back to article Pioneer slaps 80s LASERS on cars for driverless push

Sound system bods Pioneer have completed a trial of laser-outfitted cars in a bid to have vehicle mapping technology in commercial use as early as 2018. Retro-tech fans will be tickled to hear that the Japanese consumer-tech giant based its work on Laserdisc technology with its roots in the 1980s. Pioneer hacked its tech into …

  1. Voland's right hand Silver badge

    Signal to noise and interference

    It will be interesting to see how this fares when every car has got one.

    Ra(and respectively Li)dar works quite nicely when you have one or two working in the same area. Make that a few 100 as in your average 4 lanes traffic jam and watch the show.

    1. Anonymous Coward
      Anonymous Coward

      Re: Signal to noise and interference

      AIUI, LiDAR works using specially encoded pulses which return out of phase. The unit calculates this difference to determine its situation. Perhaps this pulse sequence can be unique to each unit so that it can distinguish its own emissions from those of other units.

      1. Naughtyhorse

        Re: Signal to noise and interference

        I'm pretty sure the time of flight is too fast for any meaningful encoding to be possible. not that it makes a lot of difference at the hardware level. I'm a little confused also that they take the view that laserdisk tech is 'useful' 'cause as far as i can see taking a couple of hundred thousand samples a second is not going to cut the mustard, by a couple of orders of magnitude.

    2. Infidellic_

      Re: Signal to noise and interference

      Question to all: Is LiDAR affected by rain? If the range is reduced what kind of impact will that have on autonomous high speed (i.e. motorway) driving? What about autonomous LiDar/RADAR based cars in (heavy) snowing conditions?

      1. Alan Brown Silver badge

        Re: Signal to noise and interference

        "Is LiDAR affected by rain?"

        Not that I've noticed (Adaptive cruise control), although the car maker has erred on the side of paranoia and hands control back to the meatsack if the wipers go continuous for more than 12 seconds.

        1. Naughtyhorse

          Re: if the wipers go continuous for more than 12 seconds.

          rather suggests it is affected then. makes sense.

          using survey lidar you get more than 1 bounce from a pulse, so for eg, a pulse from an aircraft hitting the canopy of a tree will give a return from the canopy, and the light continues sending further returns as it encounters more leaves, and finally the ground. usually you can take upto 4 returns from each pulse.

          A raindrop would surely generate a return, and even moderately light rain would trigger a few returns over the sort of distances you would want to measure in a driving system which would be okay for the laser, but really up the workload on the the processing end of things!

          1. Anonymous Coward
            Anonymous Coward

            Re: if the wipers go continuous for more than 12 seconds.

            The thing about rain is that most of it is gaps. You can easily prove this for yourself: when you look at a person walking in the rain, you are seeing mostly person and hardly any water, thus proving that most of the light from the person reaches your eyes without interruption by raindrops. It is very rare that rain is so heavy you can no longer see the person, and if the rain is that heavy cars should pull over and wait anyway.

      2. Anonymous Coward
        Anonymous Coward

        Re: Signal to noise and interference

        s LiDAR affected by rain?

        Not at the power levels that My Executive Vehicle will have: I want something that will blow the smouldering husks of window cleaners, über drivers and other "street artists" straight into the next borough with a simple voice command.

        Rick and Morty nailed it:

        https://youtu.be/2IjOXGtsvpE

      3. Jaybus

        Re: Signal to noise and interference

        With LIDAR, the light will back-scatter from multiple targets most of the time, including dust, fog, diesel particulates, and other so-called soft targets. But the vast majority of the scattered light being returned is coming from a hard target (vehicle, ground, etc.) The distance calculation (based on quadrature detection) is not simply the phase of the return signal, but a phase,magnitude vector. The return signal seen by the detector is the convolution of the signals from every scatter target. Signal components from soft targets in front of a hard target will cause its distance to appear closer and soft targets further away will cause it to appear further away, but only to a small degree, since the magnitude of a soft target signal is tiny compared to the magnitude of a hard target.

    3. Anonymous Coward
      Anonymous Coward

      @Voland Re: Signal to noise and interference

      As someone who has worked with LiDar in mapping, I can tell you its pretty cool, but won't solve the problem.

      The biggest problem is trying to determine the lanes on the road.

      You think its easy... maybe yes, maybe no.

      If the road is relatively smooth and you have painted lane markers, sure.

      If you have a road that has pot holes, or a rough road... maybe no.

      Construction? Forget it.

      The interesting thing is being able to capture GPS and Lidar over 100s of vehicles driving on the same stretch of road. now you can separate the streams of data in to lanes to get a better average line

  2. chris 17 Silver badge
    Mushroom

    How immune to a laser pointer will this type of system be?

    1. ZSn

      The same way yours eyes are dazzled by a radar gun

      1. Tom 7

        Dazzled by a radar gun?

        I'd imagine a well pointed hand held laser would overwhelm a detector - unless they are intending to cut their own roadways as they go.

        1. Jaybus

          Re: Dazzled by a radar gun?

          Assuming you could hand hold a laser pointer and actually hit a tiny detector on a moving vehicle and hold it on target for many seconds, then it might be possible. A standard laser pointer would not be nearly bright enough. The laser pointer would appear as a DC signal on the detector, whereas the real signal is modulated. Since a bandpass filter will be removing signal components not within a small band around the modulation frequency, the DC laser pointer would have to be bright enough to completely saturate the detector in order to much affect the measurement.

          1. Charles 9

            Re: Dazzled by a radar gun?

            So what about you use something like a dazzler (which is very bright and has a wider beam) tuned to match the frequency of the LiDAR laser?

  3. Pascal Monett Silver badge

    "irrational human fear from loss of control"

    Sorry, but given how everything computer-controlled has demonstrated a marked ability to go haywire with, sometimes, dreadful consequences, the fear of loss of control is decidedly not irrational.

    That said, I will welcome a car that drives itself properly and securely at all times, under any conditions, with open arms. The thought of sitting in my own private taxi, telling it where to go and then opening a book for the trip to work is one I would really, really like to see come true.

    Alas, it most likely won't happen before I retire. Oh well.

    1. Dale 3

      Re: "I will welcome"

      I would welcome a car that is driven by a human properly and securely at all times, but alas, that most likely wouldn't happen either before I retire.

      1. Charles 9

        Re: "I will welcome"

        Probably because the human simply cannot fulfill the "at all times" provision, meaning another solution is needed.

        PS. Perhaps Pioneer can demonstrate a little more confidence in its technology by showing field tests in less-than-ideal conditions such as heavy rain (which can refract the LiDAR) or whiteouts (which can cover up necessary details).

    2. Graham Marsden
      Alert

      @Pascal Monett - Re: "irrational human fear from loss of control"

      > given how everything computer-controlled has demonstrated a marked ability to go haywire with, sometimes, dreadful consequences, the fear of loss of control is decidedly not irrational.

      Given that people can't even manage to design computer systems to send e-mails safely and securely, I wholeheartedly agree!

      1. Denarius
        Terminator

        Re: @Pascal Monett - "irrational human fear from loss of control"

        Aircraft computer control fear ? Not for 50 years. It has gone too far in fact. The computer is blindly trusted despite it being obvious that the sensors are not reporting correctly. See AirFrance splat in MidAtlantic. Commercial airlines have begun to quietly check their long serving pilots can still actually fly well without computers. As for computer auto-drivers, bring it on. A 7x 9s reliable machine would beat the double 9s or less driver I have to avoid or endure daily. Complete inability to observe 100 meters ahead or anticipate is common. I suspect the main problem is who to sue when things get bent being the elephant in the room.

        1. MaxHertz

          Re: Air France Flight 447 accident

          Erm... the Air France crash was due to the pilot not recognising that one of the air speed indicators was malfunctioning. If a computer had been in control it probably would not have crashed.

          1. BlinkenLights

            Re: Air France Flight 447 accident

            Erm... The computer was in control until it got conflicting readings from the sensors and gave up control. At that point only the pilots could figure out what was going on, and they got it wrong, on this occasion.

            1. Alan Brown Silver badge

              Re: Air France Flight 447 accident

              "The computer was in control until it got conflicting readings from the sensors and gave up control."

              What amazes me is that Airbus pitots don't have automatic sensors to turn on the heaters under icing conditions.

              Or that it didn't use inertial/GPS nav input as well as the mechanical airspeed inputs (there are a number of ways of estimating speed sans pitots and pilots are taught them. This should be part of the flight control program too)

              The biggest bug is that programmers didn't anticipate something like this and make allowances for it. There are a number of cases of crashes caused by programmers working on the basis that "a pilot would never do this in real world conditions", only to find out that pilots DO. The same issue will happen with cars, but you have to factor in the wildcard of erratically controlled vehicles/pedestrians in the immediate vicinity. I do tend to wonder when the first instance of road rage against an automated vehicle will occur.

              1. Charles 9

                Re: Air France Flight 447 accident

                "What amazes me is that Airbus pitots don't have automatic sensors to turn on the heaters under icing conditions."

                Now you got a sensor ON a sensor. And that sensor can still fail.

                "Or that it didn't use inertial/GPS nav input as well as the mechanical airspeed inputs (there are a number of ways of estimating speed sans pitots and pilots are taught them. This should be part of the flight control program too)"

                Which doesn't help much if you're getting conflicting readings, meaning one sensor says one thing while another says a different thing. At this point, it can't figure out which one's the right one and having additional sensors isn't likely to help since you can always have a failure cascade that makes multiple sensor say the same WRONG thing at the same time.

        2. Alan Brown Silver badge

          Re: @Pascal Monett - "irrational human fear from loss of control"

          "Complete inability to observe 100 meters ahead or anticipate is common"

          "3 second drivers" are extremely common - that's 3 seconds on the throttle, 3 seconds off - which means they're not controlling their speed properly or anticipating more than 2-3 seconds ahead, which in turn means a lot more braking than should be necessary.

          There's a 20-30% fuel consumption penalty for this kind of driving so improving that pays dividends.

          Almost all drivers believe they're "better than average". This kind of self-delusion is why there will be a lot of resistance to automated vehicles, but insurance companies will overrule them. Do you think £2000/year premiums on small cheap cars will stay restricted to the under-25s?

          What gets even more interesting is when automated vehicles start tracking and reporting badly driven meatsack vehicles. If you think that data won't end up in the hands of the insurance industry you're naive. Everyone cocks up - regularly - and if every event starts being reported you can expect to find things getting costly.

    3. Filippo Silver badge

      Re: "irrational human fear from loss of control"

      It doesn't have to. It only has to drive itself properly and securely *more than most human drivers*. That's very definitely achievable.

      If you accept some non-zero error rate from a human driver, but won't accept a lower error rate from a computer because it's still non-zero, that's irrational.

      1. DropBear
        Devil

        Re: "irrational human fear from loss of control"

        "If you accept some non-zero error rate from a human driver, but won't accept a lower error rate from a computer because it's still non-zero, that's irrational."

        I'm sure you'll have no problem with me testing my almost-entirely safe automated machine gun turret at the same shooting range you go to, when you're there in person. I promise it isn't any less safe than most humans can be expected to be handling a gun...

        1. macjules

          Re: "irrational human fear from loss of control"

          You mean like in South Africa ... http://www.wired.com/2007/10/robot-cannon-ki/?

      2. Ben Tasker

        Re: "irrational human fear from loss of control"

        It only has to drive itself properly and securely *more than most human drivers*

        wrong

        For me to get in one, or be comfortable having my family in one, it has to drive itself properly and more securely than me.

        I'm not claiming to be the jesus of driving here either, my point is I have a high mileage (min 600 miles/week) with a very low (0) accident rate. That, undoubtedly, is in part due to good fortune, but it means I set the bar pretty high for how good an autonomous car needs to be before I'd feel comfortable giving up control.

        So, no, it doesn't just need to be better than the average meatsack, it needs to be better than each individual before they can sanely make the choice to trust an autonomous car with their lives (feels like an exaggeration, yet its not....)

        1. Alan Brown Silver badge

          Re: "irrational human fear from loss of control"

          "For me to get in one, or be comfortable having my family in one, it has to drive itself properly and more securely than me."

          Your insurance company will beg to differ.

          Seriously. The driving force behind vehicle automation will be insurance premiums.

          As for your misplaced confidence in your abilities: Not being involved in any crashes does not make you a "good" driver. The rules of the road mean that someone has to be pretty spectacularly _bad_ to get into a crash and there are plenty of bad drivers who cause other people to crash but get away scot-free.

          1. Ben Tasker

            Re: "irrational human fear from loss of control"

            As for your misplaced confidence in your abilities: Not being involved in any crashes does not make you a "good" driver.

            Leaving aside the fact you have no idea whether my confidence is misplaced or not (especially given I gave you a single datapoint, you know nothing of experience/training etc), I totally agree. However, being computer controlled also does not mean that autonomous cars will be better in every situation a driver might encounter - especially anything which may be unique.

            It's a fairly simple risk assessment. My experience and ability is (more or less) a known quantity. The experience and ability of the bloke programming my robocar is based entirely on faith. I'm not comfortable gambling with my family's lives based on faith.

            Over time, the balance will likely tip, as robocars will gain experience collectively.

            As for insurance, I'd hate to go to 2k a year premiums, but you know what - if I believed my family were safer with me driving I'd pay it. That doesn't mean I/we would never make the odd journey in one, but that's a big difference to scrapping the car and making every journey in one.

            A cynic might also suggest that excess financial pressure from the insurance companies would probably just lead to more uninsured drivers on the road, given that not everyone will be willing/able to change car.

            It's an interesting technology, but it's got quite a way to go, including real-life testing on a wider range of inhospitable roads.

    4. Commswonk

      Re: "irrational human fear from loss of control"

      I am struggling to work out exactly why it is in any way "irrational".

      "However, the industry, much like the aviation sector which would also benefit from computer-controlled pilots..."

      At the moment a lot of flight is indeed computer controlled, but at the same time the pilot is unambiguously responsible for the safety of the aircraft and its contents and is able to override the automated systems if circumstances demand it. Unlike land - based vehicle drivers pilots generally have a "bubble" around them which gives them a bit of thinking and manoevering space, a luxury not always available to cars and their drivers.

      "...sensor can detect lane line markings..." yeah right, if the markings are actually there in the first place; quite a lot near Chez Commswonk are barely visible due to normal wear and tear. Oh and one or two roundabouts have mysterious markings - understandable to the human eye and brain - that _cross_ the line of travel to move vehicles into the most appropriate place for where they are going. Well that's the intention anyway, but I'm far from certain that any computer will be able to make sense of it.

      Oh and lane markings often vanish after a light fall of snow or even after rain under overhead lights.

      Another solution in search of a problem.

    5. Alan Brown Silver badge

      Re: "irrational human fear from loss of control"

      "given how everything computer-controlled has demonstrated a marked ability to go haywire"

      Usually in less spectacular ways than human controlled systems.

      Computers don't have to drive cars perfectly, just better than meatsacks. Fortunately for computers, this isn't very hard.

  4. Joey M0usepad Silver badge

    frickin Lasers?

    missed a trick there , the reg usually so good on the quirky article titles too

  5. quattroprorocked

    It's not an irrational fear of loss of control. It's a rational fear of handing control to large organisations with poor records re security and fault acceptance.

    With aircraft, the pilot has skin in the game. If his pilot error kills you, he kills himself.

    If Google's error kills you, it's just you what's dead.

    1. Charles 9

      "It's not an irrational fear of loss of control. It's a rational fear of handing control to large organisations with poor records re security and fault acceptance."

      Then how come we don't live in constant fear paralysis given we HAVE to hand control to SOME large organization in order to pretty much function as a modern society? If not private enterprise, then the State. Kinda hard to avoid, you know?

  6. Sgt_Oddball
    Flame

    What I'd like to know....

    Is how susceptible to dirt, sap and all the other crap that builds up on a car?

    I mean my own jalopy isn't the dirtiest thing on 4 wheels but it's not what you'd call showroom condition.. And what about stone chips to lenses/lens covers?

    Fire because I'd want it to be quick....

    1. Permidion

      Re: What I'd like to know....

      future cars will certainly rely on several technologies and not *only* lidars, so I dont think we have to fear any technology limitations.

      beside, if you take current ABS and ESP, there is already sensors reading allowing to "know" (calculate) when the car is on the right trajectory or not and afaik nobody is having second thought about these systems.

      when future systems will be implemented in production series, I guess they will have at least as much testing as already implemented systems.

  7. Alistair
    Windows

    Dunno bout this one

    From the diagram that lidar has a blind spot almost worse than the average meatsacks.

  8. Kepler
    Terminator

    "[T]he irrational human fear [of] loss of control"???

    "However, the industry, much like the aviation sector which would also benefit from computer-controlled pilots, is being forced to move slowly due to regulations and the irrational human fear [of] loss of control."

    Since when, and why, is the desire to be able to control our machines irrational?

    (I see now that others have already argued some of the details of the matter above. I merely ask the question, and note that Mr. Pauli's glib assertion was tossed out casually and without any support, or even attempt to support.

    Mr. Pauli and his writing are new to me — not sure how new he is to The Register (I was away for about 6 months; first noticed his name in a byline two or three weeks ago) — but this is at least the second article by him I've read in which something whose truth was far from obvious was asserted without any support or explanation at all. Hope these two instances are anomalies.)

  9. Alan Johnson

    Regulations and irrational fear is not the problem

    "However, the industry, much like the aviation sector which would also benefit from computer-controlled pilots, is being forced to move slowly due to regulations and the irrational human fear from loss of control. "

    Nonsense.

    Automatic pilots for routine tasks have been used in the aviation industry for a very long time but it is recognised that they have limits and human pilots are much better at handling exceptional circumstances, the comination is better than either alone so both are used. Generally the aviation task is easier because it is a much simpler less cluttered and more controlled environment, even so human pilots are needed for situations that automatic pilots cannot handle.

    The same will be true for cars except there are probably many more exceptional conditions. At least one human has already died as a reuslt of an inappropriate level of trust in an automatic driving system. A healthy appreciation of the limitations and failings of human and automatic systems is a good thing. Regulations or irrational fear is not the limiting factor at the moment so much as immature technology. Some regulation is absolutely necessary and appropriate.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like