Please turn on javascript in your browser to play chess.
Science Forum

Science Forum

  1. Subscriber Kewpie
    since 1-Feb-07
    06 Oct '12 08:52
    from The Future Of Self-Driving Cars, by Colin Lecher in Australian Popular Science Magazine.

    Car designer Chris Bangle has spent years designing forward-thinking vehicles, so now, with self-driving cars just legalised in California, we decided to pick his brain on what's next for the automobile that no longer needs its master.
    Today's self-driving cars mostly look like the human-commanded variety; they just act automatically. But as they become more viable for mass consumption - by 2020, Bangle believes - they'll need to be streamlined. Here are some thoughts assembled from Chris Bangle Associates S.R.L. about what the car - if there is a "car" - might look like when it's ubiquitous, in 2050. There are three main categories that a discussion about self-driving cars can fit into, Bangle says.

    YOU INSIDE OF CARS
    It's the future. All that stuff in your car's cabin? You'd think we wouldn't need it, freeing us from the tyranny of the steering wheel forever. Theoretically, Bangle says, seating position could change several ways, making it closer to the way a bus is designed, for instance. But that doesn't necessarily mean everything we associate with driving will be done away with. As an example: "Do you have to wear seatbelts in a taxi? Yes. Do you drive the taxi? No."
    There's also the matter of how far into the future we're talking about. A still-drivable self-driving car - in other words, one that functions both ways - might act as a stopgap toward a full-fledged driverless vehicle, but in the meantime, designers won't be taking anything out that's necessary for traditional driving.

    CARS RELATIVE TO CARS
    Self-driving cars (fully self-driving ones, since we've made that distinction) might not rely as heavily on "optical signals," Bangle says. Headlights, brake lights, other forms of automatic communication - those were created for humans, and if the humans deem them unnecessary, we might go without.
    Bangle also points to the so-called "invisible Mercedes," which uses LED lights to blend into its environment. If we're not there to cause the accidents between cars, and if the cars can handle the driving on their own, why not put the cars out of our vision entirely?

    CARS AND THE ENVIRONMENT
    Some of this category ties into our already-changing attitudes about cars. Younger demographics are less interested in cars, Bangle points out. If we make those cars self-driving, they might be seen by consumers as the same as an automated taxi, even if they own the deeds to those taxis. If that happens, we might want something else out of our cars: we don't care what a taxi looks like, after all, so maybe a point-a-to-point-b machine would be enough. Ownership in general gets "put into question" with self-driving cars, he says: what does a parking space mean when the car parks itself?
    As Bangle sees it, that mindset might also have effects in advertising. Maybe car owners could earn money for featuring a company's logo. When we're not in control of cars, our sense of ownership could dwindle, at least psychologically. If that happens, we might be more open to using them for advertisements.

    Those shifting attitudes don't make Bangle particularly upset. The field could use "some new references," he says. "I think self-driving cars can do that. They'll make us think again."
  2. 06 Oct '12 12:55 / 5 edits
    Most car accidents are caused by human error. So if we can design an AI to be more reliable than the human brain, we could design one to drive cars and massively reduce the incidents of car accidents and that includes fatal car accidents. Wouldn't that be great!

    In fact, if and when AI gets proven to be so very reliable that it becomes even more reliable than the most competent responsible human driver, I predict it would become law that all cars be driven by AI and none from humans simply for safety reasons. Then the same will become true for trains, aircraft, ships, trucks, cranes, tractors, harvesters industrial saws and all heavy machinery -all will be driven by AIs by law.

    AIs can be given advantages over humans such as lightning near-speed-of-light reactions and having massive amount of relevant knowledge of terrain and navigational skills etc that would all help it become a safer driver than any human.

    My only concern here is the designers of such AIs might make the mistake of making the AIs totally dependent on GPS for navigation so that there is no backup if GPS one day fails for some reason. I hope the governments would have the good sense to make it law that, without exception, all such AIs must have reliable navigation alternatives to using GPS else I predict a disaster.
  3. 06 Oct '12 13:39
    Self driving cars could also be significantly more efficient, especially in conjunction with other self driving cars.
  4. Standard member sonhouse
    Fast and Curious
    06 Oct '12 16:26
    Originally posted by twhitehead
    Self driving cars could also be significantly more efficient, especially in conjunction with other self driving cars.
    So you have 10 million self drive cars on the road and a big EMP pulse from somewhere hits, then what? How many accidents would that cause in 10 seconds?
  5. 06 Oct '12 19:04 / 1 edit
    Originally posted by sonhouse
    So you have 10 million self drive cars on the road and a big EMP pulse from somewhere hits, then what? How many accidents would that cause in 10 seconds?
    No problem; providing the governments are not so stupid as to fail to insist that all AIs that drive cars, aircraft etc must all by law be given complete electromagnetic shielding which would protect from EMP:

    http://en.wikipedia.org/wiki/Electromagnetic_shielding

    this should be cost effective and give adequate protection.
    -this is another thing that I would say should become law in addition to having reliable alternative navigation for all such AIs to GPS navigation in case GPS completely fails.
  6. 06 Oct '12 19:48
    Originally posted by sonhouse
    So you have 10 million self drive cars on the road and a big EMP pulse from somewhere hits, then what? How many accidents would that cause in 10 seconds?
    Not very different from the case where they are not self drive cars. Or maybe a whole lot safer. I would expect a self drive car to automatically stop upon loss of power.
  7. 06 Oct '12 19:50
    The first place self drive cars will be mass used, is for short commuter trips around city centers (as rental cars).
  8. Standard member sonhouse
    Fast and Curious
    06 Oct '12 23:59
    Originally posted by humy
    No problem; providing the governments are not so stupid as to fail to insist that all AIs that drive cars, aircraft etc must all by law be given complete electromagnetic shielding which would protect from EMP:

    http://en.wikipedia.org/wiki/Electromagnetic_shielding

    this should be cost effective and give adequate protection.
    -this is another thing that I ...[text shortened]... reliable alternative navigation for all such AIs to GPS navigation in case GPS completely fails.
    Actually, EMP shielding was tried in a specially built aircraft flown a few hundred miles from an actual H bomb test. It failed. It was a skin within a skin thing and just did not work, the emp pulse came through tiny cracks in the surface, like places where rivets were, solder joints and such. Any effective EMP shielding would cost more than the rest of the car and still wouldn't work against a strong pulse. Besides, who would buy a car with two bodies built in and 8 inches thick?
  9. Subscriber Kewpie
    since 1-Feb-07
    07 Oct '12 00:30
    "Google says human error was to blame for a crash involving its automated, driverless car.
    The Internet giant says its modified Prius was being manually driven when it caused a five-car smash-up near Google’s headquarters in Mountain View, Calif., last week.
    Auto website Jalopnik.com first posted photos of the robotic car apparently crashing into the rear bumper of another Prius, sparking a chain reaction of fender benders. It was the first reported incident involving Google’s driverless car, which has racked up more than 250,000 km.
    Jalopnik worried the crash was caused by a glitch in the self-driving car, which uses a roof-mounted camera, radar sensors and lasers to drive without human input. But Google says the car was not in auto pilot mode at the time of the collision.
    “Safety is our top priority,” the company told Business Insider. “One of our goals is to prevent fender benders like this one, which occurred while a person was manually driving the car.”
    Earlier this year, Nevada became the first state in the U.S. to approve the use of driverless cars on the state’s freeways." ~http://www.thestar.com/wheels 29-Aug-2011

    I'm hoping that driver-assisted cars will arrive before I'm too old to keep my driving licence, maybe another 10 years. I won't be around to see the totally automated no-driver version but I can foresee a lot of teething problems. Even today's cars become pretty deranged when their small computer systems fail ...
  10. 07 Oct '12 08:10 / 14 edits
    Originally posted by sonhouse
    Actually, EMP shielding was tried in a specially built aircraft flown a few hundred miles from an actual H bomb test. It failed. It was a skin within a skin thing and just did not work, the emp pulse came through tiny cracks in the surface, like places where rivets were, solder joints and such. Any effective EMP shielding would cost more than the rest of th ainst a strong pulse. Besides, who would buy a car with two bodies built in and 8 inches thick?

    EMP shielding was tried in a specially built aircraft flown a few hundred miles from an actual H bomb test.


    no point in shielding against the powerful pulses from H-bombs -only natural sources. If the H-bombs started falling, we would be all screwed anyway so what would be the point?

    Do you see the lack of EMP shielding against the powerful pulses from H-bombs for conventional manually driven cars being somehow less of a problem than for AI driven cars? If so, why?


    the emp pulse came through tiny cracks in the surface,


    why cannot EMP shielding be designed without cracks?

    Any effective EMP shielding would cost more than the rest of the car


    nonsense! For a starters we are not talking about EMP for the whole car but rather Less expensively just the AI brain part. An effective EMP shield against credible natural sources such as solar storms and lightning would consist of little more than a very thin layer of relatively inexpensive metal surrounding the AI brain part -how on earth can that be more expensive than the whole car!? The whole car would probably consist of vastly more metal than that.

    and still wouldn't work against a strong pulse.


    -like from a H-bomb? if so, there would be no point anyway even if it worked just fine.

    Besides, who would buy a car with two bodies built in and 8 inches thick?


    again, we are not talking about EMP shielding for the whole car but rather just the AI brain part. -Do you see the lack of EMP shielding for conventional manually driven cars being somehow less of a problem than for AI driven cars excluding the AI brain part? If so, why? if not, why bother shielding the whole car rather than Less expensively just the AI-brain part?

    Also, to the best of my knowledge, effective EMP shielding would be typically from paper-thin to up to, say, ~3mm in thickness depending on requirements and certainly NOT “8 inches thick”! where did you get “8 inches thick" from?

    I also predict that conventional electronics will be eventually replaced with spintronics for all computers including AIs.
    I would imagine that spintronic circuitry would generally be much less prone to interference from EMP because the signals in spintronic circuitry are not current-based. Not sure but I think they might not even need any EMP shielding!
  11. 07 Oct '12 09:14
    Originally posted by sonhouse
    Actually, EMP shielding was tried in a specially built aircraft flown a few hundred miles from an actual H bomb test. It failed. It was a skin within a skin thing and just did not work, the emp pulse came through tiny cracks in the surface, like places where rivets were, solder joints and such. Any effective EMP shielding would cost more than the rest of th ...[text shortened]... ainst a strong pulse. Besides, who would buy a car with two bodies built in and 8 inches thick?
    -I forgot to also point out that a typical modern conventional manually driven car already has a lot of EMP-sensitive electronics in it.
    Why would an AI driven car necessarily be much more venerable to EMP than such conventional cars?
  12. 07 Oct '12 09:36
    http://what-if.xkcd.com/5/
  13. Subscriber Ponderable
    chemist
    07 Oct '12 18:06
    Can we agree that the EMP discssion is not to the point?

    The question most probably reaching us first is the judical one:

    If a computer steered vehicle is causing an accident who is responsible? The driver or the company creating the car, or the company creating the software?

    As long as the driver can overrule the system SHe shouold be responsible. But if not? Most saving scenarios need more computernetwork steered cars than human steered ones, since humans are much more unpredictable than computers and tend to ignore rules...

    I see fully autonomous vehicles first where it is too dangerous for people to be.
  14. 07 Oct '12 19:44
    Originally posted by Ponderable
    If a computer steered vehicle is causing an accident who is responsible? The driver or the company creating the car, or the company creating the software?
    If a car is in an accident because of a brake malfunction, who is to blame? Surely it would be essentially the same scenario?
  15. 07 Oct '12 21:00 / 7 edits
    Originally posted by Ponderable
    Can we agree that the EMP discssion is not to the point?

    The question most probably reaching us first is the judical one:

    If a computer steered vehicle is causing an accident who is responsible? The driver or the company creating the car, or the company creating the software?

    As long as the driver can overrule the system SHe shouold be responsibl ...[text shortened]... re rules...

    I see fully autonomous vehicles first where it is too dangerous for people to be.

    If a computer steered vehicle is causing an accident who is responsible? The driver or the company creating the car, or the company creating the software?


    The practical answer to this is very obvious to me: change the law so that all car insurance has special prevision for when an accident is nobodies fault so that you can still claim on your insurance if it is nobodies fault.
    I hope the law would not be so stupidly designed as to never recognise when NOBODY is to blame!
    And only blame somebody or some company where and when there is actual specific reason to believe that a particular person really is at fault and not otherwise -I get the very bad impression that this is not how the law works at present.
    I just cannot stand this blame culture we have where we just have to always have an escape goat -sometimes bad things just happen and its nobodies fault! Its called life.