driverless cars

[PepperElf]

http://www.detroitnews.com/article/2...less-cars-bill

I dunno I don't trust this thing.

Especially since they're trying to suggest it will be good for say, people who are blind or are drunk... but then they admit you need a licensed driver to take over in case of emergencies.

[telecom_goddess]

They are suggesting that those who are blind use these. I don't think I trust them either just because of the stupid humans that would still be out there.

[Chromatix]

At the moment, the technology isn't entirely proven. That's why the competent driver needs to be there, in case a situation arises that the car responds dangerously to.

In the future, when it's been *proved* to be orders of magnitude safer than human drivers, it'll be available for the blind and incompetent.

[PepperElf]

At the moment, the technology isn't entirely proven. That's why the competent driver needs to be there, in case a situation arises that the car responds dangerously to.

In the future, when it's been *proved* to be orders of magnitude safer than human drivers, it'll be available for the blind and incompetent.

when? I'm not sure I'd completely trust a computer in a case like this.


like when I was driving in ny one day. my car wouldn't stop, even with the brakes applied, even with antilock brakes and low gear. what would the computer have done that I didn't do?

me, i crashed my car on purpose into a snow bank. no damage afaik but still, would a computer have decided to crash on purpose?

[LillFilly]

Nah, they won't REALLY be driverless; there will be a short yoda-like blue haired granny behind the wheel. Not new technology at all!

[Ironclad Alibi]

Will those driverless cars read the "permit required" signs in parking lots?

Will the "permit required" signs have a QR code so the driverless cars can read them?

Will Argabarga's company buy a driverless tow truck to tow them away?

Will driveless cars complain that they were not programmed to read the "permit required" signs?

Inquiring minds want to know.

[Chromatix]

when? I'm not sure I'd completely trust a computer in a case like this.


like when I was driving in ny one day. my car wouldn't stop, even with the brakes applied, even with antilock brakes and low gear. what would the computer have done that I didn't do?

me, i crashed my car on purpose into a snow bank. no damage afaik but still, would a computer have decided to crash on purpose?

Chances are, the car would have detected or anticipated the low-adhesion hazard, and either refused to drive onto that street, or proceeded at a much lower speed to permit a reasonable stopping distance. The computer can make rational and informed decisions about such things much more easily than a typical human.

Mind you, over here you would have been driving on studded winter tyres. Those are much better at gripping on ice than plain rubber is.

[Racket_Man]

My probelm with computer controlled cars would be during bad weather or serious road conditions like a blinding snow storm, ice storm, icy roads, flooded roads, and alternate routes around such obstecles.

Let's see if a computer could do my job in the conditions I drive in.

[Chromatix]

In a snowstorm, chances are the radar can see better than you can. In any case, the computer has more patience than you, and is therefore safer when the correct driving speed is slow.

As for alternate routes, once it's decided to take one, I should point out that human drivers already tend to rely on a computer to tell them where to go.

[PepperElf]

In a snowstorm, chances are the radar can see better than you can. In any case, the computer has more patience than you, and is therefore safer when the correct driving speed is slow.

As for alternate routes, once it's decided to take one, I should point out that human drivers already tend to rely on a computer to tell them where to go.

oops I forgot to mention my speed... 2mph. Kinda hard to be more patient than that.

and that's a good point about the parking. Maybe they might be programmed ahead of time but traffic is an ever-changing situation.


sure it works for google but they also have a situation where it's controlled-access. driving around their complex =/= driving on a highway

[mikoyan29]

I suspect that at first it's going to be in a situation where little driver interference is required like driving on a freeway and they will start to add more complicated situations to their algorithims. Kind of like auto-pilot in planes.

[Chromatix]

As it happens, Google's cars have been on the public road, in both freeway and urban situations, for quite a while now. They've had *one* accident, a minor rear-end collision, which was incurred while under manual control.

The system is already much more sophisticated than some people seem to realise.

[wolfie]

This thread reminds me of a site I browsed a while back.

[ADeMartino]

NO. Just........ NO.

[wolfie]

The big issue with driverless cars is that it changes the probability/severity mix for accidents. With driven cars, you get a lot of "fender benders", and a few serious crashes, but they're INDEPENDENT. Driverless cars (especially if they have to cope with the randomness of idiot human drivers by running on the same roads), if programmed properly, can avoid the "fender benders", and will have very few "normal" serious crashes, but if something goes wrong it will go wrong BIG TIME (i.e. the same problem will affect large numbers of cars at once). The liability issue (imagine 50-car pileups happening in a few dozen places at once) means no insurance company will want to touch it - even if you reduce the probability to near-zero, the potential cost of one incident raises the "expected loss" to unacceptable levels, especially since you're getting so far out into the "tail" that actuaries have no experience with it.

It's roughly similar to an issue with light aircraft. I'd estimate that over 50% of piston singles use carburetors and magneto ignition, a system that's more likely to run into problems than the electronic fuel injection and electronic ignition used on modern cars. On the other hand, most of the failures with carb/magneto fall into one of 2 categories:
1). !@#$%^& thing won't start. This is not a safety issue, since it happens when the plane is on the ground and not moving.
2). Problem develops gradually, giving you time from when you first notice the rough running until it becomes unusable, so you're able to pick out a place to land.

EFI/EI has eliminated virtually all of these cases, at the expense of having "running engine quits without warning" being the dominant case. Not the sort of thing you want to have happen when you're airborne.