By on March 17, 2017

autonomous hardware

President Donald Trump received a tour of the American Center for Mobility this week. He did not, however, discuss the federal funding of the Michigan-based autonomous testing and development facility. Instead, the site was used as a location for the president to discuss regulatory policies and meet with automotive executives. Little was said on the subject of self-driving cars.

Still, automakers routinely remind us that autonomous vehicles are right around the corner. Ford says it can have autonomous cars rolling out by 2021, Audi and Nissan have said 2020, and Volkswagen has claimed it’ll be ready for self-driving models in 2019. Tesla — which has been pioneering the technology longer than most — has stated it has the hardware necessary in its current production vehicles and would have a bulletproof system installed in 2018, anticipating regulatory approval in 2021. However, suppliers are predicting much less optimistic timelines for self-driving cars — and the dates given vary wildly. 

Mobileye, which was recently purchased by Intel for a massive $15 billion, is one of very few suppliers anywhere near most carmakers’ deadlines. It and Delphi Automotive both plan to provide an off-the-shelf self-driving system by 2019, but it would only be SAE level 4 — a level some manufacturers say they would skip due to minor safety concerns. Many more, including Ford, have said they would also avoid level 3 for the mass-market.

One of Mobileye’s biggest competitors, Nvidia, has said that anything before 2025 is unrealistic. And that year, decided by company co-founder Jen-Hsun Huang, is only possible thanks to automakers accelerating the development of autonomous technology over the last few years. “Because of deep learning, because of AI computing, we’ve really supercharged our roadmap to autonomous vehicles,” he said in a keynote speech to the Bosch Connected World conference in Berlin.

According to Reuters, the company hosting the conference estimates it will take as much as six years longer to reach to the final stage of autonomous vehicle development. Bosch even declined to officially forecast when a totally autonomous car might become publicly available.

Apparently, there are too many gray areas for anyone to say definitively. Many of the dates set by automakers are misleading when you parse out the language used to describe what they mean by “autonomous.” The vast majority of companies will actually only have SAE level 3 capabilities — meaning the car could navigate on the expressway but not in cities and would still require the driver to be prepared to take over at a moments’ notice. Others, like Tesla, say they’re on the cusp of level 5 and won’t need drivers to intervene beyond setting a destination into the computer.

Technology analyst Richard Windsor wrote this week that he remains dubious of any automakers having genuinely autonomous vehicles by 2020, mainly because the liability issue is unresolved. “This is good news for the automotive industry which is notoriously slow to adapt to and implement new technology as it will have more time to defend its position against the new entrants,” he wrote.

This brings us back around the presidential visit. Regardless of how how quickly manufacturers and suppliers can bring the technology to the table, the government is the final piece of the puzzle. These cars will need to be regulated and evaluated before making their way to consumer garages. John Maddox, president and CEO of the American Center for Mobility, was privy to Trump’s discussion with automotive executives but didn’t hear word one about his own project.

Maddox said in an interview with Crain’s Detroit Business that the president was notified of the planned $80 million site’s development but the two did not discuss any funding, despite the facility needing another $60 million to be completed. “Today was not the right time to talk about funding,” Maddox said. “We were able to talk about what we’re doing here in broad terms and about how our automotive needs to stay competitive and to do that we need to build facilities like this … and others.”

[Image: Ford Motor Company]

Get the latest TTAC e-Newsletter!

23 Comments on “Suppliers Say Automakers are Just Guessing the Timeline for Self-Driving Cars...”

  • avatar

    I still see the main two obstacles for self-driving cars as A.) cost and B.) a (crumbling) infrastructure that was not designed with them in mind. Seems like there is all the development in the world on the cars and their tech but zilch anywhere else.

    And, you have to keep in mind the vast majority (distance wise) of the country where public transportation is not an option and the working poor must drive themselves (in their 15-year-old GM 3800 W-Body or 2002 Dodge Neon) where they need to go. How are they going to afford self-driving tech? Simple. They’re not.

    • 0 avatar

      Most of the cost is in the software development. A couple smartphone-grade cameras and some sensors cost very little these days.

      • 0 avatar

        Can they be retrofitted to be used on turn of the century Chevy Impalas and Ford Escorts? If not, that leaves a lot of people out. Keep in mind, the average age of cars on the road in the U.S. is reaching 12-years-old now.

      • 0 avatar

        Did you see the picture at the top of the page? Saying those are a “couple smartphone-grade cameras and some sensors” is like saying a Saturn-V rocket is just some fuel in a tube with an igniter on the end.

      • 0 avatar

        as the couple of Autopilot mishaps have shown, you need more than a “couple of smartphone grade cameras and some sensors.” There’s a reasons actual self-driving car prototypes use a mix of technologies; LIDAR, RADAR, and ultrasonic sensors can “see” things which photographic cameras can’t.

        Just look how quickly George Hotz gives up and bails any time the problem becomes too difficult.

  • avatar
    SCE to AUX

    Liability issues will ultimately limit or stop self-driving car deployment.

    Mfrs won’t write blank checks to plaintiffs, and buyers won’t utilize or even purchase the tech if they’re held liable.

    On this subject, both Volvo and Tesla are dreaming.

  • avatar

    If truly self-driving cars are unveiled and sold in the mass market, with sufficient infrastructure improvements to ensure that they are able to function even remotely satisfactorily, before the year 2040, I’ll be genuinely shocked.

    Also, when that moment arrives, the Jetson’s flying cars with flux capacitor propulsion systems will have already arrived and 86% of the population will be living in vertical cities and have Rosie The Robot Maids/Nannies.

  • avatar

    No, they said they’d skip Level 3.

  • avatar
    Land Ark

    I had a site visit with a representative from the Federal Highway Administration at their campus just outside DC (directly next to the CIA HQ in Langley) about a year and a half ago. They wanted me to help identify the locations of our facilities so they could determine what it would take to do some work on their campus. I inquired about what they were doing and I was told they were setting the whole place up to be able to do testing of self-driving/autonomous cars.

    I didn’t get any of the details of course, but it definitely seemed like at least they were gearing up to be able to help support the need should it arise.

    Edit: And I just looked at a satellite image and I see they’ve already done the work. If you check it out, you can see they put in some crosswalks and 2 traffic lights. That’s all supposed to simulate city driving situations.

  • avatar
    Master Baiter

    Tesla’s auto pilot is like the gull wing rear doors on the Model X: It’s an otherwise worthless parlor trick you can show off to your friends.

    I talked to a guy yesterday who owns a Model S. He likes the car, but he said he can’t trust the auto pilot. It gets confused at times, in particular when going through intersections where the lanes shift direction.

    And I agree that liability issues will kill the technology.

  • avatar

    Of *course* they’re guessing at the timelines. Do you mean opposed to all the prescient truthsayers out there who can already see the future? Can we have less Hyperbolic headlines, please?

    As to the liability issue – lets not forget that this isn’t done in a vacuum. If autonomous vehicles begin to reduce the fatality rate, then there will be significant inertia, possibly at the regulatory or government level (ok, ok, everywhere BUT the US) to deal with the liability issues. these systems are going to save lives, and there is an ever-aging population that would very much like to continue to have autonomy thankyouverymuch.

    Yes, the combination of legal issues in the US and current political climate may well prevent it from taking hold south of the border, but it’s not going to stop progress in the rest of the world, and these multi-billion-dollar bets that the likes of Intel are making aren’t going to be stifled just because they have to go to Canada or the EU to be realized.

    • 0 avatar
      Matt Posky

      Automakers are promising deployment on specific years, when those dates don’t match the timeline offered by suppliers it’s my job to say so. Inertia (with a whole lot of money) is already pushing this technology forward and TTAC publishes examples of this throughout the week. It’s absolutely coming, as you said. However, we have to acknowledge that that there are inconstancies cropping up while manufacturers rebrand themselves as mobility companies.

  • avatar

    Isn’t the more significant and difficult accomplishment going to be getting meat drivers off whatever roads upon which future AVs are deemed fit to operate?

    Don’t want the nice Eloi getting et by Morlocks =:-o

  • avatar

    If the Washington DC Metrorail system can’t restore automated operation after nearly 8 years since the multiple-fatality system error in June 2009, how can anyone foresee safe self-driving cars (even if older cars could be feasibly retrofitted) operating in a trackless system?

    Assuming a system with tracks can indeed be made foolproof to avoid another 2009-type crash, it seems to me that a feasible system with cars would involve embedding “tracks” directly in roads, along with heating elements to prevent the sorts of damage that freeze-thaw cycles, deicing chemicals, etc., can do to roads. Costly, yes, but so are the present roadbuilding + maintenance cycles; we all want infrastructure that is less likely to crumble. Also, a system that depends on communication between an individual car and its “track” wouldn’t be as likely to be hacked, as car-to-car communication wouldn’t be necessary.

    • 0 avatar

      I agree. The concept of extensive (and expensive) sensor packages to spot every possible threat and oportunity in a very complex environment is a huge problem. I would think a much more feasible solution would be implanting homing / tracking wires in roadways (at least designated ones) and let the cars follow those as primary navigation. And use the optical/radar sensors as they are currently used – mostly for automatic cruise and braking. And only use those sensors for navigation as a last resort.

  • avatar

    All I would like is autonomous driving in low speed highway traffic. That is it. Maybe I’ll see that.

    If we don’t have completely autonomous planes and completely autonomous trains by now…how are we supposed to have autonomous cars? Jack brought this up a few weeks ago. You’ll see a plane that is completely autonomous before you will see a care that is completely autonomous. The operating environment is far too complicated.

    A lot of this is talk by auto companies is HYPE to raise STOCK PRICES.

  • avatar

    I hope to never see these cars on the roads. I can’t imagine the damage that can be done by one good hack.

    • 0 avatar

      Did you see the Tesla on autopilot slam into the Jersey barrier a couple of weeks ago? No hacking required!

      How the car even with today’s technology can completely miss the fact that the lane is gradually disappearing (or shifting to the right for construction) is deeply troubling.

  • avatar

    “Tesla — which has been pioneering the technology longer than most — has stated it has the hardware necessary in its current production vehicles and would have a bulletproof system installed in 2018, anticipating regulatory approval in 2021”

    I admire Tesla but I hope they knock this BS off soon. Putting a hard deadline on vaporware leads to buggy roll outs, and in this particular case that means people dying. Sure, they may be most of the way there, but the part that’s left to figure out is the really hard stuff, it’s going to take trial and error and you can’t just timeline it out based on previous progress.

  • avatar

    Had just this conversation today with a systems engineer…Does the car kill you, or kill another ?

    If it is programmed to sacrifice the driver, no one will ever buy one, yet….
    If it is programmed to kill the pedestrian/other motorist, is the owner, the maker, or the software programmer responsible ?

    Can it tell the difference between a dog/deer crossing the street (kill), or a child (no kill, subject to the above question).

Read all comments

Back to TopLeave a Reply

You must be logged in to post a comment.

Recent Comments

  • Garak: I could see quite a few places where you’d might want an electric model. Countries with expensive gas,...
  • DungBeetle62: Don’t know whose apple my Dad polished but in the early 80s after a parade of awful Cutlasses...
  • JD-Shifty: lowest gas prices were under Clinton. But that’s none of my business. We’ve seen wild price...
  • Buickman: anyone notice AutoNews has eliminated their comment section?
  • slavuta: You know what is pathetic – you keep spreading lies that Russia created some kind of holodomor. My...

New Car Research

Get a Free Dealer Quote

Who We Are

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