves a lot of vacant spaces to be filled, so to speak, but it shows confidence among automakers that autonomous driving can be achieved in a matter of years not decades, and it shows desperation at Uber that it must be achieved pronto for its business model to have a chance of being sustainable.

About a quarter of the $1.2 trillion in auto loans outstanding were granted to borrowers with a subprime credit score. Read…  Auto-Loan Subprime Blows Up Lehman-Moment-Like

Enjoy reading WOLF STREET and want to support it? You can donate. I appreciate it immensely. Click on the beer and iced-tea mug to find out how:

Would you like to be notified via email when WOLF STREET publishes a new article? Sign up here.



  159 comments for “Uber’s Survival Strategy? “Tens of Thousands” of Self-Driving Volvo-Ubers Starting in 2019

  1. Derek says:

    Humans are terrible drivers. Human drivers in the US caused 40,000 traffic fatalities in 2016. Self-driving technology doesn’t have to be perfect. It just has to be a lot less terrible than human drivers. That’s the promise.

    This apologia is also a recurring player. Does this claim really check out? I have a hard time believing computers, as they exist today and for the conceivable future, will ever drive well. There is too much randomness for them to deal with.

    That cars are so necessary being a fundamental, and probably unrecoverable, design flaw in postwar society is the never-raised Big Problem.

    • Old Engineer says:

      Wolf is correct. Traffic deaths in the U.S have been between 35000 and 50000 per years since the 1930’s. What has declined is traffic deaths per mile driven. That number has declined from about 150 deaths per billion miles driven in the late 1930s and has become asymptotic to 10 deaths per billion miles driven since about 1996.

      I think the hardest thing for self driving cars will be co-existing with human drivers. Once all cars are self driven it becomes algorithmically much easier.

      Another problem is that in many cities it is only through selective violation of the traffic rules/laws that traffic flows at an acceptable rate. If everyone is “driving to rules” the actual road system capacity is much less.

      • Shawntay says:

        what will police do if they can not set speed traps and hand out moving violations? No human divers mean traffic violations which means finding another source of revenue.

      • Enrique says:

        Quote: “Another problem is that in many cities it is only through selective violation of the traffic rules/laws that traffic flows at an acceptable rate. If everyone is “driving to rules” the actual road system capacity is much less.”

        This is an extremely interesting point and had never occurred to me in the slightest, TBH. I think that is a real and obviously unintended consequence of mass-adoption of autonomous vehicles.

        Road capacity in many/most major US metros is an enormous problem that will under no circumstances go away within the lifetime of anyone reading this. The entire country is utterly non-optimised for public transit and it is more or less impossible to build infrastructure on any scale due to the almost Ottomon-esque level of political dysfunction.

        • Joe says:

          The promise that self driving vehicle is better or safer with no hard test data or the actual finished product is based on hope, faith, and lollipops.

          I thought this was the “be critical of the hype blog…”

          The vehicular accident totals rarely even split solo, etc. accidents.

          Lots folks die because mechanical failure. Will corporate penny pinching and a computer with even more parts solve the natural defect rate?

          But its probably a mute point. Whose got credit to buy one, and outside of urban areas can tuber compete with a 2k 97 civic?

          Cars are actually crazy cheap.

        • Smingles says:

          @Joe

          “The promise that self driving vehicle is better or safer with no hard test data or the actual finished product is based on hope, faith, and lollipops.”

          Uber has driven over a million miles in the last year alone, with passengers, in Pittsburgh, Tempe, and San Francisco.

          One accident… not the fault of the AV… and nobody was hurt.

          “I thought this was the “be critical of the hype blog…””

          Is it the blog where we don’t know anything about the topic at hand but throw out nonsense anyways?

      • RangerOne says:

        I don’t agree that people breaking the rules makes traffic move faster at all. In fact people behaving erractically, cutting in line or merging both to aggressively and passively is what slows traffic down. As long as people are in the mix this is a problem.

        But if a highway is 100% autonomous these problems go away because everyone can drive the same. They could in fact bump the speed limit because once you are 100% certain you don’t have to compensate for poor driving going fast won’t be as unsafe.

        100% compliance with traffic flow laws, signaling and merging would likely mean an end to stop and go traffic and we would have the fastest possible orderly flow. Computers could optimize every highway flow based on density and synchronize car driving to meet traffic needs. Highways will become like train tracks.

        Humans on average only make driving slower and more dangerous.

        That being said I find the idea that a Chinese company will provide autonomous vehicales by 2019 that are anything near full autonomous that can drive alongside real people laughable. The best experimental technology made public has not filled this role, I’m think MIT races and Google’s cars. China isn’t exactly known for innovation and their is no one to steel a working version from yet….

        • J Bank says:

          Your comment comes off a little brash, but I had a very interesting conversation recently that supports it.

          I have a client that made a ton of money off of microchip design and is in talks now with major American manufacturers regarding a new design. He mentioned that once you have an American company, ANY American company, sign off on it, you can sell to China had a dizzying pace.

          But you must, MUST have that first Amercian contract. Basically, if Intel or Qualcomm or even one of the smaller companies sign off on your work, that’s all the due diligence these Chinese companies need.

          China is much less about innovation and more about jumping on the bandwagon as soon as possible.

        • Kraig says:

          But if a highway is 100% automated these problems go away.
          This is why automated light rail is a thing.

      • Kraig says:

        Then the road network and poor public transit uptake is the problem

    • nick kelly says:

      In today’s Globe and Mail in biz section there is a piece by a Mike Isaac from San Francisco about Uber’s intention to buy 24,000 Volvos with self driving tech ‘when it is ready’

      There is no mention of early 2019 or any year, just ‘when ready’.

      If early 2019,or late 2019. is proposed for mass production pretty much all the bugs will have to be gone now.

      • Prairies says:

        Volvo America website states “At Volvo Cars we believe that our first unsupervised autonomous vehicles will be in the market by 2021”

        Not listed on the Canadian website so likely only to be released in California, The Unicorn State.

        • Wolf Richter says:

          Prairies,

          READ my article, for crying out loud, before you write this nonsense. That’s exactly what my article says. It’s a quote from Volvo’s press release. You just didn’t read the article. You’re just commenting without having read the article and without knowing the facts.

          Here is the thing, which you would know if you had read the article: the Uber vehicles will use Uber’s AV technology, and the cars Uber is buying from Volvo are ready to have this technology installed.

          Volvo is working on its own AV technology and will have its own AV technology ready by 2021, it said.

    • Smingles says:

      “This apologia is also a recurring player. Does this claim really check out? I have a hard time believing computers, as they exist today and for the conceivable future, will ever drive well. There is too much randomness for them to deal with.”

      Waymo (Google) alone has driven over 1 million miles autonomously in the last year.

      Some of this has been as part of a taxi service with actual human passengers (and a person in the driver’s seat to take over if necessary). Presently, the “safety” person now sits in the back seat and can’t take over, but can force the car to pull over and stop. In a few months, they will have the option to do away with a “safety” person altogether.

      These are small scare experiments in well defined areas (e.g. specific Phoenix suburbs, for Waymo). But they’re moving on to harder ‘playgrounds’ (e.g. Detroit) where poor weather and poor infrastructure will pose more problems.

      To put things in perspective, since 2014 there have been 34 accidents in California involving autonomous vehicles. About 1/3 were fender benders where a human hit a stopped AV. Only 1/34 were the fault of an AV in autonomous mode (3 more were the fault of the AV, but in manual mode).

      Based on the evidence thus far, it’s pretty clear that human drivers will continue to be the biggest threat on the roads today.

    • Kraig says:

      This is why mandatory ai assistance works well. Humans are bad at following the rules, they are good at dealing with randomness. Computers are good at following the rules they are bad at randomness. If speed limit is 30 then reduce speed of cars (this is even easier since Sat nav already know the speed limits and we already have speed limiters just link those straight to the Sat nav and take the human out of the loop. Of course then manufactures have to be convinced to sell them and drivers have to buy them. Telling the ai to stop at yield signs is tricked, you probably have to update them with special markers for the car. Computers are also even better at informing other computers. If one has an accident cars further back can slow in anticipation.

      • d says:

        “Of course then manufactures have to be convinced to sell them and drivers have to buy them.”

        In developed countries, Heavy trucks have had this for over 20 years. As standard equipment. Along with reverse alarms, reversing cameras, Etc.

        It is interesting to note, that a lot of this equipment, that is standard fitment in some regions, is still “Optional extras” in others. Its not about convincing manufactures to fit them, or sell them, its about convincing Manufacturers/Dealers, to stop using them, to MILK the customer, in certain markets.

        The customer that wont pay for “it” as an optional extra. Also wont pay to have “it” removed which is what happens in a lot of developed markets now.

        Dont want, Sat nav, Reversing camera, CD shuttle, Heated Electric adjustable mirrors, Headlight wipers, Factory alarm and Immobilizer, ETc.

        Fine, Vehicle will cost X much more, and you will have to wait X much time after payment, whilst it is customized to your requirements. Or Built to your instruction at the factory.

        There is a quirk in the law, as in many countries. Like ours, Speed limiters and reversing cameras, are a mandatory fitment. But, nobody mandates, they actually work, or can not be turned off, by the Driver/Operator.

        Those that are inbuilt in the EMS, can also be bypassed easily. As Most heavy Vehicle EMS are command-able through extra sensor ports, and easily service personnel tun-able, via any almost any laptop, due to multiple environment, load, and task, operating requirements.

    • Kraig says:

      Absolutely. All of the speeding related deaths could be stopped by mandating all cars be fitted with automated speed governers (already used by court order for some speeders) l. If all cars have a speed control ai, then nobody will speed, there will be no speed related deaths. Same for breathlyser interlock eliminate all drunk driving and watch deaths go down. (And lots of demand for rides from pubs/clubs) stop all red light running and watch deaths go down (little.tricker as computer has to recgonise red lights)The biggest problem for computers is the human drivers on the road, slower moving traffic is not quite as problematic. Automated technology on its own right of way has already been solved. See docklands light rail, vancouver skh train. Kerb guided buses offer the same fixed guideway on road surfaces. Although these both require investment in infrastructure. We already have active self driving minibus albeit slow
      https://www.google.co.uk/amp/s/amp.theguardian.com/technology/2016/aug/18/self-driving-buses-helsinki
      Remember too that computers can react in microseconds which can give much shorter stopping spaces in an emergency (child runs in front)

      • d says:

        ” Same for breathlyser interlock eliminate all drunk driving and watch deaths go down. (And lots of demand for rides from pubs/clubs) stop all red light running and watch deaths go down (little.tricker as computer has to recgonise red lights)”

        You dont get it guy who wants to speed drive after drinking run red lights, Turn off the Finance company immobilizer.

        WILL, and somebody WILL be there to provide the tech service for a small fee.

        Just like the private guy who turns of the service light when it comes on even though the vehicle has not been serviced.

        SInce ther hav ebeen vehicle sther hav ebeen no factory people customizing them .

        All those after market camshafts, fitted to GM street vehicles in California. With NOT FOR HIGHWAY USE IN CALIFORNIA written on the thrown away boxes, are the simplest example of this.

  2. dinn says:

    3 cheers for the robots: ha ha ha

  3. Mike R. says:

    Self driving cars will never happen. Period. It’s silicon valley hype and pipe dream. The complexities are too great to automate.

    Regarding human drivers, fatalities could be reduced greatly if we didn’t have so many illegal immigrants in this country. Many police forces are reticent/reluctant to aggressively enforce poor driving habits because of fear of picking up illegals. I know this for a fact from Charlotte, NC a pseudo sanctuary city. Driving behavior there was abhorant but no crackdown.

    And how about interstate driving? Very little to no enforcement of speed limit. Even 5 over. And the worst cause of accidents on the highway….. strings of cars following each other with one or two car lengths between them and all going 20–25 over speed limit. No reaction time whatsoever. Never seen cars pulled for this. No consistent and widespread enforcement.

    • alex in san jose AKA digital Detroit says:

      Reporting again from the actual, outdoors, so-called “Silicon Valley” here. One store I go to a lot is OSH, AKA Orchard Supply Hardware. Only the original store will do, the one on San Carlos street here in San Jose. It started as a co-op in the 1930s to enable fruit growers to buy ladders, baskets, etc. It’s now developed into a middle to high-end hardware store with things like patio furniture, nifty fireplace things, Craftsman tools, etc.

      I was there to buy some tire sealant and was helped very much by a guy who turned out to be the manager. And how he helped is how 90% of the helping goes on; I tell him I’m looking for “Slime” tire sealant, he knows what “Slime” is and where it is, and than has me follow him to where the item is.

      Exactly what, a few years ago, OSH supposedly launched a fleet of “OSHbots” who’d trundle around the store “taking” customers to where things are, since frankly it’s a pretty simple task. I’ve looked for the OSHbots, even just one OSHbot. I asked about ’em a couple years ago and I was told “they’re always in the shop”.

      The helpful guy seemed like he knew his stuff (he turned out to be a manager) so I asked him about the OSHbots. He told me that “Since Lowe’s bought OSH…” (a thing I highly doubt) “They’re over at Lowe’s now”. When I asked him which Lowe’s he said the one on Brokaw, and I said I go there almost daily and have never seen an OSHbot there. Well, another Lowe’s maybe ….

      Such optimism! I frequent a site called Reddit and if there were OSHbots active anywhere, there’d be videos and stories about them on there.

      All these things had to do was “know” where things are and have the customer follow them. In a controlled environment, no cars just people and carts and rows of merchandise. But they apparently have never emerged from the shop.

      So when I hear about self-driving cars, I think this must be the “flying car” of times. Meanwhile it’s the bicycle industry that’s doing well right now.

    • Anders Grönskog says:

      “I think there is a world market for maybe five computers.”
      Thomas Watson, president of IBM, 1943

    • Rates says:

      I am definitely in the camp that believes this whole AI thing is overrated and overblown, but saying that it can’t be done is not true.

      You can always dumb down the system such that the dumbest car can go through it.

      That’s how AI will ultimately work in practice. Just dumb down the population and declare the whole thing to be a huge success.

      • r cohn says:

        The simpler the task the easier it is for AI and robots to dominate.
        So the simpler the driving task ,like interstate trick driving ,the more likely that driverless vehicles will gain a foothold.
        But what happen when someone is driving thru the Donner pass.Or how will a driveless car react to bicyclists pedestrians and large potholes in NYC

    • ger says:

      Best advice is put an ambulance chasing lawyer on autodial …

    • David g LA says:

      Exactly. Have you compared our fatality rates with European counties? It’s pathetic.

    • Smingles says:

      You were wrong the last time you said self driving cars won’t happen, and you’re wrong now.

      They’re already happening. I don’t understand how you can be in such denial. They’re literally already out there, driving people around, today, in multiple cities.

    • BrianC says:

      Oh – I think you are overly optimistic here…

      You have to “think outside the box” and mix software and the law to get the required output…

      So if the SW engineers can’t get the technology to work… We just need to change two things. The inputs and the law!

      1) All metro areas required to embed wired control infra in all road ways.
      2) Pedestrians required to only cross in marked areas on the roadway. Outside the marks? Too bad for you if you get hit by an AV car…
      3) Bike riders and other non AV users required to carry a smartphone or equivalent device with SW/HW designed to interoperate with AV on public road ways. Preferably requiring a subscription fee payable for AV dev and maint…
      4) Blanket immunity from tort claims for AV manufacturers and municipalities that provide/mandate the above infra…
      5) Profit!

      We just need a way to privatize all of the profits and socialize all of the potential losses to make this really work.

      Cheers!

      Speaking from ~33 plus years in embedded real time T&M, OS Kernel, Aviation, and Medical fields. Doing SW design, implementation, maintenance…

  4. Guido says:

    “Humans are terrible drivers. Human drivers in the US caused 40,000 traffic fatalities in 2016.”

    This argument is fraught with problems. Humans drive under all kinds of conditions. They are required to do something called ‘going with the traffic’, which a lot of times means driving over the posted speed limit. Sometimes, you get a ticket for not driving fast enough even though the traffic is going over the speed limit. On 101 in Bay Area, posted speed limit is 65 on the peninsula and 55 near the city. Nobody drives under 70. Once I was doing a 73 and a CHP officer behind me gave me a glare asking me to vacate the lane (this was not an emergency and he had no lights on). He and the rest kept going at 75 after I vacated.

    In Michigan, for example, some roads post a minimum speed limit. From what I understand, the self driving cars drive slowly or stay with in the speed limit. The rules of the game are different for these two segments.

    Then there is the statistical argument. Elon Musk used to boast about his Tesla being safer than other cars because they had no accidents. He stopped making this argument after a passenger got beheaded. The fallacy in that argument is his assumption that events of interest occur at a uniform spacing, i.e. if one witnesses 7 accidents for 1000 miles driven *on average*, his argument claimed that 2000 miles were driven and no accidents occurred. The nature of the distribution of these accidents is probabilistic (because we don’t know the process generating the data, we can only guess by looking for patterns). Once these cars really hit the road and play in the major leagues, we will see some real results.

    In cases, where a self driving car does get into an accident, it is blamed on the human driver. At least in this sense, the self driving cars are like human drivers — it is always the other guy’s fault.

    • ger says:

      Uber will likely pad the pockets of politicians to make it a federal crime to collide with a driverless car. They will claim technology makes it impossible for a driverless car to make a miscalculation or fail to adjust to conditions….. such as America’s disastrous infrastructure failure.

    • Smingles says:

      “They are required to do something called ‘going with the traffic’, which a lot of times means driving over the posted speed limit. Sometimes, you get a ticket for not driving fast enough even though the traffic is going over the speed limit. On 101 in Bay Area, posted speed limit is 65 on the peninsula and 55 near the city. Nobody drives under 70. Once I was doing a 73 and a CHP officer behind me gave me a glare asking me to vacate the lane (this was not an emergency and he had no lights on). He and the rest kept going at 75 after I vacated.”

      That is the explicit law in many states, though. If you’re driving in the left lane, you must be traveling faster than the “normal speed of traffic.” That includes even going over the speed limit. If traffic is going 75 in a 70, and you’re going 70 in the left lane, you are breaking the law.

      People going below the speed of traffic in the left lane actually cause MORE accidents than people who speed in the left lane. You are actively being a bad driver by slowing down the left lane. It forces people to pass you on the right, and screws up the whole flow of highway traffic.

      The best system, in my opinion, is one where the speed limits more accurately reflect the speed most people drive at (which might be 10 or even 15 mph over the posted limit), but are also rigorously enforced. As are laws about passing lanes and travel lanes. It leaves less up to the discretion of the police, and makes things more clear for commuters.

      “In cases, where a self driving car does get into an accident, it is blamed on the human driver. At least in this sense, the self driving cars are like human drivers — it is always the other guy’s fault.”

      I’m not sure what you mean. In the vast majority of the very few accidents AVs have gotten in, they were fender benders where a human driver rear-ended the AV.

    • Kraig says:

      Driving with traffic also .means braking smoothly and keeping the correct distancr. Study after study has show that driving closer to the other car does not improve traffic at all ( in fact distance for smoother stopping improves things. Stopping or sharp accel/decal is what causes traffic. This is why roundabouts exist. As for minimum maximum speed that is pretty easy for a computer if lower than min, speed up and if greater than and slow down.

  5. Kent says:

    Wolf,

    You are very optimistic. Google has far and away the best self-driving technology and even they admit to human drivers having to take control of vehicles every 5000 miles. In a real self-driving taxi that might do 5000 miles/month that is a possible accident every month.

    Uber’s technology, if it has any, is going to be orders of magnitude less capable that GOOGs. No way this works. I will put my money on this being a desperate attempt by Uber to game up some chatter so that some early investors can bail before reality sets in.

      • c smith says:

        There will be millions of accidents involving self-driving cars, once they become prevalent. And, as with the wreck mentioned in the article, 99 out of 100 will be found to be caused by human drivers.

        • Wolf Richter says:

          There are millions of accidents now, caused by human drivers.

        • Old Engineer says:

          No, the accident rate should go down once self driving cars are prevalent. When most cares are driven following the same algorithms, which will lack human impulse, road rage, and other emotional responses, the accident rate should be minimal. It is the mix of self driven and human driven which will cause the problem. I do agree that teaching self driving cars to anticipate and deal with the irrational decision making processes of humans reliably will be next to impossible. No two humans can be reliably expected to react to the same set of circumstances in the same way.

    • Rates says:

      You might have nailed it. It’s interesting how this “news” is cropping up while Uber is negotiating with SoftBank.

    • Wolf Richter says:

      Google doesn’t build cars. It has to work with automakers. The problem for Google will be to find a major automaker that wants to partner with Google on this key technology where Google will extract all the gravy (as it does in advertising), leaving the automakers with the hard work of building and selling cars at a minimal profit. There is resistance. All major automakers are working feverishly on their own technology. So for Google, I think, it will be a battle royale to get its system into the cars of major automakers.

    • Smingles says:

      “You are very optimistic. Google has far and away the best self-driving technology and even they admit to human drivers having to take control of vehicles every 5000 miles.”

      Waymo (Google) currently already has self-driving cars on the road which can’t be taken control of by humans, except to stop. And in the next few months, they won’t even have that.

      “Uber’s technology, if it has any, is going to be orders of magnitude less capable that GOOGs.”

      Uber’s technology is Google’s… they stole it, according to Google (via a senior engineer who left Google for Uber). It doesn’t matter, anyways. They’re all going to end up with the same tech in the end. Advanced mapping, laser systems, cameras, etc.

      I don’t think people quite appreciate just how much time and money is being put into autonomous vehicles across multiple industries. There are a number of companies that deal with JUST mapping cities alone. This isn’t some hare brained idea from 15 years ago. This is happening, and everyone in the know realizes it and is competing to be the first big player there.

  6. Jeremy says:

    Tough luck for drivers. Tougher luck for pedestrians.

    • Wolf Richter says:

      I live on a dangerous intersection for human drivers. There are accidents all the time. A couple of weeks ago at night, three vehicles collided, and there were injuries, because one came down the hill and ran a red light, and the other two didn’t react fast enough. Humans are terrible drivers.

      • DV says:

        Wolf, the problem is speed…At what speed self-driving vehicles will be able to drive more or less safely? The same problem with EVs – the faster you go, the less range you have. The existing road infrastructure will be unable to handle lower speeds.

      • Steve M says:

        Here are the parts I don’t understand.

        Today, the child runs onto the street chasing the ball and gets hit by the car and is killed. The driver is led away in handcuffs.

        He won’t be driving again. Depending on circumstances, he likely might be doing time. He killed a child.

        Who gets cuffed when a self driving kills the kid? Who pays the criminal justice (not civil lawsuit) price for the crime?

        Second, crap happens and I’m running late for a most important meeting. Big bucks are on the line, and if I don’t make it, game over; no excuses and no second chance.

        It takes 20 minutes to get there usually and I got 15, but I get a driver who knows how to push it past the limit for an extra 100 bucks. I make it with two minutes to spare.

        Happens all the time with people. Can it happen with self driving cars?

        Just curious is all.

        • Wolf Richter says:

          Concerning your “20 minutes” question, and the five minutes you’re running late, the answer is simple: leave on time and don’t run late. Don’t risk your life and the lives of others just because you’re running five minutes late and want to make up for it. This is precisely what turns humans into “terrible drivers.”

        • rex says:

          “He won’t be driving again. Depending on circumstances, he likely might be doing time. He killed a child
          Who gets cuffed when a self driving kills the kid? Who pays the criminal justice (not civil lawsuit) price for the crime?”

          What crime has been committed in this instance assuming the driver was obeying traffic laws and was sober? Yes, a tragedy but no criminal act and thus no arrest for killing a child.
          Understand now?
          rx

        • RagnarD says:

          Great point. This seems to me to be the ultimate deal breaker.
          I doesn’t seem that any automaker/software developer/ insurer would be willing to shoulder this financial risk. And what person/driver would to give up that sort of control / responsibility?

          Regards Wolf and Rex, ever watch a too close to call play in a football game? They happen in real life on the roads. A kid darting out in front of car? Who saw it? Who is going to define whether it was legal / illegal? This is a HUGE gray area.

        • Smingles says:

          “Today, the child runs onto the street chasing the ball and gets hit by the car and is killed. The driver is led away in handcuffs.”

          Was the driver obeying all traffic laws? Did the child run onto the street in front of the car with no time for the driver, who was obeying all traffic laws, to stop?

          Then nobody is getting led away in handcuffs, and all you have is an unfortunate tragedy.

          “It takes 20 minutes to get there usually and I got 15, but I get a driver who knows how to push it past the limit for an extra 100 bucks. I make it with two minutes to spare.

          Happens all the time with people. Can it happen with self driving cars?”

          So you put your economic self-interest over the safety of everyone else involved, and what, you’re asking if you could force a self-driving car to do the same?

          Does that hypothetical really deserve an answer?

        • Steve M says:

          Let me respond this way to those who responded to me.

          Rex, first and most important is thank you for taking time to try to clarify the matter but I’m left more confused.

          When I took drivers ed long ago, I was taught that the number one law/rule of driving which superseded all others (speed limits, right hand turn on red, etc.) was Do not hit and/or kill anyone when driving.

          If a police car gets behind you and orders you to pull over NOW!, should you hit someone while doing so, you’ll be in bigger trouble, even though you were literally obeying the law.

          So who do the police investigate and question when such an incident occurs? The computer? Maybe VW can regain market share by designing the onboard monitoring system.

          And Mr. Richter, all I can say is my mother would love you, and I love my mother. I salute you. Your values are impeccable.

          I’m always grateful when you make time to reply to me. Thank you.

        • Steve M says:

          Smingles, I note that you chimed in and thank you is the order of the day.

          So I plead with you not to take any personal offense if I respond in the following way.

          Like Rex, from your extremely reasonable response to the child incident, I can deduce with near certainty that you are not the parent, sibling, aunt/uncle or neighbor of the child.

          Even if ultimately exonerated, I can’t imagine any driver who kills a child is not being led away in “cuffs.”

          I suppose they’ll tow the offending vehicle away, but I don’t see how that can be nearly as satisfying to the parent, sibling, etc.

          Law is rarely clear cut, which is why there is an entire industry and whole careers based on examining its daily application.

          It should be difficult for society to accept that a tragedy is the result of a “glitch” without human responsibility but I realize we’re being conditioned that way.

          I’m just a humble anachronism.

          As for the late scenario, I used the first person just to shorten the question. “I” is quicker than “that person.”

          I tend to agree that machines can’t do worse than us.

          Whether the hypothetical deserves an answer or not, my hunch is that it will be answered eventually. Who has the greater stake usually prevails.

          I was merely pondering possible scenarios. From my experience, it’s rarely a case of if technological innovation will be abused. Usually, it’s a question of how.

          Thanks again. I’m pleased by the quality of intelligence on this site.

          I’m sorry to detract from it.

      • Mary says:

        Wolf, I’m with you. Whatever it takes to get human beings out of the driver’s seat. Human beings who just snorted cocaine or consumed a six-pack. Human beings whose eyes are glued to their phones. Human beings who just had a fight with their spouse or got fired from their job and are now homicidal/suicidal. Human beings with seasonal insanity like the girl who tried to swerve past me, first to the right, then the left, on a tiny street in Pasadena.

      • Wolf Richter says:

        Just now, a minute ago, there was another accident at our intersection. Fender bender. A taxi and a guy in a ragtop. Two human drivers. Humans are terrible drivers.

  7. IAV8 says:

    I won’t say I told you so when the affected drivers use pickups and other larger vehicles to push these driverless vehicles into ditches when they are also passenger-less.

    At some point the “disruptors” of our economy need to be told to evolve less quickly. I can imagine a suddenly unemployed army of taxi drivers in our major urban areas becoming disenfranchised with the American ideal of pulling yourself up by your bootstraps.

    If you aren’t expecting violence, you aren’t paying attention.

    • alex in san jose AKA digital Detroit says:

      I can see this happening. There are a lot of angry people out there. And while any normal person would be appalled at the idea of hurting a taxi or uber or delivery driver, if the car’s just putting around on its own yeah it’d be pretty tempting to mess with it. Add in that if you can “capture” the car without damaging it too badly, that’s a fortune in parts you can sell on the black market.

      • JA says:

        Alex, I totally agree with you. I doubt the self driving software engineers are thinking about all the cleaver ways dishonest people will take advantage of expensive cars with no one driving. I’m sure cars will disappear. I’m sure there will be folks who figure out flaws in the software to stage accidents or do other things no one has yet thought of.

        • Kent says:

          A can of spray paint and knowledge of the location of the lidars.

        • Smingles says:

          “I doubt the self driving software engineers are thinking about all the cleaver ways dishonest people will take advantage of expensive cars with no one driving. I’m sure cars will disappear. I’m sure there will be folks who figure out flaws in the software to stage accidents or do other things no one has yet thought of.”

          Really… so you think random joes on the internet who spend 2 minutes a year reading about autonomous cars come up with the idea of taking advantage of these cars, but the companies who have spent billions trying to make this a reality, and the engineers who spend all their time trying to make this a reality…. aren’t thinking about it? It just slipped their minds…?

    • fajensen says:

      I can imagine some teen HaX0rZ running real-life GTA with the self driving cars while streaming the passangers funny faces over the anonymous pay-per-view live streaming service that will no doubt be available from the more douche’y amongst those “tech innovators and disrupters”.

      Maybe there will be a service for trading the controls of Pwned Cars too?

      Anything goes if it’s provided with an App?

    • Smingles says:

      “I won’t say I told you so when the affected drivers use pickups and other larger vehicles to push these driverless vehicles into ditches when they are also passenger-less. ”

      And I won’t say I told you so when these affected drivers get hauled off to prison for acting like total idiots and endangering everyone, including themselves, on the road. Deal!

      Like, what, you think they’re going to be able to drive them off the road and then just putz away with nobody being the wiser?

      Even without a passenger I bet that’s likely to be charged as felony aggravated assault AND leaving the scene of an accident. Or did you not think that part through…?

  8. TJ Martin says:

    Horsefeathers ! No State , County , City or Town will allow this to happen regardless of Uber’s grand delusions due in no small part to all the potential litigations involved but more importantly because the infra structure nor the technology exists in order to make autonomous cars even the least bit viable in the real world by 2025 .. never mind 2019

    Then add in the reality that the current Tax Reform has eliminated all funds earmarked to develop the infra structure needed for autonomous vehicles .. and ..

    … I’ll hazard a guess this ‘ so called ‘ announcement by the delusional minions and marketing mavens at Uber’s sole intent is to drive up their rapidly waning stock prices .. hoping the general investing public is as stupid as Uber thinks they are

    Begging the question … what is them folks at Uber smoking ?

    • Wolf Richter says:

      Laws at the state and federal level are already being passed to accommodate self-driving cars. Insurance companies are preparing for it. Everyone is working on it to make this happen.

      • Michael Fiorillo says:

        Captive governments, federal state and local, might be happy to give these companies everything they want, but the technology may crash against the shoals of bipartisan policies that refuse to raise the taxes necessary to maintain its infrastructure. It’s a question of political economy, and the contradictions inherent in neo-liberal policy, as much as technology, but no one wants to face that.

        It’ll be more than a little ironic if all this disruptive innovation is delayed, or worse, by a polity that insists on extracting wealth from public resources, not producing and improving them.

    • Smingles says:

      “No State , County , City or Town will allow this to happen”

      They’re already out there.

  9. Prairies says:

    This autonomous hype and the EV semi’s from Tesla popping up at the same time smells of stock manipulation. Trying to pump up Q4 and pad pockets before the 2018 wake up call.

    I know a person in B.C. who order the new Tesla car when it released in the summer. The expected delivery date is end of 2019-early 2020. Yet Volvo will release these Uber cars in 2019 along with Tesla semi’s showing up at Walmart and Loblaw’s warehouses.

    All propaganda, all in an effort to get tax subsidies and more inflated stock prices.

    Anyone interested in going halvesies with me on my future project. I want to build a car that runs on everything, but also runs on nothing. It will go everywhere but get you nowhere. Best of all, it will cost you nothing and make you money. A unicorn on 4 wheels, but better than wheels because they won’t go flat.

    • Wolf Richter says:

      Did the article mention Tesla? Nope. Not once. This wasn’t about Tesla. Tesla doesn’t matter.

      • Prairies says:

        I assumed it was an article on autonomous vehicles. Just pointed out a market competitor and if you look at the stocks for Volvo and Tesla you would see they both are on downward trends, Tesla fell close to $80/share since Sept and Volvo has fallen $10 since Nov. 1. Both stocks have since seen a pause to the decline. It may be short lived, or it may be the start to a 3 year rise.

        So that is why I will assume this to be hype for stock values to climb. Or is this not a place for observing the financial gains of the suppliers of pipe dreams.

  10. Truth Always says:

    1. Who is writing the software for Self driving cars?

    2. Isn’t there a pending Waymo (Alphabet) – Uber lawsuit over IP related to autonomous cars?

    3. What precludes Toyota/Mercedes etc. from launching self driving car service with (or without) help from Waymo since the former can do it way cheaper?

    I think the story on this still needs to be written even if you discount the fact that self driving software will be ready/ (I am in software and i think it maybe ready in 3 years if Nvidia/Intel can keep pushing their chips to process more information)

  11. uuu says:

    Some other concerns w/ this business model:
    1. Insurance rate of driverless vehicle vs regular vehicle?
    1a. due to higher vehicle value
    1b. due to novel/untested tech
    2. lack of driver = disrespect factor
    2a. by passengers – leaving garbage etc
    2b. by other vehicles on the road’
    2c. by vehicle maintenance staff in the event they’re pushed too hard
    – who will complain if they cut corners on service or cleaning?
    3. reaction by uber drivers with “conventional” vehicles – more
    unpleasant realities coming out of the woodwork?
    4. competition from future players (i.e., google) with second generation, lower cost / higher reliability vehicles

  12. walter map says:

    Robotic CEOs will be coming off the drawing boards sooner than you think. It turned out to be easy to program a robot to stack the board and hire a compensation consultant. Figuring out a mediocre golf game was a little tricky. Overpromising and making excuses would have been trivial except for the size of the database.

    Humans won’t be fully obsolete until they come out with robots to replace beggars in India. Creating a market for them is likely to be as big a challenge as creating one for FB.

  13. Shawn says:

    Level 5 autonomous vehicles, aka self driving taxis are still a long way off. All autonomous tech currently on the market is Level 2. The Tesla model S that killed a guy (watching Harry Potter when he should have been watching the road, not sure if that is actually true) last year was a Level 2 autonomous vehicle. Essentially, the truck the guy’s car ran under was white in color and the car’s AV tech couldn’t see it. Currently, there are only two vehicle that have Level 3 AV tech, the Audi A8 and the Cadillac Escalade. Both very high end vehicles.

  14. Michael Fiorillo says:

    I know that NYC is an outlier in many ways, but the idea of it working here seems preposterous.

    There are tens of thousands of vans and pick ups for local service industries, the cable companies, all vying for advantage on cramped streets, many of them swerving and double parking. The streets are being torn up by the utility crews, ambuklances and fire trucks demanding right-of-way, bicycle messengers and restaurant take-out maniacs whizzing by in all directions… and that’s in good weather. NYC may be an extreme case, but every area has its variables and x-factors that will impair these vehicles.

    I wouldn’t presume to say it’ll never happen, but I don’t see Level 5/fully autonomous vehicles in full, real time use for a while. And whoever it is that succeeds in doing it, it ain’t gonna be Uber.

  15. nick kelly says:

    ‘and it shows desperation at Uber that it must be achieved pronto for its business model to have a chance of being sustainable.’

    Indeed. But if a car is to enter mass production in 14 months (early 2019) the prototype has to exist almost now. Does it exist now?

    One thing is certain, Uber’s desperation won’t guarantee its practicality.

    No doubt self-drivers are ready for limited travel around defined urban routes. much like street cars without tracks.

    In fact the street car without tracks, or bus, is obviously the more economically feasible app because the expensive self- driving tech would have much the same cost, but be utilized by a much more expensive vehicle, capable of transporting many people around the aforementioned urban route.

    So why are we putting the car before the bus, at least in the media where announcements about self drivers are almost always about cars.

    One answer: there is more languishing common stock in the car biz that like Uber, desperately needs a boost.

  16. GSH says:

    I used to do real-time multi-tasking operating system software development. The number of people comprehending inherent race conditions in large multi-layered software stacks is very limited. We should all be scared of these largely untested blobs of million lines of code. I wish every one good luck with their outsourced software. A large CASE or IF-THEN-ELSE statement is not AI regardless how much data you process.

    • Derek says:

      This is the heart of the matter. Even with true AI (whatever that’s supposed to mean), nobody knows how it works or what it’s doing. This stuff is created in great haste. This will not end well.

  17. IAV8 says:

    Dream on. This battle will start with the slashing of tires.

    It will evolve into something much worse.

  18. Charles says:

    Self-driving cars in a contained, controlled environment without any human drivers could work. This would probably be for only affluent communities.

  19. roddy6667 says:

    Uber has a system where they use the vehicles the drivers pay for. The driver has the responsibility of owning and maintaining the vehicle while getting paid peanuts for operating it. They lose millions of dollars a month under this system. Now they are going to buy very expensive cars, taking on the added expense of ownership, maintenance, repairs and taxes? How will this be better? They could lease them to drivers, but then the drivers will make even less because of increased overhead.
    Self-driving cars are a lawsuit waiting to happen. I imagine that litigation attorneys are already drawing up the lawsuits. They will just wait until the accidents(s) and fill in the names and places later. Don’t forget how these lawsuits work. You don’t have to prove scientifically that something caused death or injury. You only have to convince twelve people too stupid to get out of jury duty.

    • Greg says:

      “Now they are going to buy very expensive cars, taking on the added expense of ownership, maintenance, repairs and taxes? ”

      This is a profound insight that for some reason Wolf has missed. Uber will also be responsible if lack of vehicle maintenance results in a car accident.

      This driverless car craze, like the bitcoin and dot-com craze and the current stock market mania are all part of the ‘magical thinking’ of the age.

      Whether it be Peter Thiel, Elon Musk or W Richter, these people live gilded lives with 4- star service and champagne dreams. Life one big party where anything they believe will become a reality.

      Yet, I’m still waiting for my pets.com stock to make a come-back.

      The driverless car revolution may very well foment real revolution after the first 100 real, not virtual children are killed by those things.

      • Bruce Adlam says:

        I drive public transport buses and we have the latest computers on board as long as you rely on a signal from the sky you will never be able to garentee all the time .the hype from wall street to Uber is unbelievable and the cost of the sensors are prohibitively high this is more about keeping the money rolling in on a dream way in the future I wouldn’t hold my breath .people are so gulable they believe anything.

    • d says:

      “You only have to convince twelve people too stupid to get out of jury duty.”

      Criminal Prosecutors have a 60% hit rate convincing those same juries, that something happened so somebody must be held responsible.

      Which is how so many peopel, who did noting wrong, end up in jail.

    • Mel says:

      In fact, Uber has already tried paying for its own cars, in the guise of cheap auto-purchase loans for its drivers, and it couldn’t afford that.

      For your closing point though, this is America. EVERYTHING is a lawsuit waiting to happen.

  20. Halsey Taylor says:

    It’ll work fine in the end. They’ll eventually run on rails – either steel or electronic.

    • alex in san jose AKA digital Detroit says:

      Halsey Taylor – I spend a fair amount of time riding my city’s light rail system, and if there’s one thing you’d think could be automated, it’d be a light rail system, right? Well, there are no end of instances where having a human being at the controls heads off an accident, a passenger missing their stop, etc. And, because $2 for a ride instead of the cost of owning a car is much more do-able for the poor and people who don’t function well in general, there are a fair number of times the police and Sheriffs have to meet a light rail train at a station and do … police and sheriff stuff.

      Wasn’t it a thing a while back that some fast food places with drive-throughs, the person behind the mic at the drive-through was states, even countries, away? Is that still a thing? Because I see a situation where even if the light rail system were automated, it’d not save time or money, and it’d end up clunkier than it is now, with a person at the controls.

      • fajensen says:

        If you’re going to Copenhagen, you can try the metro.

        It is fully robotic, no drivers.

        And it works well, there is a train every 5 minutes or so.

        The sucky part is buying the ticket, one can, but: They favor some demonic hell-spawned IT-horror called “rejsekortet” which tracks Everything and require a danish SSN (to better track Everything).

        The had problems with one or two morons forcing the doors causing shutdowns. Since they were actually morons in the medical sense, they are now living somewhere else.

  21. Raymond Rogers says:

    I’m still waiting for the flying cars that popular mechanicspromised in the 90’s.

    I wonder will these cars handle navigating stuff falling off the back of trucks and blown tires. I wonder how they deal with construction zones, especially when the lanes are narrow and sometimes not leveled. When an accident occurs and a lane is closed rather suddenly what happens then? How do these cars act with snow and Ice storms? How about snow, ice, and parking lots?

    My guess is the next economic downturn erases many pipe dreams. It’s just a guess, by it seems there is much figuring out to do and probably not enough never-ending funding to do so.

    • Wolf Richter says:

      In the 1990s, you could never imagine having a super computer in your pocket. Now it’s standard (smartphone).

    • Smingles says:

      “I’m still waiting for the flying cars that popular mechanics promised in the 90’s.”

      Dubai has already started testing flying taxis. They don’t look like cars, though– more like a combination of a small helicopter and a drone.

    • roddy6667 says:

      One thing I realized from driving 50 years in CT, is that you can’t see the lines or writing on the road if there is snow, or even just rain and headlight glare. Things like the lane markings, left turn only, and all that just disappear. If these self-crashing cars depend on them, they won’t be there when they need them.
      Also, being almost 70, Popular Mechanics has been promising flying cars since the Fifties. Still waiting.

  22. Crysangle says:

    Well you seem to be somewhere on board with the idea of driverless, but your reporting is just that, and the arguments balanced. Mish loves this topic, is a full proponent (or if not a qualified polemicist), and we had a lot of fun with that until he changed site ( improvements in a week or never, but he has lost almost all his long time commentators, probably a lot of traffic).

    I agree from a safety perspective, BUT you are not going to run a new system and qualify it as acceptable by ratio to the previous one. It does not work like that, not for someone who is injured or worse by AV, not by rational logic that demands an existing comparison . Commonality is subjective, as you will not present how things would have been under the previous reality as comparison – no can do. You are not going to tell someone injured by AV that it is OK as a whole, that they are injured for the common good. Nope.

    So let’s see.
    You want to put large robots on the roads, ok.
    They are going to have to prove they are as good (safe) at least as current drivers, fair no?
    How?
    Well you build a very realistic test circuit, with stretches of highway, introduce all kind of circumstance, with all kind of traffic, and evaluate till you reach a point of approval.

    Then you go to real world tests (with standby driver). Maybe this is the path covered so far… maybe not. The normal auto regulation for standard vehicles has been watered-down already apparently, to allow AV access.

    Then what? Well AV causes an accident, and you take away its license (as with any faulty vehicle)…sorry Uber/your transport is not working until a patch is installed and tested….and why not prosecute builders and those who passed its roadworthiness at the same time. That is right, all tens of thousands… millions… may be grounded, as they all have the same “driver”.

    Do you think it will fudge through, that because there is enough public pressure (from owners) to continue use, that it will be OTG upgrades? I don’t, because those injured will not be only users, and they will sue the hell out of any favour to AV that causes injury, and that means government – because there is no driver to blame, only those that put AV on the road…those that allowed it to be put on the road.

    It is very fraught, it might get steam rolled through legislatively, but I think the reality is much further off than is imagined by some. In fact as a project, it is a major project, as it requires vast changes to many facets of the use of (public) transit space.

    • Guido says:

      @Crysangle, the process you mention is what one would take if one were releasing any product with company’s life on the line. That is a very expensive process and no company can go through that unless it is subsidized by govt. The subsidies are now coming in the form of dmv giving permits, the lawmakers changing laws, etc. — behaviors driven by obsequiousness and an eagerness to show they too are tech savvy.

      This is exactly the behavior that led to the pension funds losing a ton of money in 2008. The investment banker would show up in the office of his classmate who would be working in Calpers, say. The IB fellow is quite smart and very savvy with what he does. The Calpers classmate is not in IB because he doesn’t understand the math. The IB fellow also practices graft. He then throws math all over the place and everybody nods so as to not betray their lack of understanding (emperor being naked and all). They buy and lose, while the seller makes boatloads of money.

      The same thing is happening with AV. Those in the know have been telling us that these cars will show up any day now. The guy doing it has the brain power to wade through the issues involved. The consumers and reporters of these developments? Not so much.

      I think this is where capitalism works best. If their tech really works, the AV folks will make enough to not have 10 generations work again. If they are wrong, they sink unless they are bailed out (bet on this).

      OTOH, I keep wondering about the dmv and the legislators who keep championing this by removing all safeguards. It will be fun to watch these roaches run for cover when the witch hunt begins to explain why somebody died.

      • Crysangle says:

        Yes, it is all being prepped.

        Personally I dislike having anybody else’s agenda forced on me, and especially from someone I don’t know at a desk I don’t where, and especially especially that they do this in the name of everyone else for “their own good”.

        If it stands and works by itself fine, but if it is going to anything else than fit in under existing legal discipline, then it has to at least go through some kind of direct public vote. Clearly the power structure of most western countries is not really representative, and so it will come down to avoiding and quelling public dissent in a real life fudge. If it does not take off, if it is widely rejected for whatever reason, it is likely to be on a scale that shakes out a system which holds these sort of ideas as an absolutely necessary next step of progress. To that end it is likely to become a harsh battleline at some point, for one reason or another, very likely an ideological one.

  23. johns says:

    Yes in the beginning there will be accidents but they will introduce AI to self driving cars and they will learn all tricks of defensive driving, professional driving, and at the end they will all be professional drivers.
    Not so sure about Uber, they might be using this as a leverage to get more debt but soon one car maker will achieve this perfect self driving vehicle, it may take another 5-10 years but we will see this.

    • Dan Romig says:

      I wonder if AI cars will be able to ‘learn all tricks of defensive driving, professional driving, …”

      I pay attention to all around me when in my car or on my bike and motorbike. Little things like a driver in front of you glancing over a shoulder are predictive indicators that that car may soon change lanes. Will the sensors on an XC90 see the glance, or will it just be quick enough to react to movement from the car in front?

      Technology is pretty amazing; time will tell how far it will go for driverless vehicles.

      • fajensen says:

        Will the sensors on an XC90 see the glance, or will it just be quick enough to react to movement from the car in front?
        The XC90 is designed to run over an elk on an icy Swedish forest road without slowing down too much to miss Systembolaget’s closing time!

        If they get the “Green”, T8 Twin Engine AWD version, they can get a 320 BHP Turbo-Charged petrol engine on top of the electric motor for a total of about 407 BHP, making everything more exciting.

        The Thinking behind that particular design, I believe is: Armoured Knight of Rightful Justice (management) vs. Drafted Peasants with Pointy Sticks (A Picket Line :).

        They can easily work out the kinks in their algorithms (and roo-bars) in those places “abroad” where the people we drone for having the wrong used mobile live. Recycle the signature strike AI as it were.

        • Crysangle says:

          Ah yes, the old elk test meets twenty first century hominid evolution.

          “Are yer fast enough elky? Well are ye? Are ye?? ”

          The US is “abroad” for me, and that is all I need to know… for now. I’m just waiting to see how they try to introduce this to mainland Europe… French taxi drivers, or the diametrical to a 2CV … the Spanish youth love tech, but they are broke too, and often mischievous.

          Maybe AV will end up becoming the symbol of the useless elite.

          Monsieur Sansmains , or Cojonudo de Gilipolla.

      • Johns says:

        You will be amazed what AI can achieve,
        An AI computer beat the world GO game champion, the computer had played this game with a human first time, it played by itself all the time and learned all possible moves, developed itself and became a pro Go player. Why not drive a car , which is a very easy task, everybody can drive a car, good or bad.
        But other things can happen, like battery failure, equipment, sensor failure etc. then probably parks the car and calls for mechanic :)

  24. Realist says:

    What a wet dream for hackers with bad intent. Those cars will be connected by a datalink to HQ … And security will probably be priorized as in the usual IoT thingys ….

  25. JR says:

    Uber is desperate. No doubt about it.

    T. Boone Picken: “If you want a deal real bad, you get a real bad deal.”

    The probabilities go something like this:
    1) Uber is still able to scrape together $1B to purchase Chinese Volvos
    2) The regulatory framework is ready in 2019 or sometime after to allow Level 5 (no human intervention) autonomous driving
    3) The technology is actually capable. No doubt that it will be capable some day, but it will likely have to travel up Moore’s law for a number of years before this is a reality. That or roads and cars have to start communicating (traffic lights talking to cars, etc). It is much more likely that another $10T will be blown up in the desert before this type of infrastructure investment happens.

    My $0.02 and worth every penny.

  26. ian says:

    Uber’s original business model was to bring supply – a car and driver willing to act as a taxi and demand – somebody needing a ride, together. They would make the sale possible and take a cut of that sale. This they do but do not make a profit. They do not own the cars, employ the drivers (yet), maintain or insure the cars, put gas in the cars, nothing. Theoretically very little overheads. Can someone please explain to me how they will make money when they have to buy, maintain, insure, gas up, garage a huge fleet of cars? Sure, without drivers but with all these other costs which they don’t now have? And if I was mayor of a town watching all those taxi driving jobs going up in smoke, unless I was hosting the business, I would perhaps think twice of giving them a license.

  27. Snotfroth says:

    I think a lot of people are in denial over autonomous vehicles. With the amount of money, talent, and interest being poured into AV, it is effectively inexorable. How much power does a modern CPU/GPU have? A lot. How many of these processors can you fit and power in a car? A lot. Hardware is not an issue, and there are many more sharp programmers piling every day onto this project.

    Wolf’s point about AV needing to only be better than human drivers has been on the back of my mind for a while. AVs don’t need to be perfect, they only need to be, on average, safer than human drivers. At that point insurance companies will probably get on board. They may also like the fact that AVs will be covered in cameras and sensors so the liable party of any accident will be clear. Eventually as safty improves and statistics are collected, insurance will start charging higher premiums for human-only drivers because they are high-risk under the new standard.

    And when there are accidents, especially in the beginning, you can be sure the AV manufacturers will carefully analyze the performance of their systems. Any improvements they make will be applied through software updates to the entire fleet. Thus, all AVs will learn and get better from each individual accident. That’s something human drivers can’t really match.

    Regarding the “so where are the flying cars” point, people already do enough damage with ground vehicles. Regulators would never allow those same people to pilot large machines flying overhead at 300 MPH. The only way we’re getting flying cars is if they are computer controlled — autonomous. So, AV may in fact finally facilitate the fabled flying car.

    • JR says:

      Hey no denial here. It is a matter of paradigm shifts and economics. Clearly the truckers will be the first to lose their jobs to autonomous interstate transport convoys. The domain of the problem is limited, and maybe there can be autonomous friendly regulations regarding deviations and such surprises. After the truckers are out of a job, then we can discuss the future of Uber, in whatever form it might exist by then.

      • d says:

        Those autonomous trucks, will be so easy to: hack, spoof and hijack, it will be amusing to watch.

        No more Human drivers, no more nasty assault and battery Etc, charges.

        • nick kelly says:

          Jesus D, I know we’ve differed but we must share a certain criminal gene (well controlled! ) that makes us think of sh&t like hijacking an AV.
          I had an email read on air from London ( Squawk Box Europe) about Amazon’s plan to deliver stuff by drone. I said they’d be too tempting to shoot down. Tongue in cheek, host Steve Sedgewick reads it and says: ‘Nick, we cannot condone criminal behavior’

          My latest: if you can hack into unoccupied AV Volvo, you can drive it into a semi or anywhere and then to the chop shop.

          A ‘known unknown’ about this whole thing about AVs, especially ones with goodies (semis) is the reaction of criminals to unmanned vehicles.

        • d says:

          Think on this

          Today they hack GPS trackers in stolen cars , and put them on long-haul trucks, which, owners, insurance, and the police then chase after. Whilst the car goes the other way and gets chopped, Rebirthed, used in crime, Whatever.

          AI trucks report on their journey. So when the truck arrives and turns out to be a stolen AV Car impersonating the truck.

          The truck has been missing for now many hours/days by then???

          40 Ft semi, loaded with Auto tires is worth over 200K quick cash. Full of flat screen TV’S Over 1M Quick cash.

          Driverless trucks, are a truck hijackers dream.

          You can buy Military quality GPS (and GSM) Buster/Blocker on line for less than $200.00. Now.

          For very little more, you can buy the neat little toy, the cops and spooks use to spoof as a Celltower and collect everything moving over it, at the tower it either isn’t or is only lightly encrypted.

          Driverless trucks will get destination changes and report using the same systems.

          We haven even talked here about the potential for criminals to make billions hijacking crew-less automated Ships.

          “Jesus D, I know we’ve differed but we must share a certain criminal gene (well controlled! ) that makes us think of sh&t like hijacking an AV.”

          Its only a crime if you get caught.

          Its not a criminal mind, its an engineers mind, where are the Vulnerabilities, what will be the first failures, and what will be the result of those failures. How do I counter this.

          What happens when X occurs.

          Automated Garage door/Gate openers are great until somebody buys a box full of replacement remotes, and opens ever garage in town.

          Movement activated light.

          Counter, BB Gun, or crossbow (quieter), or spray paint sensor if accessible unobserved. Or depending on circumstances, remove main power supply fuse at street, or cut power to building at supply point (Pole) switch (May trigger Alarm).

          Its Just, fault, cause, remedy. Applied differently.

    • Mel says:

      “you can be sure the AV manufacturers will carefully analyze the performance of their systems. ”

      I’m not so sure. People could have carefully analyzed Credit Default Swaps, but they didn’t (or refused to believe the few people who did) until the world blew up. We can see lots more systemic risk and TBTF when it comes down to closing half the trucking industry and its biggest players, as somebody pointed out up there. Profits are at stake. Dieselgate.

    • fajensen says:

      How much power does a modern CPU/GPU have? A lot. How many of these processors can you fit and power in a car? A lot. Hardware is not an issue, and there are many more sharp programmers piling every day onto this project.

      These things are still stupid as snot, being basically an abacus with an operating manual (software) for moving the little balls. To get elegant and efficient AI, similar to the adaptable intelligence that any small animal has, I believe that some form of quantum computing is needed.

      Effective AI cannot be brute-forced into existence while using the wrong models, there need to be an undestanding first on what the physical link is between “minds” / “brains” and “the universe”. It should be clear that “intelligence” works differently from “computing”; Simulating intelligence by adding more computing is going down the wrong path, I think.

      Maybe they can use the current approach to somehow get something to “work” – but – I bet it won’t be elegant or pretty or power efficient or resilient in any way.

      PS:
      There is some interesting work being done on “Reservoir Computing” which does hint at something “smarter” being possible. Researchers using physical systems to perform the non-linear mapping needed for this technique to work is a possible way to include quantum states into the model and perhaps get at something that is not fixed, mechanical.

      https://www.nature.com/articles/srep00287?WT.ec_id=SREP-639-20120301

      http://minds.jacobs-university.de/sites/default/files/uploads/teaching/MLSpring16/ReservoirComputing.pdf
      https://www.elen.ucl.ac.be/Proceedings/esann/esannpdf/es2007-8.pdf

      https://www.nature.com/articles/s41598-017-10257-6

  28. kevin says:

    This AV topic keeps recurring here ans elsewhere, so I’m putting mostly similar points as I had made earlier.

    1) The ipso facto argument that AV can never replace human drivers (for whatever technical and non-technical reasons) already shows that human thinking is understandably biased against it. Who wouldn’t naturally be against something that threatens our place in the world?
    Remember the lake Wobegon adage where everyone always thinks they always are above-average? I too think that my driving skills are far above average but my wife doesn’t seem to think so. lol.

    2) I agree that Uber pushing for 2019 to put AVs on the road is rather ambitious and reeks of desperation. The technology may be feasible, but the regulatory, legal, insurance and a host of other social issues that city mayors, and state jurisdictions face are the hurdles that will likely derail their schedule. Old world systems and institutions are not going to take these disruptions too well, and Kalanick himself, in his mad ego-trip, has totally messed up the relationship with regulatory forces, by acting like a hard-assed, lawless cowboy conquering virgin country (and countries).

    3) Despite pt 2), whether Uber / Lyft / Volvo / Google / Tesla succeeds or fails is NO longer the point at all.

    If Uber / Google / Tesla goes bankrupt, some other company somewhere will eventually buy over their patents/tech/expertise and push ahead with AV vehicles. It is not a question of if, but when.

    All tech once invented will be used/abuse/misused and improved upon. Whether its nuclear energy, gene-therapy, self-driving cars, legal weed smoking, virtual reality, sexbots or whatever you can imagine.

    Saying that it won’t work because humans will try to break them or find ways around them (via tire-slashing, hacking them or deliberately crashing into them etc.) is NOT going to stop their inexorable march towards implementation. Some humans will always want to “hack” things no matter what you say or do. If we were still using horses today as the primary mode of transport, do you think humans will not “hack” their horses for a faster ride?…Maybe by pumping them with steroids or whatever that is available in that age and time? lol.

    Society and the regulatory environment plus these AV companies will adapt and move forward, dragging large swathes of humanity along kicking and screaming and then everyone will eventually not only accept the tech but pretty much get used to it.
    Tell me if I’m wrong here…. like how you cannot live without your smart phone now ;-)

    4) The AV technology is here, and it will keep on improving in leaps and bounds…generally slower than sci-fi predictions but definitely faster than we expect. For instance, Smart phones swamped the entire global market within barely 5 years, resulting in Nokia turning from a bellweather to a has-been within the same time period.

    It maybe 5, 10, 20 years but usually technology goes “slowly at first, then suddenly its everywhere”. Exponential rates is difficult for human brains thinking in linear mode.

    5) Old Engineer made a good point about the difficulties during the transition period where there is a mix of human-driven and AI-driven vehicles, but all transitions are troublesome. Thats why they are alled “transitions”.
    Again, its just a question of time. My guess is, it will not be in 2019 but very likely in 2025 and almost certainly in 2040. In the long march of human progress, 25 years +/- is just a rounding error, although to any individual its a big deal because it is like more than a quarter of a person’s lifetime and getting replaced by AI during one’s viable working lifetime is not a comfortable thought….which goes back to my pt. 1).

    • Michael Fiorillo says:

      You may be right, but that “long march of human progress” idea is questionable; progress is spasmodic, not orderly, and can go in reverse. The Romans had piped water and flush toilets, which disappeared for a over a millennium in Europe after the empire’s collapse.

      With most human knowledge now in digital form, one could perhaps argue that it is more vulnerable to going poof, via a naturally occurring or military-induced EMP. In the unlikely, but possible, event of that, expect some very rapid social regression.

      I’m not predicting an imminent collapse (though the potentialities are there for all to see), but just pointing out the flaws in the idea that we’re in a linear move toward ever-upward “progress.”

      • nick kelly says:

        There is only one weapon that has never been used, the H-bomb.

        It uses a small atomic explosion as detonator and produces a fusion reaction like the one powering the sun. It creates a small brief star on the earth.

        Arms negotiations have lowered their numbers to about 4000 per side for the two main powers.

        Their use en masse would halt progress.

        So….do we feel lucky?

    • Ken says:

      Well written comment. Thank you Kevin.

  29. Shawn says:

    I often give Uber rides in Boston, and I bring this issue up quite a lot with riders. It’s interesting to hear everyone’s thoughts. The thing I can’t quite picture is how (soon) the AVs will be able to handle a snowstorm with over a foot of snow like the one I was giving rides in last year. As a side note: The surge pricing was over triple in that instance since so many drivers opted to go offline.

    Also, here’s a page which links to the first nine parts of a comprehensive look at Uber from a business perspective: http://horanaviation.com/Uber.html There’s also a newer part 10 at Naked Capitalism.

  30. Citizen AllenM says:

    One of the thoughts that is being missed is the economics of it all. If I can get a ride to and from work, and go get groceries on the weekend and stop at the hardware store for less than $100 a week I will switch in a heartbeat.

    My cost per vehicle runs about $5k per year including everything, so below that price point, a vehicle goes away…period. I have 3 of them for 3 drivers.

    Economics will dictate reality, along with reliability and convenience.

    The lowest hanging fruit is bus service and long distance trucking. Night local delivery is next, then tight urban environments. As for the nightmare that is NYC- banning most private cars would solve the problem- or simply start congestion pricing for permits.

    The real push will come from insurance costs- and I suspect that insurance costs for cars that do not enroll in the electronic monitoring system will increase significantly (human controlled, no interface), cars with interface will be cheaper, and cars partially controlled cheaper still, and autonomous will be the cheapest. Also, place like NYC will use this for automatic congestion pricing. LoL.

    People babbling about fatalities are being foolish- kids get run over every single day in America, and nobody gets led away in handcuffs, just another accident and funeral. The thought that there is punishment is simply when you have a real fault present, and in the current system, even with fault punishments are not equitable.

    Horses killed riders and bystanders, yet the car killed even more of them and we still got in and drove.

    Rural America will be the last stand of the independent driver-and just consider how traffic laws will become meaningless when most of the vehicles are moving as part of the collective.

    And as was pointed out above, many law enforcement jobs consisting of ticket writing and toll collection are going away forever.

    Crime is already becoming much more difficult in a wired society. A guy I know in a low rent neighborhood no longer locks his gate to his repair shop outer areas because he now has total camera coverage in place that is run through his phone and records offsite. Just drop off the pump and put it on the table. It felt strange, but he even texted me a pic showing the pump was received with the time. Clarity of the pic was outstanding. He also has infrared.

    In short, technology is overrunning so much of what we assume is normal, but cars are just the most potent symbol of what is happening in so many fields.

    The real question is what happens to employment in such a society?

    • Crysangle says:

      Outside of fuel my cost per week ( including s/h purchase cost over 3yrs) is at most 20 dollars ( all in, inc maintenance) for a small 50000 mile, yr 2000 +, hatchback in average/good condition.

      Fuel price is fuel price.

      Won’t beat that under any scheme except government subsidy…but then “you” still pay.

    • Kent says:

      “One of the thoughts that is being missed is the economics of it all.”

      Agree 100%. And here’s the deal. Just about everyone has to get to work around 8:00 and leaves work around 5:00. Which means just about everyone will still have to have a now driver less car available only to them. If it made economic sense to have someone else own a car and drop it off to you every morning, that service would already be happening. But it doesn’t make sense.

      So why does having a car without a driver suddenly make sense? You still have to have one just for you so you have to pay for the entirety of the vehicle, maintenance, repairs and fuel. No savings there. And you have to pay for all the extra electronics to make it “driverless”. And you get to give up control of your trip in terms of speed, directions etc…

      Would it make sense as a taxi service? Probably. Is that a big enough market to absorb the billions being pored into this? Probably not.

  31. Nicko2 says:

    The automated vehicle revolution is happening…but it’s happening in faster growing, more prosperous economies rather than the US.

    For example, China — investing billions in AI and associated technologies, Tesla is working on a deal to open a factory there.

    Japan – they will deploy a fleet of several thousand automated cars for the 2020 Olympics

    UAE – heavily investing in automation, partner with Tesla, automated taxis, and automated taxi Drones.

    If you want the future, look to the east.

  32. yoopar says:

    Many issues to play out here. Insurance, Maintenance, reliability,sustainability,repeatability, accountability,cost,and last but not least…. will people want to buy them. I believe all will eventually be worked out.First, by gradually integrating them as they are doing now and then totally AV in the bigger cities where they are needed most. The people will go where the cost is the least, which means eventually nobody ( except those with $) will have one in their driveway. The American dream will not be dead…. but will creat much more jobs for a different kind of vehicle. Maybe theme parks where one can go to actually drive, motorcycles,atv’s,or side by sides. It will be illegal to drive one on government roads. Only your land. It will save many lives and use cars for what they were intended for…… to go from point a to point b. Probably new and better ways of transportation will be invented creating many jobs. Who knows, we may get back into walking. We can see how the internet has been so secure, safe, trustworthy and totally private. I can not see how AV would be any different. After all …. The horse always comes before the carriage.

  33. prepalaw says:

    Wolf,

    please think the consequences of decriminalizing auto accidents cause by robot drivers. The same no consequences will apply when a robot breaks down your door and shoots your family. When a robot breaks into your bank account and steals your money.

    Corporate activities decriminalized. Human activities remain subject to criminal laws.

    You have elevated the synthetic person over the natural person.

    • Wolf Richter says:

      Motor vehicle accidents are not “criminal” unless a law is broken (running a red light, DUI, etc.). AVs are programmed to observe the laws (and they don’t drink). They will still get into accidents occasionally, but not because they broke the law. So the legal issues involving AVs will be “civil” (who pays for what) rather than “criminal.”

      • Crysangle says:

        Disagree.

        You are purposefully placing an object that is not controlled by a person in the roadway. If it causes damage, that is criminal.

        An object taken on the roadway, that is controlled by a person who is equally subject to possibility of injury, then that person is considered in principle “a mutual subject” with regard to mishap. Error is therefore civily attributable, unless purposeful intent is demonstrated, when it becomes criminal.

        I say this as common sense supposition.

        • Michael Fiorillo says:

          “I say this as common sense supposition.”

          Aye, and there’s the rub.

          Common sense can be easily ignored when there’s elite consensus among the political, financial and technocratic classes, as seems to be developing with automated vehicles.

          Get ready, ’cause this is gonna hurt.

      • Hiho says:

        Wolf, you do not seem to get that there are actual people writing the algoritms, so damn, these cars will still be driven by humans. The only difference being that instead of a wheel they will use a piece of code.

        You seem to uncritically worship a technogy that you do not understand, no offence intended.

        I used to work as an IT analyst and I can assure you that provided you saw how developers tend to work, you would not sleep well at night knowing that these monstrousities could be soon in the streets rolling over your wife or kids.

        Computers are not gods, coders are humans and fail, bugs happen, shit happens.

        Two words: precautionary principle.

        • kevin says:

          @Hiho, I really don’t know why I seem to be “defending” the concept of AV when I do not own shares of these companies…..but I feel there is a lot of misguided logic and irrationality that I feel compelled to point out.

          I work in IT for many years too and I am familiar with the whole 9 yards of software engineering, waterfall methods, Agile,, testing and QA etc. Yes, there will be bugs and there will be errors… and they will be improved and corrected over time, so what was your point about?

          Look, commercial airplanes today have millions of lines of codes too. Airliners run on AUTO-PILOT at cruising heights.
          Does that mean there aren’t going to be s/w bugs in airline codes? Or do we avoid flying altogether until we perfect everything? Modern airplanes are certainly not without imperfections and defective designs, but on balance, they are pretty reliable versus the cost of reaching technological perfection (if its even possible).

          In fact, the USAF have already visibly demonstrated that aircraft can take-off and land autonomously too, but obviously aviation authorities today are not so daring to suggest we should do away with commercial pilots……..yet.
          Pilot salaries have been under pressure of late though….

          http://mashable.com/2016/09/13/f16-autopilot-saves-unconscious-pilot/#sn9QYZOCVmqL

          On the whole, airplane manufacturers does a pretty decent job for safety, balanced against the potential human and economic cost of implementation. Remember, it wasn’t always the case. Historically, there were many accidents and errant airplane manufacturers/contractors, but over the decades, these got weeded out by competition and by regulation.

          Yet, today lots of bad shit still happens with airliner accidents too, right?
          The fallacy is assuming anything “autonomous” is all bad simply because we do not have a man-in-the-loop. I understand it because we fear losing control and/or we always want a human to place the blame on.

          My take is society as a whole, will still move ahead, perhaps with some sort of blended approach where a human driver might still be required as emotional reassurance for large-passenger formats (such as autonomous buses) but not necessary for smaller vehicles such as sedans etc.

          Other folks saying it won’t happen or categorically saying its all bad tech development is just not being realistic at all.

          The regulatory environment, legal, insurance and other technological aspects WILL improve, adapt and move forwards no matter how vehemently we object today. Whether you like it or not, we are all “going along for the ride” (pun intended).

          But Yes, I agree you might personally avoid riding in the initial AV prototypes or the newer models :-)
          Perhaps, wait for the next iterations with better updates (hopefully without newer bugs introduced) or more stable technologies…. but again its just a question of time to generally-acceptable platform stability.

        • d says:

          Since there have been electronic Auto pilot systems in Commercial aircraft there has also been the issue that terrorist may take control of the plane from the ground and or the cabin and redirect it for whatever purpose.

          Including 9/11 style events.

          The talk of Autonomous Ships is band enough. The talk of Autonomous passenger carrying planes is in the current global environment insanity.

          when you can loose a loaded commercial flight (MH 17) almost without trace. That was piloted.

          How can Autonomous passenger carrying commercial flights, be even considered.

        • Crysangle says:

          Various pilots have commented on the lack of similarity between AV and aircraft autopilot. Basically they say that autopilot is just an aid, that they do a lot of work flying the aircraft besides, and that it is not relied upon plus often has to be taken to manual control… that is besides takeoff and landing plus all the parameters they have to set and adjust to keep the flight in order and autopilot functioning.

  34. breamrod says:

    you know I guess we’ll all be chipped and have a basic income and will eventually evolve into mechanical beings. Won’t that be nice? NOT. Glad I won’t live long enough to see it.

  35. Rates says:

    LOL. You can’t make this stuff up. http://www.zerohedge.com/news/2017-11-21/uber-admits-it-hid-57-million-users-hacked-paid-keep-quiet

    They can’t even keep their servers secure, and now they are going to do driverless cars.

    Just take a look at the latest Fast and Furious movie to see what will happen. A hacker taking control of multiple cars in a city center. The lives saved due to eliminating bad drivers will be offset by hackers taking control of the command center.

    The future is coming and there’s no place in it for you.

    • walter map says:

      “The future is coming and there’s no place in it for you.”

      Thanks for the confirmation.

  36. Ion storm says:

    This self driving car hype is a reflection of concentrated media (and thus concentrated advertizing dollars more than anything else.)

    We’re in a dead zone as a society. Once we hoped and cheered for a man to the moon now the main stream media tell us we should hope to

    1) be spied on all the time by our cars. Everywhere you go will be reported to the police, the president, your congressman and every 3rd party partner of the company that made your car’s bumper.

    2) lose our jobs if we drive anything for a living

    3) work on our way to work as your drive will become company time.

    4) expect to be killed by a driverless car one day as many have cited above, no one will be held accountable if there is an accident, hence quality standards on these vehicles will be low.

    5) expect to be brought in for questioning by our cars everytime the police get a bug up their ass about anything.

    DRIVERLESS CARS- YEAAAH!!

    • kevin says:

      @Ion storm, I read your rant (which I shouldn’t have because its like many of the unbalanced comments here & elsewhere on this heated topic).

      Firstly, a disclaimer: I do not own shares of any AV company or aspiring-AV company, such as Uber / Tesla / Waymo / China BYD or whatever.

      1) You are already being spied on, if you’re carrying a smart phone in your pocket. You’d better go smash your smart phone now and swear off it forever or try getting used to it.

      https://qz.com/1131515/google-collects-android-users-locations-even-when-location-services-are-disabled/

      2) Well, there IS some merit to this point. Fact is, we will lose a LOT more than just merely driving jobs.
      Me thinks this AI cycle of innovation is fundamentally different from all the previous technological leaps (e.g. discovery of fire by man, invention of gunpowder, the gutenberg printing press, powered flight, internet etc.)…. AI applications has far-reaching implications because it is mimicking / even surpassing in some respects, human brain performance.

      3) Minor point. Aren’t you already technically “working” if you read or reply to work emails on your smartphone today?
      My parent’s generation never even had to bother with phone calls from their boss after office hours, since a pager didn’t even existed then. lol.
      Again, do you realize my point that you have already been dragged forwards by technology kicking and screaming?

      4) Again, IF we were all still using horses for transport today, do you have a good chance of being killed by your horse?
      Your horse is “autonomous” and with a non-human mind of its own too.
      There is no human remotely controlling its brain, although you can direct it by pulling the reins left or right, but it doesn’t guarantee that your horse will not suddenly throw you off or kick you to death right?

      Hmm… I wonder how modern-day insurance will handle the horse?
      Or who should we sue in the event of a horse accident? The rider of the horse, the horse rental company, or the horse trainer for not properly “programming” the horse?
      See my point now?

      5) See my pt 1 above.

  37. Hiho says:

    Not going to happen, never. This looks like a, desesperate attempt to keep the hype and prevent the bubble from collapsing before insiders get out.

    Keep in mind that shifting capital risk and cost onto the driver, who financially captive, is the key feature of uber business model along with looking for monopoly power.

  38. Cambric Finish says:

    In 2016, it is estimated that Americans drove a record 3.2 trillion miles (3,200,000,000,000). Waymo said recently it logged 1 million miles on public roadways in 7 months (1,000,000) . Now, as a thought experiment replace the 3.2 trillion miles of human driving with 3.2 trillion miles of the current vintage of self-driving technology (or the technology 3-5 years from now or 6-10 years if you can see that far). This is the goal, correct?

    • Nicko2 says:

      China is will have two dozen mega cities (population greater than 10 million) in a few short years….with population densities like that, automation is the only solution to combat congestion. As I said before, the automation revolution is happening, just not in the US.

      • Hiho says:

        And could you tell me how on the earth will AV improve congestion? Will these cars fly? The way I see it, for each human-driven car that you substract, there will be a new AV.

        The solution is far simpler: public transportation. But ofc that’s not so cool as AV and does not allow hype-pump and dump-take the money and run operations in the global casino.

    • Wolf Richter says:

      I think your projected timeframe is way too tight, simply by the way vehicles age: they last decades. Some vehicles are kept on the road for 30 years or longer.

      So when these AVs are starting to be sold to the public (which will be years after taxi and delivery companies have been using them), it will start the clock ticking. From that point on, it will probably be another 20 years before AVs have replaced a large number of human-driven vehicles, simply because no one is going to throw away a car that is still running well. Over time, people will sell their human-driven cars and buy an AV, but someone is going to buy that cheap used vehicle and drive it for a long time.

      The US sells between 15 and 17 million new vehicles are year. At that rate, it takes a long time to replace all existing vehicles on the road today.

      • Robert says:

        “Some vehicles are kept on the road for 30 years or longer.”

        And driveless cars will require endless upgrades and costs to stay compatible with the ‘newest upgrade’.

        Everytime a government official creates a new law you’ll be forced to upgrade. Sounds very very expensive. But good for the car industry.

        • Wolf Richter says:

          For many people (not all), vehicle ownership would no longer be necessary. A vehicle that gets driven 20 hours a day as part of a big fleet that takes people where they want to go won’t last 30 years. Or even ten years. By the time it has 200,000 miles on it, it’s time for it to head to the salvage yard. If the vehicle utilization is high, it will only take a few years to get there.

      • d says:

        {“The US sells between 15 and 17 million new vehicles are year. At that rate, it takes a long time to replace all existing vehicles on the road today.”

        Untill they pass a little law like the one in Singapore.

        Which sates you can not operate a commercial vehicle older than 10 years on the public or private road, in Singapore.

        I do not believe you would be foolish enough to Bet AGAINST them, doing such.

        Or registration and inspection laws in Japan which make it cost prohibitive to keep vehicles over 10 Years old on the road, even if they are in perfect condition.

  39. prepalaw says:

    Wolf,

    you know Japan very well. How many years can you keep a used car on the road in Japan before being required by bureaucrats to subject the vehicle to costly inspections and repairs. You are coerced to sell the car and buy new.

    This method will be employed in the US to force people to stop using conventional cars. Either, retrofit your car with automation technology, or we will not renew your registration.

  40. Edger Unner says:

    Uber desperate? You must be joking? Ubee will lose money on this per vehicle. Uber drivers currently have to maintain their own vehicles and suck up the depreciation value. At the end of a year they will all find they’re making a minimum wage. Uber has always been known as an a**hole Corporation. They view their own employees as a problem. And yes, an Uber driver is a direct employee of the company, not a freelance driver. I own a limo company, so I know the reality of what is really happening when these cases go to court.

  41. nick kelly says:

    I’m not going to replay my skeptical tune re: self drivers but I am going to make a point about human driving.

    Many people LOVE it! Others just like it.
    Does anyone commenting watch TV ads? They aren’t selling steak they’re selling sizzle. The luv affair is far, far, from over.

    If safety was our only concern motorcycles would no longer be legal. You are 70 times more likely to die on a motorcycle than in a car. Most US MC drivers don’t own them for economy, they like driving them.

    Neither the sports car or the MC is going to disappear because a bunch of folks want to live in a futuristic hive.

    On a personal note: I would consider an electric for one reason: as an economical second car. l don’t want an electric to drive or even an automatic. To me an electric sports car is almost a contradiction. I want to feel some vibration and hear some exhaust rumble.

    And to those who say the personal vehicle per se is obsolescent, take a look at a map of the US. Not everyone lives or wants to live in a hive and, honest, at least a hundred million don’t. There is not going to be any bus, Uber,Lyft or self-driver operating in the forseeable future in most of the rural or semi-rural US no matter what happens in the hives.

  42. My3cts says:

    Aside from any technical/financial consideration about the advancement of self driving technology in the coming years, I would also like to pinpoint a fact as someone who has worked in urban electric car sharing schemes: Many people are pigs, and do not respect the common property. It might be because I’m in Paris and this is an accute problem here, but no doubt it’s a common problem the world over, to varying extent. Enough for a customer or two to start leaving an empty bottle or bag of chips in the car, and at the end of the day you end up with a vehicule no one in his right mind would like to set foot in. Without any sort of human supervision, god knows what’s gonna happen in those cars.

    • Pete Franklin says:

      There are ways around this. Sensors , and cameras that detect “left behinds” and automatically ding your credit card and provide you with a picture/video in case you feel like contesting.

  43. Pete Franklin says:

    Wolf, good article. One of my programs is investing/developing AV-related tech (for use in other robotic industries, not auto). Volvo is using, like most others, the Nvidia CUDA platform (Drive PX). Once enough real-time hours across millions of miles of varied conditions are gathered, I firmly believe AV will be feasible. The estimate is about 3-4 years down the road (lol), maybe 5. The tech itself isn’t the issue, relatively cheap hardware already linking existing sensor systems (and LIDAR if need be). It’s the experience needed. I think the Germans will be first out of the gate. I also think the trucking industry will be the first to adopt, and then commuter-related routes. I also think the precautionary principle will govern the ethical dilemma scenarios (which in reality almost are never deployed).

    There is, however, a much cheaper and more effective way to get auto-deaths substantially reduced. That is, install alcohol locks on every truck and vehicle. Alcohol-related deaths account for about 1/3rd of all vehicle deaths in the U.S.

    • nick kelly says:

      Glad to see someone who apparently knows something about it upping the ante to 5 years. Ya a lot can happen in 5: like between Sept 39 and May 45.
      Microwave radar, jets, missiles, A bomb.

      However, once someone opines that a program is so difficult it might take 5 years, the corollary is they don’t know how long it will take.

      That’s why they are giving themselves a buffer that in terms of tech development that is highly motivated and financed (although not quite as much as 39-45.) is half way to infinite.

      • Wolf Richter says:

        You said: “However, once someone opines that a program is so difficult it might take 5 years, the corollary is they don’t know how long it will take.”

        Well, even if it takes five years: It takes about 3 years to develop a basic new car (nothing fancy), from the day it’s decided to the day the first retail customer buys a mass-produced vehicle. That’s something every major automaker routinely does. No biggie. It used to take four years. Five years is not long when it comes to developing, building, and testing complex manufactured products for the mass market.

        By contrast, check out how long it takes to develop a passenger jet.

        • nick kelly says:

          I was a bit needlessly caustic in that comment, which is not required to be skeptical.

          One bit of newish info described by ‘d’ but also by the US Army.
          The latter has recently made some changes based on observations of the combat in Ukraine.
          The Russians deployed a cheap, portable low- power transmitter that jams GPS for 30 or so miles.

          As a result the US Army is re-introducing some troops to low tech navigation, apparently including star charts.

          When each vehicle has its own driver, the system is decentralized. There will be accidents, but not all at once.
          If self- drivers depend on GPS, their navigation is centralized. They have, in effect, one driver.

  44. Coaster Noster says:

    Wolf, whenever you wish to create bigger numbers for viewership, run something about autonomous vehicles…anything.

    The same kind of “run over kids”, “hacked software”, “out of control speeding” “who’s liable” -commentary has erupted all years, every year, since the concept of driverless cars became “a thing”. It’s all been put to bed…for years…get over it…don’t display your ignorance, folks.

    It’s tough to get rid of “village mentality”. All your commentatorii are sitting in their armchairs, trying to come up with “gotcha” arguments.

    Almost to a person, the critics of the move to autonomous vehicles fail to understand the increasing velocity of information, and the amount of data and information. I believe I read that total amount of information, that was produced since the time of the Egyptians up until 2007, was doubled by the year 2014. I’m not saying you should understand and “feel” what it means, the huge increase in data and information, but you should not think that a single person’s perception is unique enough to warrant writing down an “insight” as to what will happen with AVs.

    “What???!!! I stick my check in the ATM, and it =reads= the handwritten amount??!! Wow, I can’t wait until someone “hacks” that, and it goes away as a failed idea…or, mebbee not..!”

    • d says:

      ““What???!!! I stick my check in the ATM, and it =reads= the handwritten amount??!! Wow, I can’t wait until someone “hacks” that, and it goes away as a failed idea…or, mebbee not..!””

      They have been hacked, and are still being hacked, with stripe readers and key loggers.

      Billions in total, has been stolen that way.

      • Wolf Richter says:

        I’m depositing all handwritten checks that way. The scanners work. The amount shows up in my account instantly (though it’s “on hold” for a few days).

        ATMs now have safeguards that make card scanners nearly impossible to install. You also need to visually inspect the ATM to make sure no add-on equipment, such as a keyboard overlay, has been installed. If something looks fishy, go to a different ATM. Basic precautions help you avoid a lot of problems.

        • d says:

          True and yes.

          They have also found at least 1 completely fake machine being used to harvest data.

          Just like various instore EFTPOS machines that also harvest Data.

        • Wolf Richter says:

          Sure. I see dubious machines all the time, owned by private parties, with no bank brand on it. Only morons use those machines. I only use machines owned by major banks and unless I’m traveling, I only use the machines from my bank.

    • Wolf Richter says:

      Actually, people love to comment on the driver-less-car topic, and they’re having fun doing it, and I think much of what you see has some humor in it, but in terms of readership, this article was one of the weakest ones in recent times, with only 4,800 “reads” so far. Some of this may have to do with Thanksgiving week, which is always lousy for financial sites :-]

      In my experience, the number of comments is not correlated to the number of “reads.” Some of my most-read articles (10x as meany reads) have fewer comments.

      • Coaster Noster says:

        Thanks! The “real story” noted, as opposed to my speculation. Noted as well: comments do not correlate to “reads”. While I understand your point of view, I personally do not feel the “humor” in the comments, but rather a strident “lashing out” against the obvious erosion of one of the last bastions of personal power projection: the automobile, owned and driven by an individual, in a manner that suits their taste.

        Too many people on any given landscape these days.

        I’m thinking about it. Perhaps your most-read articles with the fewest comments are your most-detailed, understandable, and complete articles, so the reader goes, “Hunh, yeah….” and moves on. No comment required, except, “Good job Wolf!!”

    • nick kelly says:

      ‘commentary has erupted all years, every year, since the concept of driverless cars became “a thing”.’

      That’s the point, It’s been quite a while. So where are they, and I don’t mean tests ?
      It’s a bit of a puzzle that more people don’t see that the bus, not the car, is the obvious app. Same route, dedicated parking spot, cost of self- driving equipment spread over much larger vehicle cost, and currently, unlike Uber. a well paid human driver. Around here they get 50K a year, which easily finances a million dollars, but say 500K to be safe.

      Note that this is NOT the cost of the bus, you’ve got 500K to spend JUST on the self- driving tech… IF IF IF you can operate it without a driver.
      If this worked reliably most buses should be self drivers by now.

      And I think most buses WILL be self drivers at some point, I just disagree with the time frame. I also think that the bus route will be ‘smart’ with beacons and sensors, communicating with the bus.

  45. Boiled Coffee says:

    Self-driving vehicle enthusiasts need to reckon with these key issues:

    1) Bad code can lead to disastrous consequences.
    https://www.theatlantic.com/technology/archive/2017/09/saving-the-world-from-code/540393/

    2) To wit, self-driving cars, while perhaps reducing risk of lower scale accidents, likely increase risk of significantly larger scale fatal catastrophes.
    https://ftalphaville.ft.com/2016/10/26/2178004/the-autoignition-temperature-of-manual-cars-is-much-higher-than-fahrenheit-451/

    3) There will never be a world where self-driving vehicles would not need to coexist with human driven cars, therefore any potential efficiency gains are moot and the poor ability of driverless cars to anticipate and interact with human drivers (e.g. you cannot head nod or wave a driverless car through a four-way stop) will likely make our roads less safe.
    https://ftalphaville.ft.com/2016/10/26/2178004/the-autoignition-temperature-of-manual-cars-is-much-higher-than-fahrenheit-451/

    While I understand why techno-futurists are so keen and want to believe the hype, this is just not the future of transportation.

Comments are closed.