Stories
Slash Boxes
Comments

SoylentNews is people

posted by martyb on Thursday August 10 2017, @06:04PM   Printer-friendly
from the semileaks dept.

https://arstechnica.com/cars/2017/08/reuters-tesla-looking-to-start-testing-autonomous-semi-in-platoon-formation/

Tesla CEO Elon Musk announced in April that the company is working on pushing a long-haul electric semi truck to market, which is set to be formally revealed in September. Now, Reuters has viewed e-mail correspondence between Tesla and the Nevada Department of Motor Vehicles that indicate that the company has discussed testing semi trucks on the state's roads.

The Reuters report also mentioned that the semis would be outfitted with autonomous functions, so they could traverse the nation's highways without a driver in the front seat. The e-mails seemed to indicate that Tesla's semis would "platoon," that is, drive in a formation such that a number of trucks could follow a lead vehicle. It's unclear whether the lead vehicle would have a driver, or operate autonomously with a person in the front seat to monitor safety.

[...] Reuters also reported that California DMV officials will meet with Tesla this week "to talk about Tesla's efforts with autonomous trucks."


Original Submission

 
This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 2) by fyngyrz on Thursday August 10 2017, @08:20PM (7 children)

    by fyngyrz (6567) on Thursday August 10 2017, @08:20PM (#551847) Journal

    For trucks that big, the follow distance does not have to be as tight as you imagine.

    For trucks that big, the stopping distance is significant, and particularly so when heavily loaded. I'm quite familiar with the distances involved in drafting trucks. I've done it using semis, pickups, sedans, and motorcycles. There's not nearly enough room to stop if the truck in front suddenly piles up. That's why I don't do it any more. :/

    There's no minimizing the potential problems if trucks are drafting in the usual fashion and the leader impacts something without warning. There's going to be a considerable mess.

    Starting Score:    1  point
    Karma-Bonus Modifier   +1  

    Total Score:   2  
  • (Score: 2) by bob_super on Thursday August 10 2017, @09:19PM (6 children)

    by bob_super (1357) on Thursday August 10 2017, @09:19PM (#551876)

    > There's not nearly enough room to stop if the truck in front suddenly piles up.

    That's only if you rely on seeing the truck brake before you act, losing precious seconds. With automated systems, the first truck deciding to brake can instantly trigger all of them braking.
    The question becomes: does hitting the obstacle cause the front truck's braking distance to be reduced by more than the distance the second truck was following at? (for equivalent loads)
    In most cases, because trucks are much heavier than random obstacles, the answer will be no. Occasionally, you'll end up with a pileup.

    • (Score: 2) by frojack on Friday August 11 2017, @12:04AM (3 children)

      by frojack (1554) on Friday August 11 2017, @12:04AM (#551960) Journal

      because trucks are much heavier than random obstacles,

      But roads are not exactly full of random obstacles.

      Trucks are likely to hit bridge support, cars, and Other Loaded Trucks, or worse, on-coming loaded trucks.
      In at least three out of four of those have a good chance of stopping a truck shorter than brakes could ever muster.

      --
      No, you are mistaken. I've always had this sig.
      • (Score: 2) by bob_super on Friday August 11 2017, @01:00AM (2 children)

        by bob_super (1357) on Friday August 11 2017, @01:00AM (#551992)

        Arguably, the incapacity of the trucks to properly handle an A380 doing an emergency landing on a highway is a good reason to avoid automating trucks.

        Weird shit happens. If someone drops a bridge support in the middle of the road, or an on-coming truck swerves in the way, maybe you do want a train of driverless trucks to get mangled, rather than the current solution of losing a train of driven trucks.

        • (Score: 2) by fyngyrz on Friday August 11 2017, @01:58PM (1 child)

          by fyngyrz (6567) on Friday August 11 2017, @01:58PM (#552270) Journal

          If someone drops a bridge support in the middle of the road, or an on-coming truck swerves in the way, maybe you do want a train of driverless trucks to get mangled, rather than the current solution of losing a train of driven trucks.

          Nah, you just don't want them drafting or otherwise following too closely. With proper following distances, only the first truck will crash. That's why we're taught about proper following distances.

          Mind you, the majority of the drivers I've observed out east and on the west coast ignore the entire idea of proper following distances, but hey, that's just evolution in action: taking themselves out of the reproduction and/or support chain.

          The premise of the convoy following at distances adequate to save fuel is bankrupt based on safety concerns. Any such behavior is inherently unsafe.

          • (Score: 2) by bob_super on Friday August 11 2017, @04:14PM

            by bob_super (1357) on Friday August 11 2017, @04:14PM (#552397)

            Except that it's not, because saying the second automatic truck will crash is the same logic as saying that the second half of the first trailer will crash (or the second trailer on current dual-trailer setups), so it should be in a different truck.
            Pushing that logic, the only logical setup is a bunch of automatic cars, carrying tiny payloads, and following each other with hundreds of yards of spacing.
            Those Australian coal train-trucks with 5 or 7 trailers? Can't have that, we' could lose more than one payload in a crash!
            Trains? forget it, those lose dozens of cars in a crash!

            As I said already, if you network them and put the longest-braking in front, your odds of more than one crashing are low enough, and the gains of having only one guy at the front (or two, for non-stop, and eventually zero) will make it a compelling solution for the fleet operators.

    • (Score: 2) by fyngyrz on Friday August 11 2017, @01:40PM (1 child)

      by fyngyrz (6567) on Friday August 11 2017, @01:40PM (#552260) Journal

      That's only if you rely on seeing the truck brake before you act, losing precious seconds.

      No. It isn't. At 55 MPH, you're traveling at ~80.6 feet per second. The truck in front hasn't "braked"; the truck in front has collided with another vehicle, perhaps one that crossed the centerline. Or it has fallen into a sinkhole. It went from 55 MPH to a pile of crumpled metal. Something without warning to the leader. If the following truck is instantly informed of the problem by whatever means: communications, sensors, it has (following distance / 80.6) seconds to come to a halt at 55 MPH. If the truck is loaded, as it will surely be for at least half the journey it takes (and all of it if the truck's itinerary is managed well), the stopping distance is significantly extended. Each following truck in the convoy has the same problem based on whatever their following distance is - they have no extra time, because they have no extra warning. If the 2nd truck can't stop in time, neither can any of the others if they are following at the same distance.

      Drafting efficiency has been measured using an automobile (I couldn't find an example with a semi) but let's roll with it: Measured drafting effectiveness for reduction of fuel consumption at 55 MPH is approximately, according to our friends at mythbusters [autoblog.com]:

      100 feet: 32.0 MPG - baseline
       50 feet: 35.5 MPG - 11%
       20 feet: 38.5 MPG - 20%
       10 feet: 44.5 MPG - 27%

      So at 50 feet, for an 11% savings, at 55 MPH (for a car, savings for a truck will likely be considerably less, though following distances will be similar because they are based on getting into the spot behind the leading truck's displacement of the air mass, which doesn't change based on what is following), available stopping distance with instant knowledge of the collision is 50/80.6=.61 seconds. Now ask yourself: Can you imagine a fully load semi coming to a complete halt in about half a second? Again, this assumes zero delays in the information that the following truck needs to come to a complete halt, 55 MPH (likely faster, so less time) and the application of full stopping capability.

      So there's the math. I stand by my earlier assessment. Thump thump thump thump.

      • (Score: 2) by bob_super on Friday August 11 2017, @04:33PM

        by bob_super (1357) on Friday August 11 2017, @04:33PM (#552404)

        Whoa, lots of typing, yet ignore fundamental physics...
        Step 1: you don't bring 40 tons to a stop instantly, unless you run head-on into a very solid wall (I'm ignoring crumple time to please you). Even when hitting a truck swerving from the oncoming lane, that takes a lot of precision to hit exactly centerline exactly straight. The front truck may slow down faster than the second can handle, but every fraction of a second gained is a chance to avoid a chain reaction.
        Step 2: Why exactly do you assume that something which can stop a truck on a dime somehow cannot be detected a second or two or five before it does? Crunch! You didn't sense a fucking wall, so now your followers are in trouble... A few engineers and test engineers need to be fired!
        Step 3: Getting back to my original statement, I didn't imply it was foolproof, but would work in most cases and wrote that you may get a pileup in extreme circumstances. But your objection is ridiculous, because if we worried about sinkholes and pop-up-insta-truck-stop obstacles on our roads, we would never, ever, get in a vehicle, and probably not even on a horse.

        Also note that trucks being shaped like bricks (in most countries), drafting benefits are higher than you think. A car not drafting has a pretty good Cx.
        See also my other reply to you about the horror of potentially losing multiple payloads.