Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 


Forgot your password?
Close
typodupeerror
×
Transportation

Ford Self-Driving R&D Car Tells Small Animal From Paper Bag At 200 Ft. 207

Posted by samzenpus from the where-you-going? dept.
cartechboy writes "Autonomous driving is every car manufacturer's immediate R&D project. In car-building terms, even if a new technology isn't due for 10 years — since that's just two full vehicle generations away-- it has to be developed now. So now it is for autonomous car research and testing, and this week Ford revealed a brand new Fusion Hybrid research vehicle built for autonomous R&D with some interesting tech capabilities. Technologies inside the new Fusion Hybrid research vehicle include LIDAR (a light-based range detection), which scans at 2.5 million times per second to create a 3D map of the surrounding environment at a radius of 200 feet. Ford says the research vehicle's sensors are sensitive enough to detect the difference between a small animal and a paper bag even at maximum range. More road-ready differentiations include observation and understanding of pedestrians, cyclists, and plain old stationary objects. Ford is working on this project in cooperation with the University of Michigan."
This discussion has been archived. No new comments can be posted.

Ford Self-Driving R&D Car Tells Small Animal From Paper Bag At 200 Ft. More

Ford Self-Driving R&D Car Tells Small Animal From Paper Bag At 200 Ft.

Comments Filter:

  • Mis-read the title (Score:3, Funny)

    by Maow ( 620678 ) on Friday December 13, 2013 @06:35AM (#45678739) Journal

    I read it as "Ford Self-Driving R&D Car Smells Small Animal From Paper Bag At 200 Ft." and my first thought was, "What the hell kind of test is that?!?"

    Split second later, "Waaiit a second, that can't be right."

    But hey, my truck smells like a small animal in a paper bag - from 2 years ago.

    *goes back to sleep*

    • Re: (Score:2)

      by sycodon ( 149926 )

      Can we get a "swerve to hit cat" mode?

    • This is funny. But there is something about it that is not.

      200 ft. at highway speeds is less than 2 seconds. In order to avoid said small animal, the car would have to rather radically swerve, and disrupt both traffic behind it, maybe in adjacent lanes as well, and would most certainly disturb the occupant(s) of the car.

      It has to do better than that to be practical.

  • Yes but (Score:5, Funny)

    by Rosco P. Coltrane ( 209368 ) on Friday December 13, 2013 @06:36AM (#45678743)

    Can it tell if the small animal is *inside* the paperbag? I'm thinking of cats specifically. Cats and paperbags... cat lovers know what I'm talking about.

    • I'm thinking of cats specifically.

      Actually, I was thinking more Monty Python "Four Yorkshire Men":

      "When I was a lad, we lived in a paper bag in the middle of the road . . ."

    • Re: (Score:2)

      by Radres ( 776901 )

      In China, only old people tell small animal from paper bag.

    • I don't think so, but given a speed of 60mph It will be able to tell you if it's alive or dead in about 2.27 seconds.

    • Can it tell if the small animal is *inside* the paperbag? I'm thinking of cats specifically. Cats and paperbags... cat lovers know what I'm talking about.

      You got modded "funny" but you make a good point. Unless you know what's in the paper bag, you should try to avoid it providing you don't have to do something even more dangerous. And you never know what's in the paper bag . . .

    • Cat haters presumably are more familiar with burlap sacks?

  • salt and de-icer (Score:4, Insightful)

    by hunter44102 ( 890157 ) on Friday December 13, 2013 @06:42AM (#45678763)
    lets see what it can detect in the Northeast after 2 days of snow, salt and de-icer puts a 'film' of gunk covering 90% of the vehicle

    • You miss a crucial point : no small animal will be out and about when it snows in the northeast - and paperbags will get soggy in a hurry too - so the detection feature is not needed in those conditions.

      • Right, because detecting paperbags and small animals are the only things a self driving car needs to be aware of. Certainly not objects like street signs, lane markings, people or rusted off car parts.

    • That's a good point. My car currently looks like it's covered with dirty white sand and it's get covered with snow when I drive. My backup camera is useless unless I clean it every time.

    • lets see what it can detect in the Northeast after 2 days of snow, salt and de-icer puts a 'film' of gunk covering 90% of the vehicle

      Northeast. What about Michigan? At least this is being developed by people who don't live in sunny California. Seriously, that may have an effect. I keep asking how Google cars do in a rainstorm, but they don't seem to understand the concept.

      OTOH, if the ability of a car to handle various types of weather reflects where the designers live, then why were most American cars for many years so bad in the snow? Seriously, never figured that out. US cars were amongst the last to adopt FWD (although IIRC Mercedes

      • I love RWD in the snow. I'd say the width of your tyres mattes more than the drive system. It also depends how much snow you get I suppose. Snow is never a problem for me, but ice can really suck when it's on an incline (as in the car park at my last flat, where I had a lot of fun trying to get going some days..).

        • I love RWD in the snow.

          Why? FWD has better traction because more weight is over the drive wheels, and it's more stable (when rear drive wheels slip the car fishtails). As for tire width, I never noticed that. The cut of the tread is another matter. We usually have only moderate snow here in the NYC area (though a few feet once in a while) so with modern all-weather tires I don't bother with snow tires. If you do get real snow tires though, they work great (used to use them before all-weather tires were as good as they are today).

          • Why? FWD has better traction because more weight is over the drive wheels, and it's more stable (when rear drive wheels slip the car fishtails)

            Well, for one thing, I enjoy drifting/fishtailing when it's raining or there's snow (I only do that if there aren't other cars around though). Having weight over the drive wheels is pretty good for grip yes, but having the drive wheels also doing the steering is not a good thing, especially in unexpected situations. I suppose that a driver that's aware of the limitations of their vehicle will always fare better in poor weather than someone who knows nothing about drive systems and weight distribution, so it

          • Why? FWD has better traction because more weight is over the drive wheels, and it's more stable (when rear drive wheels slip the car fishtails).

            Depends on the weight distribution in the particular car. Drive something like a BMW where the weight distribution is close to 50/50 and you don't have such an advantage from FWD. The type of drive system (FWD, RWD, AWD, 4WD) only actually matters when accelerating. FWD works better for many people for exactly the reasons you mention but a RWD rear engine car (porsche 911) can get excellent traction for the same reasons. Really you want AWD or 4WD if traction under acceleration is a big concern. But a

      • Re:salt and de-icer (Score:4, Informative)

        by MightyYar ( 622222 ) on Friday December 13, 2013 @08:57AM (#45679179)

        Jeep? That's a pretty American brand. :) Most of the time a rear-wheel drive vehicle with some extra weight in the trunk and a set of snow tires was pretty decent in the snow. My dad is from Pittsburgh, and if he could get around snowy hills with that configuration, I'm pretty sure other folks could too.

        Four wheel drive was complicated and expensive, and you ended up with an extra bulge and shifter on your floor. FWD was and is pretty crappy for handling in all of the rest of the year, with a few standout exceptions. FWD is cheaper and gives you a flat floorpan - that is the primary reason why it was adopted.

        • FWD is cheaper and gives you a flat floorpan - that is the primary reason why it was adopted.

          Well, it's supposed to be cheaper on gas since the engine is pulling you forward instead of pushing.

          But, I've been in a few North American cars which, despite being FWD, have a big hump going through the middle for no good reason, the same as if there was a drive shaft to the rear wheels.

          It made absolutely no sense, and the only thing I can figure is the car company decided it would be more expensive to retool the p

          • Well, it's supposed to be cheaper on gas since the engine is pulling you forward instead of pushing.

            Never heard that one, although I think the extra friction of the drive shaft is a very slight inefficiency. FWD cars also weigh slightly less too, due to the absence of a drive shaft and separate differential housing.

            I've been in a few North American cars which, despite being FWD, have a big hump going through the middle for no good reason, the same as if there was a drive shaft to the rear wheels

            Were those models available in 4WD or AWD? They might have just used the same floor pan for both. Still a bit sloppy, but marketing probably figures if you have the hump anyway, you're more likely to pay the extra for AWD.

            • Were those models available in 4WD or AWD?

              Nope. Front wheel drive only. Which means there is no sensible reason other than laziness/being cheap idiots why you would have the hump in the middle.

            • I've been in a few North American cars which, despite being FWD, have a big hump going through the middle for no good reason, the same as if there was a drive shaft to the rear wheels

              Were those models available in 4WD or AWD? They might have just used the same floor pan for both. Still a bit sloppy, but marketing probably figures if you have the hump anyway, you're more likely to pay the extra for AWD.

              Even FWD cars tend to need a little hump to accommodate the tailpipe, but if it's a big hump then your expl

          • Well, it's supposed to be cheaper on gas since the engine is pulling you forward instead of pushing.

            You'll have to cite a source for that because the wheel doesn't know whether it is pushing or pulling and despite being an engineer I cannot think of any physics reason why pushing versus pulling would make a spec of difference in fuel economy. FWD cars are often less performance oriented which means they might be designed with fuel economy higher on the priority list but that isn't an inherent advantage of FWD over RWD.

            But, I've been in a few North American cars which, despite being FWD, have a big hump going through the middle for no good reason, the same as if there was a drive shaft to the rear wheels.

            Which ones? I can't think of any that fit that description and I'm kind of a car nut.

          • The hump is there to run the exhaust and to hold the catalytic converter/muffler/etc. It's above floor pan level to protect it.

            I one cracked a head because I crushed a catalytic converter when going over a speed bump (hanger broke and the exhaust was sagging). Engine overheated on the exhaust side and the temp gauge never raised.

        • My dad is from Pittsburgh, and if he could get around snowy hills with that configuration ...

          It's not just a question of whether you can, but which is the preferable approach.

          Four wheel drive was complicated and expensive

          Agreed. AWD is very popular these days, but unless you live in a very snowy area it doesn't make sense. I don't need two extra differentials, two extra half-axles, and 6 extra U-joints (the last being the most likely source of trouble).

          FWD was and is pretty crappy for handling in all of the rest of the year, with a few standout exceptions.

          If there are a few standout exceptions, it proves it's not an inherent problem with FWD. The whole "bad handling" thing just isn't an issue in the 21st century, unless you have a serious sports

    • while (visibility 85 && hasfluid)
      {
              hasfluid = applyfluid();
              whipe(3);
              sleep(1);
              whipe(1);
      }
      if (!hasfluid) {
          auto = 0;
          manual = allertHalt("Out of fluid, Car is stopping, do you want to switch to manual?", "Yes", "No")
      }

  • If production is 10 years from now, we will have hit the concrete wall by then. 2.5 million scans per second is not going to get processed by a 10$ chip. It will be interesting to see how the end of Moore's law will affect this and similar projects.

    • Re:Moore's law (Score:4, Interesting)

      by ebno-10db ( 1459097 ) on Friday December 13, 2013 @08:06AM (#45679029)

      It will be interesting to see how the end of Moore's law will affect this and similar projects.

      Maybe programmers will learn the nearly lost art of writing efficient code.

    • The production of the flying car has always been just 10 years in the future, for about 60 years now. Wonder if this is the same situation.

      • Probably so. Of course in the 60's they also predicted that flat screen TV's were 10 years in the future, so maybe we'll eventually get it.

    • We must be pretty close to some sort of concrete wall right now. I just replaced a 3-year-old MacBook pro that had a 4 core 2.3 GHz Intel Core i7 processor. The new one has, wait for it... a 4 core 2.3 GHz Intel Core i7 processor. I know, MHz myth, yadda yadda yaddah... the fact is, it is hardly any faster. I've been using computers since the C64 and this has never happened before. (Granted, the new one is smaller, has better battery life, twice the RAM, the SSD has twice the capacity, and the screen

  • Noise (Score:5, Interesting)

    by jklovanc ( 1603149 ) on Friday December 13, 2013 @07:12AM (#45678873)

    I wonder about all these active technologies; lidar, radar, ultrasonic, etc. They work very well when there is only one vehicle in the area. What happens on a crowded freeway when there are a couple hundred vehicles an the area pumping out all those emissions? Wouldn't it be difficult to differentiate between returns due to your emitters and the emitters from other vehicles? Unless each emitter is working on a different frequency interference is a possibility. There is also the issue of sensors being sensitive enough to detect return but filtered enough not to be dazzled by the direct emissions from other vehicles close by.

    • Re:Noise (Score:4, Interesting)

      by AmiMoJo ( 196126 ) * on Friday December 13, 2013 @08:48AM (#45679151) Homepage Journal

      It generally isn't a problem because single readings are never used. They are always averaged over time and combined with other sensors. They also pulse their output and can detect interference and adjust their timing randomly to avoid it.

      Think about how many devices manage to share unlicensed radio spectrum and how few cars will be that close together. The reason for having so many sensors is that if any one fails the others can make up for it.

      Of course it will still fail from time to time, but less than a human.

      • It generally isn't a problem because single readings are never used.

        Sorry but as far as I have seen in designs there is one sensor for each area with little or no overlap. Also multiple scanners would increase the issue.

        Think about how many devices manage to share unlicensed radio spectrum

        That works because most devices do not transmit continually and there is a great deal of dead space. Whenever there is a collision the data is just re-transmitted. That is very different than an area scanner which transmits continually.

        how few cars will be that close together.

        Take an 8 lane highway at rush hour in bumper to bumper traffic with a 200 foot range. With a car every 20 feet that is 40 c

    • Re:Noise (Score:4, Interesting)

      by swb ( 14022 ) on Friday December 13, 2013 @11:10AM (#45680171)

      My Volvo S80 has radar for the collision avoidance feature and the distance-sensing cruise control.

      The only problem I've ever had with it has been in snowstorms where the radar panel gets covered with snow and ice -- the dash display will then show "radar blocked."

      On the other hand, my Valentine 1 radar detector will false on other cars radar detectors and some automatic doors on commercial buildings.

      About the only other problem I've had with the distance sensing cruise has been getting behind cars driving slightly slower than my set point and not noticing that I'm going a little slower than I want to drive. My car will basically follow the other car and match its speed transparently until it goes faster than my cruise set point.

      • A radar with a collision avoidance feature and the distance-sensing cruise control is a very simple device. all it does is look for an object that is directly in front of the vehicle and measure the distance. It creates a very course picture of the road. That picture would have to be orders of magnitude finer to be usable to guide a car along a road.

  • What are they really saying? (Score:5, Insightful)

    by jklovanc ( 1603149 ) on Friday December 13, 2013 @07:28AM (#45678911)

    Ford says the research vehicle's sensors are sensitive enough to detect the difference between a small animal and a paper bag even at maximum range.

    Given that the sensors can detect a difference here are some follow on questions that seem important.
    1. Can it detect which one is the animal and which one is the bag? (they talk about difference not identification)
    2. Can it tell if the small animal is alive or dead if it is not moving.
    3. Can it tell if the animal is on a leash and not going to be an issue?
    4. Can it detect if there is a barrier between the animal and the desired route of travel and the animal not being an issue?
    5. Can it tell the difference between a turtle and a rabbit? Turtles having much more restricted movement possibilities than a rabbit.
    6. Will it remember that the small animal went into the bag. Out of sight out of mind.
    7. Can it differentiate between an empty bag and a bag of cement? Driving over an empty bag is not a problem. Driving over a bag of cement is probably a problem.
    Detecting the difference between a small animal and a paper bag is important but it is only the first step in in a very complex decision process to determine what to do with that information.

    • What they are really saying is that they are programming their cars to identify an certain object, and in certain cases ignore it and just drive over it.

      Ford: "We are terribly sorry our autonomous car ran over your baby. To the car it looked like a paper bag. Next time dress your child in brown fur, and we promise we'll go around it."

      • The article does not say they can identify which is the bag and which is the animal it just says they can detect a difference. In an experiment where there was a paper bag and a small animal in the field of view of of the sensor there are at least two outcomes.
        1. The sensor returns "there are two different kinds of objects out there". Therefore the sensor has detected a difference between a paper bag and a small animal. That is exactly what the claim states.
        2. The sensor returns that object a is a paper bag

    • All good observations, because I'm not so concerned about it avoiding one-off obstacles encountered serially as I am about the choices it makes between two bad options, like the braking path that goes through a small dog vs. a crawling toddler.

    • Re: (Score:2)

      by fnj ( 64210 )

      Given that the sensors can detect a difference here are some follow on questions that seem important.
      1. Can it detect which one is the animal and which one is the bag? (they talk about difference not identification)
      2. Can it tell if the small animal is alive or dead if it is not moving.
      3. Can it tell if the animal is on a leash and not going to be an issue?
      4. Can it detect if there is a barrier between the animal and the desired route of travel and the animal not being an issue?
      5. Can it tell the difference

      • That type of thinking never crossed my mind till I started hauling horses. Before, I would consider doing "something" to avoid hitting some small animal. Now when I am going down the road with @ 4000 lbs of horse and trailer (and another 1500 lbs of truck) and I see a small animal in the road I say a quick prayer for it as I am not doing anything but continuing straight ahead.

        One day this idiot passes me on a double yellow section of secondary road. The car then pulls in front of me and then has to hit t

        • That type of thinking never crossed my mind till I started hauling horses.

          What goes through the mind of a horse traveling at 40-60 miles an hour? Does he go 'Vrooom...' in his head or does he think 'clippity-cloppity-clippity-clop' sped up four times?

          The other day I was driving a fully loaded dump truck descending a slightly banked dirt track covered with sleet and mud. At one point it mattered not what the forward speed was, and it was slow indeed, the side-slip matched it. Turn in the direction of the skid they say, only that direction did not lead to a happy place. At that

      • If one saw a small, live unattached animal on the side of the road it is usually prudent to slow down a bit and move a bit away from the animal. This would give the prudent driver a bit more time to compensate and may have a smaller chance of startling the animal into doing something stupid.

    • 3. Can it tell if the animal is on a leash and not going to be an issue?

      3.a. Can it tell if the animal is on a leash and is still going to be an issue?

  • OMG, No! Not Ford! (Score:3)

    by Wingsy ( 761354 ) on Friday December 13, 2013 @07:29AM (#45678915)
    I won't sit in a Ford with the engine running, and in the future it seems I won't get on the highway with Ford's self driving cars on the road. I'm terrified that there's going to be some leftover Microsoft code in there somewhere (i.e., from Sync).

  • What happens when the car realises it doesn't need the driver to get around and could easily pop down to the local garage itself when it needs spare parts or petrol/electricity? Or when it gets tired of smelling faintly of sick, or having its lovely seat fabric ruined by small humans?

    And if it can tell a paper bag from a small animal from 200 ft, perhaps it can also distinguish a rubbish bin from a human so it knows WHICH ONE TO KILL?

    I'm not saying this will happen, I'm just asking questions.

  • 200ft? That's about 60m. That puts it far to close to the ideal, best condition stopping distance of a car moving at about 60km/h. The software for detection isn't new, and ladar has had this sort of range for a while. Quintiple the range and keep processing real time, then it'll be worth news

  • While the whole system is very cool and 200' sounds like a lot, remember that at highway speeds, a car is covering ~100' per second, so 2 seconds to identify, contemplate, and react to that obstacle.

    Logically, in oncoming situations (as a worst-case), two highway-speed vehicles 'detecting' at 200' have only about one second (actually less thanks to inertia, given that control-input and -effect isn't instant) to resolve, contemplate, and react.

    I have to imagine the guys working on these systems are acutely a

  • 200 feet is just a tad under the distance required [csgnetwork.com] to decelerate from 65 mph to zero on dry pavement. In other words, the system gives gives enough advance warning so you know what you're about to gently bump into after screaming to a stop in a cloud of smoke. Or crash into, if the pavement happens to be wet.

    I'd say 10 years to mass market is optimistic.

    • Yeah but if it misinterprets a squirrel or a cat vs a plastic bag you're going to get a sudden deceleration you weren't expecting. When people are in charge you can determine that if it's a small animal which allows you to swerve or just use it as a speed bump.

  • The autonomous car detects a cat in the road, and then what does it do? Does it slam on the brake even tho you're doing 65 mph and there's an 18 wheeler 3 feet from your rear bumper? Does it try to brake and swerve even tho there's a glaze of ice on the road?

    • Re: (Score:3, Interesting)

      by blackbeak ( 1227080 )
      The autonomous car deftly picks up the cat via robotic arm extension without even needing to slow down, reconfigures the route plan to stop by the nearest animal shelter, and automatically drops cat in the shelter's autonomous stray animal receptor.

  • Can it tell a small animal inside a paper bag? (Score:3, Funny)

    by 140Mandak262Jamuna ( 970587 ) on Friday December 13, 2013 @09:30AM (#45679289) Journal
    Well the Russians are way ahead of Ford. They drive around with dash cams all the while and their systems can not only tell a paper bag from a small animal, it can tell if there is a small animal inside the paper bag. Not only that, it would take that cute cat in a paper video and upload it to the click bait web site also has a drive by download malware. Sergey Gregorovich, the owner of the malware site, says, "My R&D investment in integrating small animal in paper bag detection technology with dash cam, auto upload and drive by download technologies have given me rich dividends".

    • Re: (Score:2)

      by aiadot ( 3055455 )
      Ah the good old paper bag insurance fraud. The reason why all russians have dashcams in their vehicles.

      1. Hides inside paperbag
      2. Get's hit by car
      3. ???
      4. Profit

  • I can't wait until Ford starts making hybrid fusion cars. We only have a couple of years to make Mr. Fusion happen.

  • I have a Ford Focus that features a computer-controlled manual transmission. So it's easy to drive, like an automatic, but gets somewhat better MPG, like a manual. Problem is, the computer often shifts it like somebody who doesn't know how to drive a stick. It used to stutter when backing up, but that got fixed via a software update in factory recall. (Hackers, here's a new attack vector.) Anyway, once they can get that right, then maybe they'll be ready to drive the whole car.

  • Ford Self-Driving R&D Car Tells Small Animal From Paper Bag At 200 Ft.

    Why?

  • You often find small animals hiding in paper bags looking for food.
  • is this for keeping score? 5 points for a bag, 50 points for a cat? Just wondering, since it is not an issue to run over a paper bag, and while running over an animal is not desirable, it is also safer than trying to avoid it.

  • ... Ford says the research vehicle's sensors are sensitive enough to detect the difference between a small animal and a paper bag even at maximum range...

    That is not as difficult as it may sound. And it is not a question of "sensitivity". The paper bag does not generate heat, while a small animal does.

    .
    So big deal - the car has an infrared camera on board.

  • A ridiculous press release.

    Did they mention HOW RELIABLY it can tell a paper bag from a kid on a bigwheeler on the road?

    If it is less than 99.99% correct, the first kid that gets run over will spawn a billion-dollar lawsuit.

  • Kobiashi maru (Score:3)

    by TheCarp ( 96830 ) <sjc.carpanet@net> on Friday December 13, 2013 @12:13PM (#45680793) Homepage

    I ran over a cat once; and it was the right decision at the time. Will the car be able to make that assessment?

    I know I had no choice and it still bothers me. I still see that kitten running out from the side of the onramp, diagonally across the road. I still remember that split second where I saw no where to go but off the road into a ditch, no time or space to stop....and the look of excitement on the kittens face running towards a fate he could not have expected.

    I wonder, how will a driverless car react in a no-win situation? Because I know I am not the only person to have faced one. Someone I know was on her first long distance trip out of state and suddenly found a deer in front of her. She didn't have the experience to make the snap judgement, she didn't hit the deer...instead she swerved and ended up bouncing off gaurd rails like a pinball. Telling this to my rural living cousin his response was unceremonious: "Never swerve for a deer; just hit it" (moose btw, are another story)

    Then again, maybe if it was a kid, do you go for the ditch? Is it different if its just me in the car or a carload of people? (kids in the car?). A driver can debate these things and make split second decisions to sacrifice himself; a driverless car has to leave this decision to engineers who design how it makes decisions.

    I think it makes the most sense to constrain its emergency response to what keeps the occupants the most safe in all situations; that seems most right but, its not always easy to feel good about. I don't think I would want to be the guy who wrote that code.

  • The Ford vehicle has four Velodyne HDL-32 [velodynelidar.com] LIDAR units. This is the generation after the one Google uses. They're smaller, but the field of view is wider vertically and the resolution is lower.

    They spin and get full-circle images, so for research purposes they're usually mounted on top of the vehicle. But that has to change for production vehicles. A production system wiill need more sensors better integrated into the auto body.

Slashdot Top Deals

"What man has done, man can aspire to do." -- Jerry Pournelle, about space flight

Close