Carlos Ghosn: Fully Autonomous Cars Are Further Off Than Elon Musk Claims, Preview Of Nissan Tech

1 year ago by Mark Kane 46

Carlos Ghosn

Carlos Ghosn

Elon Musk

Elon Musk

Not everyone is happy that Tesla takes the biggest part of the media and consumer attention on autonomus driving technology.

At NAIAS, Carlos Ghosn – CEO of Nissan and Renault – who heavily bet on electric cars and autonomus driving, “pours cold water on Elon Musk’s claim that Tesla will have a fully autonomous car on the roads within three years“.

Nissan hopes to be ready with most of the autonomous stuff in 2020, gradually adding new features beginning this year.

Tesla already began in the second half of 2015 by introducing Autopilot, which consists of a set of assisted driving features, requiring the driver to double check everything the car does.

Carlos Ghosn, when asked about the 24-36 month timeframe for a fully autonomous car from Tesla that will go across U.S. without driver, reacted as follows.

“I understand there is a marketing war, a lot of announcements; that’s fair, everybody wants a piece of it, but at the same time judge on what is on the market.”

“The problem is very complicated; what is an autonomous car? If it’s a question of being autonomous on one lane on a highway or maybe changing lanes, then yes this is 2016, 2017.”

“But if you’re talking about autonomous driving in a city, with crossroads or the car making decisions in complicated situations, then frankly I don’t think it’s going to be ready before 2020. If you are talking about self-driving cars in Palo Alto, it’s not the same as in the real world.”

Well, Tesla already overtook Nissan in current sales volume of electric cars in late 2015, despite Nissan targeting the mass market, so we will move aside and wait to see who is more close to truth. It’s probable that in any scenario, carmakers will be decade or more ahead of rest of the world, which is not ready for fully autonomous cars (insurance, law, regulators, cities and towns).

Video: Autonomous driving in Silicon Valley and planned roll-out on the technology:

Recently released video on autonomous driving (in Silcon Valley) by Nissan:

Source: Autocar

Tags: , , , , , ,

46 responses to "Carlos Ghosn: Fully Autonomous Cars Are Further Off Than Elon Musk Claims, Preview Of Nissan Tech"

  1. Jernej says:

    Ghosn makes sense. Musk usually delivers, but not on time. Too ambitious target dates are for marketing and internal motivation.

    1. evcarnut says:

      I like Nissan’s Concept IDS , But they’d better come up with a car that has DECENT Range …The Mickey Mouse Range They Are 0ffering just will not cut it…Elon Is True & reliable …, It’s the suppliers That are the Culprits , They don’t deliver to him on time so He Can’t to the markets 0n time..Infact He’s suing the Falcon door Design/builders for delaying them 1 yr., He ad to find another source when they Failed Him. It not as easy as 0ne may think ..Infact it’s quite complex .This guy works just about 24/7 ..Its easy to pass judgement…Like passing gas…. l 0 l

    2. Jim Whitehead says:

      Ghosn is often right about the Leaf but his predictions about Tesla are almost all WRONG. Years ago, he said Tesla would always be a niche player. WRONG. Then they would never be more than a few thousand. WRONG. The Model S is now 50K sales as the 2013 Car of the Year. He has no clue what Tesla can do in the future.

      Ghosn in the past ignored his best engineers and cut corners. He never used Tesla like liquid cooling for batteries, and he got the “Arizona problem” of Leaf batteries wilting in heat. Its why you can find some used Leafs very cheap now.

      1. evcarnut says:

        Goshn ,= Sour Grapes over Musk’s Success’s , Talents & Smarts, That, He, Goshn will never even come remotely close to match..& Musk is Half Goshn’s age!

    3. manbitesgas says:

      I trully respect Ghosn for delivering a “mass EV” to the market first and I love our Leaf. But while he stalls 2nd gen and whines about autopilot, the world is leap-frogging Nissan in the BEV department. They better get their act together soon…

  2. Mart says:

    The primary factor for delay will be regulatory, rather than technological.

    1. SparkEV says:

      So true. Get rid of people from the road, and FULLY autonomous could’ve been here years ago.

      But the key word is FULLY. For highway autonomous, it could’ve been here 30 years ago. Anyone remember undergraduate engineer’s micromouse competitions from 30 years ago? That’s pretty much autonomous car if “walls” are things like highway k-rails.

    2. kdawg says:

      I’m not so sure about that. It’s going to take some serious algorithms to get a computer to think like (or better) than a human in complicated driving situations. I know Tesla is collecting a lot of data, but this is still a huge task. Then you have sensor/camera limitations. There’s only so much these devices can do, and they have to redundant and fail-safe.

      Auto-pilot is not as simple as many think. If someone does pull it off, I’m sure the Feds will have a bunch of tests to conduct. The automaker most likely will have done a lot of testing on their own and can provide that data along w/a stat of something like (3 million miles of autonomous driving w/no issues completed).

      1. Pushmi-Pullyu says:

        “It’s going to take some serious algorithms to get a computer to think like (or better) than a human in complicated driving situations.”

        That’s the wrong approach. There’s no reason to try to emulate human behavior in driving. In fact, when all cars on the road can communicate with each other, such things as stop lights will no longer be needed. The cars will interact according to a programmed priority, to allow all to pass through an intersection without danger of collision.

        The basic problem is one of collision avoidance, which is something that computer games already handle quite well. The difficult problems to solve, for expert-systems autonomous software design, will be getting the sensors and the software working in concert to recognize obstacles, to decipher road signs and traffic signals, and to figure out where the edge of the traffic lane is. In certain conditions even humans have a hard time figuring out the latter; for example, when there’s a blanket of snow on the ground, it can become hard to see the edge of the road.

        1. kdawg says:

          Two things.

          1) I also said “better” than humans.
          2) You talk about world where all cars have AI and there are no stop lights. This won’t exist for a very long time. Until then, the car needs to act like a human driver, because that is the data it will be getting.

          Hopefully w/the data Tesla (and others) are getting, they are developing some deep learning algorithms that can do basic situation anticipation. It’s one thing to get the pattern recognition down of the visual cues, and it’s another to be able to predict what is going to happen next.

          1. Pushmi-Pullyu says:

            kdawg said:

            “You talk about world where all cars have AI and there are no stop lights. This won’t exist for a very long time. Until then, the car needs to act like a human driver…”

            Well, the expert system needs to be able to properly react to human the way humans drive, if that’s what you mean. But no, they definitely should not be programmed to mimic human behavior. Humans don’t cooperate when they drive; they compete, and such uncooperative behavior slows down traffic and causes traffic jams.

            For example, when one lane of a highway is blocked and traffic is heavy, drivers will compete to get into the lane of traffic they think is moving fastest. But in moving from a slower lane to a faster lane, the car switching lanes blocks all the cars behind it, forcing the faster lane to slow to the speed of the slow lane. The result is a traffic jam, where all vehicles are forced to move at a speed very, very little faster than the slowest vehicle in the entire group.

            We can be sure that autonomous cars will not be programmed to drive this way!

            I have found quite interesting attempts to use computer programs to mimic traffic flow. What the software engineers have found is that instead of having all the simulated drivers act alike, you get a better simulation of reality if you use several different models of behavior. For example, some drivers are more aggressive, some play it safe.

            Contrariwise, autonomous cars will be programmed to all behave exactly alike, which will benefit everybody by making it much easier for other cars to predict their behavior.

            Traffic will flow much more smoothly, we’ll all get where we want to go faster, and more safely, when all or nearly all cars on the road are self-driving. That will be accomplished by programming cars to cooperate and to take turns, and not programming cars to drive just like humans.

            http://www.academia.edu/9142019/A_new_driver_behavior_model_to_create_realistic_urban_traffic_environment

            1. kdawg says:

              I guess I should have put the the word “good” in front of “human drivers”. It would be idiotic to program the cars to mimic “bad human drivers”.

              1. Raymondjram says:

                GM and other will develop V2V communications so all the vehicles can talk and warn each others of traffic changes. But there will also be some evil hacker who will change the V2v rules such that his vehciles will dominate over the others and take more control, thus become more aggressive.

  3. Dan says:

    I really like how down to earth and sensible Ghosn sounds in his statements. Nissan always seem to deliver close to what they promise – they don’t create too much fanfare and hype and they don’t under-deliver. That is a rare trait in the bluster filled world of marketing spin.

    1. Mister G says:

      I disagree, after borrowing billions of dollars from the US Government Nissan should have a 200 mile EV yesterday not 3 years from now

      1. Dan says:

        You wouldn’t be the first person to mistake Nissan Motor Acceptance Corp (NMAC), which is for all intents and purposes, a financial services company with Nissan Motor Company. The US government did not give any money to Nissan.

        1. MTN Ranger says:

          $1.45 Billion in 2010 for battery factory in TN.

          http://energy.gov/lpo/nissan

          1. Dan says:

            Read the article again. It’s a loan, not a grant. Nissan has to pay it back with interest.

            That would be like saying you owe the federal govt something other than the interest you’re paying just because Fannie bought and securitized your home mortgage.

  4. Alex says:

    Fully autonomous cars are far away from Model S, Model S features are cool but should be named piloted drive.

  5. Mister G says:

    I wouldn’t doubt Musk…he after all brought the world a real electric car and succeeded in landing a rocket without damage. WOW

  6. European point of view says:

    Carlos Ghosn seems to discover Elon Musk is always late , very late , when it comes to action compared to bold declarations .
    Last episode : the presentation of the model 3 shceduled for match will eventually be a draft , with sketches (?)

    1. Anon says:

      Tesla has already implemented many initial autonomous features to their existing fleet by OTA updates. How many autonomous features has Nissa released to their Leaf Fleet?

      How long do you think it will take Nissan to equip their Leaf with the same level of autonomy that Tesla currently has on the roads?

      Perhaps Ghosn’s conservatism is part of why Leaf sales have dropped like a rock in 2015, while being surpassed by Tesla?

      As for the Model III, not sure where you get your information. Not once, has anyone representing Tesla, mentioned the word “Sketches”.

      Thanks again for sharing your European POV. 😉

      1. Mark says:

        Actually he is right… They have backed off the assumption that we would see an actual prototype roll onto a stage this March.

        1. Anon says:

          Saw it. Nowhere is the term ‘sketches’ used. Please redact your earlier Model III comment, if you can’t support it.

          Also, please respond on your erroneous POV regarding Ghosen and their Leaf, in comparison to Tesla’s currently implemented autonomy.

          1. ffbj says:

            It does seem a bit sketchy.

          2. European point of view says:

            I placed a “(?)” after “sketches”

            1. PVH says:

              The stock market seems to believe it will only be sketches (wrong or right).

      2. Kacey Green says:

        According to the roadmap two or three years to match AutoPilot

  7. M. St. J. says:

    Once you drive electric you realize the advantage. But 85 miles is too short. Waiting for the 200 mile EVs before I lease another

  8. Greg says:

    Nissan improves their cars very slowly. It took then 5 years to go from 24 Kwh battery to a 39Kwh battery. So I expect that 2017 Leaf will get a 36 Kwh battery and auto stearing.

    Way to go Nissan.

    1. Greg says:

      Oops. I meant from a 24Kwh battery to a 30Kwh battery.

  9. Lou Grinzo says:

    Carlos is right about the complexity and the likely time frame for what I would call full-bore autonomous vehicles. If anything, I think even he’s being quite optimistic.

    Autonomous cars are one of those things that seems vastly simpler to a casual observer than it really is. I’ve been a programmer in one form or another since the late 70s, and I shudder to think at how hard it will be to engineer and program a fully autonomous car that has to work on current roadways (even with a lot of sensor augmentation) and share them with the (often borderline psychotic) drivers we deal with every day.

  10. Someone out there says:

    I’m with Carlos Ghosn on this one, I think Elon Musk way underestimates how difficult artificial intelligence is. In fact, people have been underestimating it for decades. Already in the 60’s people believed that we would have autonomous home robot helpers “any day now”. We still don’t. It is a very complicated problem.

  11. Tesla “autopilot” is far far far away from real autonomous driving.
    Fully Autonomous means the driver is not required to hover his hand all the time over the wheel and be prepared to take over any moment.

  12. Pushmi-Pullyu says:

    Are fully autonomous vehicles more than three years off? Probably, yes.

    On the other hand, Tesla had the corporate intestinal fortitude to release “Autosteer” in Beta mode, and allow it to be used on all roads, despite the fact that officially it was only intended to be used on limited access superhighways. That puts Tesla ahead of every other company in actually deploying expert-systems (so-called “A.I.”) assisted driving in mass produced cars.

    I think Google is still ahead in the actual ability of their expert-systems self-driving software, but Google doesn’t allow anyone but specially trained drivers operate their self-driving cars.

    It has become clear that self-driving cars will not arrive all at once, as Google is trying to do, but incrementally, with “smart” driver assist features incrementally advancing toward full self-driving autonomy.

  13. dRanger says:

    I think the thing that will help Tesla get regulatory approval quicker is their “shadow mode” data which should be able to prove how Autopilot can reduce accidents by comparing the Autopilot recommendations to what the driver actually did. Although Elon said there have been no accidents with Autopilot so far, that will change. Autopilot doesn’t have to be perfect, just provably better than human control.

    1. kdawg says:

      Yes its a race for the data. Big data FTW.

  14. ydnas7 says:

    Nissan’s autosteer hit the market about a year before Tesla:

    so they probably will release autonomous driving the same time as Tesla.

    Autonomous means a car that can be summoned and the drives a lady to hospital (through the fog, and along road works/sidetrack) while she is giving birth in the back seat, with no other humans around.

    Anything less is just a variation on autopilot/steer/super cruise….

    Autopilot, there is pilot in control

    Autonomous, it will drive by itself, even when inconvenient.

  15. Bloggin says:

    yeah….and I guess Tesla’s estimate of 200 mile EVs was further off, then 300 mile(max range) EVs are even further off. Oh wait….they are here…..it’s a Tesla.

    By 2017 Tesla should be at 400 miles. It’s all about creating more energy dense batteries. Because Tesla’s battery pack already uses all the available space.

    But seriously, I can barely get 300 miles from my gasoline car, so the current battery pack already equals a gasoline car.

    1. Someone out there says:

      Yes but you can replenish those 300 miles very quickly in a gas car, not so in an EV. That is why they need bigger batteries, so you don’t have to do it as often (on longer trips, of course).

      1. GSP says:

        DC fast charge can get 80% in 20-30 minutes. This is plenty quick enough if you have a 300 mile battery.

        GSP

        1. Someone out there says:

          No it’s not if you have to do it 10 times per day

          1. Someone out there says:

            Whoops, I was a little quick there. With a good enough battery, which I believe is around 400 miles, then charging times won’t matter much. 300 miles is decent but not great, we do need to get past that.
            80% charge of a 300 mile battery is 240 miles. You can’t expect there to be chargers at every corner, which leaves maybe 220 miles. Then you have bad weather, snow, hilly terrain, traffic pileups, lead foot and whatnot. those 220 will quickly go down to 150 miles.
            Then it shouldn’t be forgotten that the 300 mile battery is only good for 300 miles when it’s new. 5 or more years later it’s a different story.

          2. Djoni says:

            10 times a day for 80% of 300 mile?
            You should buy an helicopter for those everyday 2 500 miles errand.

  16. martinwinlow says:

    I can only assume Mr Ghosn has not had the opportunity to try the AutoPilot (AP) function on a Model S for any significant time. If he did, he would have the vision and intelligence to realise that full autonomy really is only a very short time away – a few years at most. Whether drivers or the authorities (not to mention the legal system) will be ready that quickly is another matter.

    To all intents and purposes, the Model S already has full autonomous capability. The addition of, for example, the extra order of magnitude (or 2) of positioning accuracy that the european Galileo GPS system will bring would be enough to tip it over the line, on major roads at least. It can already change lanes ‘automatically’ and follow the car in front at a safe, steady distance and it has reasonably good GPS-based mapping. Once AP is activated there is no technical reason why the system could not take you to the off-ramp for your pre-programmed final destination with no input whatsoever from the person in the driving seat. I am talking *now*, not in ‘a few years’ time.

    For those that say it still needs a driver to take over in the event of an issue, you are right but that will very soon become unnecessary. Just consider how many miles have already been driven by the MS under AP control and how few incidents there have been. And bare in mind that despite Tesla’s intention that AP should be supervised at all times (with at least one hand on the wheel), a sizeable proportion of those AP miles have been done without any such supervision. Daft but true. MW