Slashdot Mirror


Driverless Cars Are Giving Engineers a Fuel Economy Headache (bloomberg.com)

schwit1 shares a report from Bloomberg: Judging from General Motors' test cars and Elon Musk's predictions, the world is headed toward a future that's both driverless and all-electric. In reality, autonomy and battery power could end up being at odds. That's because self-driving technology is a huge power drain. Some of today's prototypes for fully autonomous systems consume two to four kilowatts of electricity -- the equivalent of having 50 to 100 laptops continuously running in the trunk, according to BorgWarner Inc. The supplier of vehicle propulsion systems expects the first autonomous cars -- likely robotaxis that are constantly on the road -- will be too energy-hungry to run on battery power alone. A fully autonomous subcompact car like a Honda Fit, for example, will get 54.6 miles to the gallon in 2025 in the best-case scenario, more than 5 miles below the U.S. emissions target, according to BorgWarner. A small pickup or SUV would be at 45.8 mpg, versus a target of 50. Engineers don't have much time to resolve this, as companies are planning to deploy their first fully self-driving cars in the next couple of years. One way for automakers to meet the power-hungry needs of self-driving systems will be to use gasoline-electric hybrid models rather than purely electric cars, said Mary Gustanski, chief technology officer of supplier Delphi Automotive Plc's powertrain business.

28 of 208 comments (clear)

  1. Didn't consider miniaturization? Moore's Law? by bit+trollent · · Score: 5, Interesting

    Really? These engineers didn't consider that processing power is constantly shrinking and becoming more efficient?

    I feel like I'm reading an article from 30 years ago about how computers will never fit inside your home because the take up large rooms!

    1. Re:Didn't consider miniaturization? Moore's Law? by ShanghaiBill · · Score: 5, Insightful

      Really? These engineers didn't consider that processing power is constantly shrinking and becoming more efficient?

      Also, the claim that SDCs will have computers consuming 2-4 kw seems implausible. Tesla Autopilot consumes WAY less than that, and is doing basically the same thing in terms of processing sensory data. Waymo will have access to TPUs that can process vision data eight times more efficiently than GPUs, which themselves consume no where near 2kw. There may be some heavily instrumented prototypes that have 2-4kw of computing power, but that doesn't mean the production version will do that.

    2. Re:Didn't consider miniaturization? Moore's Law? by rmdingler · · Score: 2

      The nascent technology of diver-less, electric autos is such that gains in efficiency are inevitable.

      It also seems likely the technology will garner substantial government subsidies on the order of alternative energy generation, so the leaps in advancement will be measured in bounds.

      --
      Happiness in intelligent people is the rarest thing I know.

      Ernest Hemingway

    3. Re:Didn't consider miniaturization? Moore's Law? by viperidaenz · · Score: 4, Insightful

      Tesla Autopilot is not much more than lane keeping and auto cruise control.
      It can't handle cross traffic without crashing into a truck. It's designed to only be used on motorways. It can't drive through a city. It can't read signs at an intersection to determine who gives way.

      Autopilot != self driving. It does exactly what an autopilot does, keeps you on course at the correct speed, and hands back control to the pilot when it detects a scenario that requires a decision.

    4. Re:Didn't consider miniaturization? Moore's Law? by rtb61 · · Score: 2

      The sensor suite for automated vehicles is key. What they can see, how far they can see it and how often they see it. How durable is it, how reliable is it. They really need to open up the sensor market and share methods until they get one that works reliably, whether passive or active. A encrypted lidar https://en.wikipedia.org/wiki/... suite seems the only way to go, encrypted because each cars signal needs to be unique but it would also be useful to share and read signals from other vehicles (much like dolphins share sonar) and of course with aimed sonar as confirmation for distance (bad idea to share data between vehicles as that can be so readily hacked with false signals, really dangerous). Ideal sensor location might well govern car shape, to ensure sensors are properly located. Sensor testing is also and interesting requirement, pretty much continuous and any failure and the vehicle stops and is a tow away (that stopping being particular tricky, slow down and stop in a straight line or attempt to pull over to the side of the road, keeping in mind your sensors are not working at this time, so that pull over from memory and hope nothing has got in the way).

      --
      Chaos - everything, everywhere, everywhen
    5. Re:Didn't consider miniaturization? Moore's Law? by mysticgoat · · Score: 3, Insightful

      You are on to something with safety in redundancy through separate virtual machines.

      Except of course for the "virtual" part.

      There has to be physical redundancy such that there is backup when something bricks one of the computers. As will happen every so often when self-driving cars are doing millions of highway miles each year. Additionally, self-driving cars need enough AI to bee able to identify the car ahead in the fast lane with the nearly flat rear tire is a threat; that the hail that is bouncing off the road a hundred yards ahead is a danger; and that little Timmy in the back seat who is making urrping noises means that there is an urgent need to stop on the shoulder so you can get his head out the door before he pukes.

      The computer network that has the smarts to handle all the stuff that can happen on the road will easily require 4 kw power, including the cooling system that will keep it functioning when you are stuck in the commuter traffic jam.

      As far as driving on the open road is concerned, the best we can hope for in the foreseeable future is improved auto-pilots, that will still require a trained and experience driver, who is alert and paying attention, behind the wheel.

    6. Re:Didn't consider miniaturization? Moore's Law? by Calydor · · Score: 2

      30 years ago?

      That was 1987. We had Amigas, I think the 386 architecture was in play at the time, the NES console was making its debut ... We already HAD computers in our homes by then.

      --
      -=This sig has nothing to do with my comment. Move along now=-
    7. Re:Didn't consider miniaturization? Moore's Law? by NicknameUnavailable · · Score: 4, Informative

      While it is possible that current designs are using 2-4kw, I'm not buying it is a long term problem. Optimize the software.

      It's possible it's not even mostly software. LIDAR powerful enough to cut through an intense hailstorm or overpower noise/reflections from other self-driving cars from its perspective could easily break 2KW.

    8. Re:Didn't consider miniaturization? Moore's Law? by ShanghaiBill · · Score: 2

      The sensor suite for automated vehicles is key.

      Tesla uses 8 cameras, consuming less than 1 watt each.
      So where are the other 1992 watts going?

    9. Re:Didn't consider miniaturization? Moore's Law? by Blymie · · Score: 2

      And, custom silicon will be the answer for sure -- and it won't even be 'custom' once deployed.

      EG, think of all the massive optimizations in Intel and AMD cpus for the decoding / encoding of mpeg/h.264/5/etc. The same, I'm sure, would be implemented for future CPUs or GPUs, so that custom won't be required.

      I admit I didn't read the article, but the summary is insane. Like the person has zero idea of how many optimizations have occurred, and how much power they save.

    10. Re:Didn't consider miniaturization? Moore's Law? by painandgreed · · Score: 2

      The sensor suite for automated vehicles is key.

      Tesla uses 8 cameras, consuming less than 1 watt each. So where are the other 1992 watts going?

      According to an article here earlier this week, LIDAR, more processing, and all the other systems that car companies say will be required for actual autonomous vehicles to function as opposed to what Musk and Tesla says will work. The Lidar also has to have sensor warts all over the car which also affects aerodynamics and fuel economy. They're building future cars with current equipment because you can't run vaporware through a government testing program. It probably all comes down to in theory, theory is the same as practice, but in practice, they are not.

  2. Easy Solution by freeze128 · · Score: 5, Funny

    The answer is simple: Outsource the processing power to an overseas call center where your virtual driver (Let's call him 'Steve') will steer the car in a simulator-like environment. A few webcams around the car will certainly use less power than that huge LIDAR pod on top of WAYMO cars.

  3. Self driving tech is a waste of money by SumDog · · Score: 4, Insightful

    Self driving cars in Europe might be neat where they don't have transportation shortages. In America, our mass-transportation infrastructure is non-existent. Except for a handful of cities, you have to own a car in order to simply function in society, or you have to find a job that lets you work from home or live in a very limited area of town.

    I wrote a post about this a while back:

    http://penguindreams.org/blog/self-driving-cars-will-not-solve-the-transportation-problem/

    Basically even if you had Interstates which only allowed self-driving cars and all of them could travel at over 120kph bumper-to-bumper and all of them were filled with four people each, you still wouldn't even get to 10% of the capacity of a traditional rail system, running on a single track, with trains arriving at 5 minute intervals (and most cities with rail systems have them arriving at 2 min intervals during rush hour. London has several automated trains. Singapore is fully automated).

    Before we start dumping billions into subsidizing self driving cars, how about we build up our self-driving train tech; a known technology which currently exists and transports millions of people every day.

    1. Re:Self driving tech is a waste of money by kamapuaa · · Score: 2

      That's absurd. Trains full works great for a city like London or cities in the US already serviced by trains. It makes no sense at all even for the Bay Area, and then when you get to flyover states - WTF, who would full trains coming by every two minutes make sense for the typical US small city with extended suburbs?

      --
      Slashdot: providing anti-social weirdos a soapbox, since 1997.
    2. Re: Self driving tech is a waste of money by Nidi62 · · Score: 2

      Because American urban and suburban populations are too spread out for efficient use of mass transportation. I live about 45 minutes from the closest station for my city's rail system, and then probably another 45 minutes ride on the train to the stop closest to my work plus another 15 to 20 minutes to wait for the bus that runs from the stop to my work. Or, i can drive myself and be there in 45-60 minutes. There are many more people like me that would benefit from autonomous cars then there are people who would benefit from increased rail service.

      --
      The only thing necessary for evil to triumph is for it to be pitted against a slightly greater evil
    3. Re:Self driving tech is a waste of money by barc0001 · · Score: 2

      You're completely missing the point of self driving cars. It's not the LONG trips on a highway that this will be hugely impactful for, it's the shorter ones in town. People don't take transit in a lot of cases because they aren't near transit. A SDC can get them that last mile on both sides. Your car takes you to the subway and then drops you off and heads back home to park, for example. Or even better, *you don't even own a car any longer* and just belong to a car share cooperative that can summon an SDC to your door for that trip to the subway, then once it drops you off, it moves on to the carshare's next user.

      I dislike using the word "disruption" to describe what is coming as it's been overused to hyperbole, but it really is the case. Entire segments of the economy are going to be screwed. For example, pay parking companies are going to take massive hits, as will any city or town that relies on moving violations and parking tickets for a significant amount of their revenue. Taxi companies, limo services, car insurance companies, auto body shops, all will see downturns in their services, possibly catastrophic ones.

  4. Even in a pretty small collective. by fahrbot-bot · · Score: 2

    ... the equivalent of having 50 to 100 laptops continuously running in the trunk, according to BorgWarner Inc.

    Apparently, Resistance isn't futile it's V / I.

    --
    It must have been something you assimilated. . . .
  5. Big-rigs will be the first autonomous vehicles by ebrandsberg · · Score: 3, Interesting

    Consider: Many long-haul trucks operate mostly on well maintained highways between distribution centers. Ten or more of them could be linked together scanning for issues, and communicating with each other, possibly with a lead car in the front that can react to accidents and incidents before they even pass, similar to over-sized loads on the road today. Even if not driving between distro centers, they can use existing rest-areas as stopping points where local drivers can take over in shifts.

  6. Re: Very inefficient programming then by Anonymous Coward · · Score: 3, Insightful

    How much power does the omni directional LIDAR array of your cluster use? Oh, it doesn't have one that works at highway speed and distance, much less allow for cross traffic?

    Well then, my welcome mat at home uses zero power and is about as relevant as your anectdote. Thanks for telling us all about your cluster. Do you not have a TV? Oh wait you must, because you would have told us if you didn't. You are that guy.

  7. Nonsense by crow · · Score: 4, Interesting

    This is calculations based on wild assumptions about what is actually required for self-driving. If Tesla is right, then they can do it with the hardware that is already shipping, so power is not an issue--they just need to finish the software. You can also look at what Nvidia just released that they're billing as being designed for self-driving cars. AMD is apparently working on a similar product.

    This sounds like a typical naysayer making stuff up to get attention (and advertising hits).

  8. This is due to using barely altered GPUs by RhettLivingston · · Score: 2

    The NVidia or other GPU approach to AI is too flexible for this application. It needs a more purpose-built chip. Perhaps something like IBM's TrueNorth or even a mixed analog/digital NN approach.

    NNs in general have potential to be much more power efficient than traditional computing with vectors makes it appear, but not when we use traditional computing techniques to simulate the NN.

    We will see this evolve quickly as the market appears. It's still quick and dirty time right now.

  9. They didn't do the math by locater16 · · Score: 3, Interesting

    A Tesla Model S has a 90 kilowatt hour battery life for it's 275 mile range. A kilowatt hour is a thousand watts, and let's assume you're going 60mph. For argument, and maths, sake we'll assume you're going above efficiency and say you'll get 240 miles out of that. That's 4 hours, so 22.5 kilowatts an hour. That's a powerdraw of 22 thousand watts in 1 hour. The new self driving chip announced by Nvidia only draws 500 watts, that's 500 watts in an hour. Or better yet, here's the empirical evidence of Tesla owners discussing their average watts/min usage: https://forums.tesla.com/forum...

    Even there with more efficiency, the new Nvidia chip uses in an hour less energy than the car itself uses in 2 minutes. This article is absolute bullshit, they had 1 damned thing their job required and they didn't do it. Self driving electric cars are perfectly mathematically sounds.

    1. Re:They didn't do the math by Ost99 · · Score: 2

      Current EVs manage a 60% duty cycle on a home charger (22kW), or 85% with a high-speed charger (100-120kW).
      The next CCS charging standard will support up to 350kW, resulting in less than 5% downtime due to charging.

      I'm convinced the savings on fuel and maintenance costs makes EVs better for fleet usage already now, with 15% downtime.

      --
      ---- Sig. gone.
  10. Re: 2-4KW my ass by Dog-Cow · · Score: 4, Funny

    Have you tried?

  11. Re:Not Another Story About Driverless Cars by kamapuaa · · Score: 4, Insightful

    The sort of person who lives at point A, and works at point B, and has to get up in the morning to drive the same route from A to B and then later in the day drives from B to A every day. Driving is an hour and a half a day (mostly) wasted. If I could sleep or work or jerk it for an extra hour and a half every single workday, I feel like my quality of life would be improved. Just that alone would be an extra 400 hours/year.

    --
    Slashdot: providing anti-social weirdos a soapbox, since 1997.
  12. Re:Very inefficient programming then by dgatwood · · Score: 2

    An entire GPU computer for highly parallel tasks that uses 200W? Sweetheart, 200W isn't enough power for a single V100 coprocessor.

    But why would you want one for those purposes? Sure, the V100 is interesting for machine learning because it has tensor-specific hardware as well as a general-purpose GPU, and I could easily see something like that being used during data acquisition and model construction. But for actually running the models, I would expect all of the non-tensor parts of that GPU to be massive overkill—probably by a couple of orders of magnitude power-wise.

    --

    Check out my sci-fi/humor trilogy at PatriotsBooks.

  13. Re:Not Another Story About Driverless Cars by dunkelfalke · · Score: 2, Funny

    It seems to me that since apparently driving a car is your only joy in life, you should be the one who is pitied. Why won't you become a taxi driver?

    --
    "It's such a fine line between stupid and clever" -- David St. Hubbins, Spinal Tap
  14. Re:Not Another Story About Driverless Cars by geekmux · · Score: 2

    Let me guess, couldn't pass the driver's test now you have a grudge against motor vehicles?

    Driving a car is one of the most dangerous things humans do on a daily basis. 40,000 people die every year doing it, so don't be ignorant as to the parents point. The old-fashioned mentality of being forced to drive to a building every day and sit in front of a computer to do a job that could easily be done from home needs to die.

    My commute time was at least 2 hours every day. That's 40 hours a month wasted sitting in a car. I told my employer I would give them half of that wasted workweek back in exchange for taking one hour to do exercise every morning if they would allow me to work remotely. Not only do I maintain my health, but I save over $150 per month in fuel costs, and my vehicle will last much longer.

    Fucking kills me that companies fail to understand that forced commuting wastes hundreds of productive hours every year and contributes to a considerable death toll.