Why Would Driverless Cars Need Rules For Crashing?

[ToraToraTora]

http://www.bbc.co.uk/news/technology-37418119
we are continually being told they are perfect.

[Gromit]

Naomi24,
I said several pages ago, they are not compulsary.

[Gromit]

Zacs,

Naomi believed the warning sensor warranted a second look. We evacuate a building on hearing a fire alarm even though we cannot see or smell a fire. We are conditioned to reacted to warnings and act upon them, even if they over-ride our own senses. The comparison of a dumb sensor with a smart computer was a bad one though, as pointed out earlier.

[EDDIE51]

Vulcan, I have already said that there will be multiple computers as there are on something like the fire control system on a warship. One computer can fail but cars will be controlled by several computers a majority will need to agree on the driving actions. More likely there will be a huge central computer with links to each vehicle, that computer will have hundreds or even thousands of individual 'computers' which will operate on a 'consensus' basis with those that give wildly different readings being 'ruled out'.

[Zacs-Master]

Gromit 'even if they over-ride our own senses.' my point entirely. Naomi was certain that there wasn't an obstruction yet she still heeded the advice of an automated system yet she's against automated cars. A strange paradox.

[ToraToraTora]

so we'll be having cars driven by a committee eddie! what could go wrong!?

[naomi24]

Gromit, /The comparison of a dumb sensor with a smart computer was a bad one//

I’m not so sure it was. How will a computer distinguish between, say, a piece of paper blowing in the wind on a motorway and a genuine hazard? Cars screeching to a halt because a piece of litter blows across their path doesn’t seem terribly safe to me.

[Gromit]

Zacs,
Not sure it is a paradox. She is just being conservative about autonomous cars. I am sure that when they come, and they are not crashing, then she might change her mind.

[Zacs-Master]

I think Eddie is hareing off down a false track. Each car will be autonomous. There may be a central computer owned by the manufacturer which harvests data but it won't control the car. The danger of such a link going down would be calamitous.

[Quizproquo]

If the cars screech to a halt once all cars are automated, it won't be a problem.
The cars behind will be far enough back and in good enough time to stop as well.
As for your dandelion, so what if it caused you to get out an look. Whether it was there when you started moving or not is irrelevant. Kids, pets and things like stray footballs can all appear in a heartbeat.
Maybe you should do the weeding instead of arguing here ;)

[naomi24]

^Oo-er!

[Gromit]

Naomi,

As explained earlier, the car has 360degree video which is analysed by the computer. Just like computers can have face recognition and recognise places, they can also tell that a dandelion or piece of paper is not a hazard or a child.

[naomi24]

Gromit, I don't believe it would recognise something flying past specifically as a harmless piece of paper. It would see a hazard. Sorry, Gromit. As I said, you buy one if you want one. Not for me.

[Zacs-Master]

Gromit, it's a bit like explaining to the early rail travellers why they wouldn't suffocate over 32mph ain't it. The suggestion that cars would be allowed on the road which can't distinguish between a hazard and a piece of paper demonstrate an over simplistic view of the sophisticated systemS which will be used.

http://www.alphr.com/cars/7038/how-do-googles-self-driving-cars-work

[Gromit]

naomi,

I would have to see demonstrated that the systems are as reliable as they say they are. I get on aeroplanes because I DO believe the system are as good as they say they are, that it can tell the difference between a cloud and another aeroplane. I will get in an autonomous car when I believe it won't be fooled by a bit of paper.

[naomi24]

Gromit, me too.

[Zacs-Master]

What of you're behind one on the motorway which slams on the brakes?

[jim360]

In the end attitudes like Naomi's will only be changed when the cars become commonplace and demonstrate empirically that all her fears are mistaken. That's not to say that the concerns shouldn't be there in the first place -- this technology obviously has the potential to go wrong, and we shouldn't just adopt without testing. But equally we shouldn't reject the idea out of hand. Every criticism that has been rasied by eg TTT is either directly addressable, or can be just as easily applied to human drivers. For example, humans are normally good at telling a flower from a small child but our vision isn't always perfect either, and people can misread what's going on just as often, and make the same mistakes.

It is just irrational to assume that because computers cannot be perfect they must also be incapable of being better. Whenever this objection has been raised in the past, it's always been utterly wrong. So it will be with driverless cars.

[jno]

ZM, if you're behind a car (any sort) that slams on its brakes, you slam on yours too.

If you hit it, you were following too close. I imagine automated cars won't be able to follow too close (just a guess, though).

[Gromit]

Following distances are variable, depending on conditions. Most human drivers have a set distance which is probably wrong most of the time. Autonomous cars will be programmable, so it is fair to assume the distance for stopping will be generous, not the minimum.

[jno]

that sounds like a case where autonomous cars will be rather safer than humans, who frequently follow too close (I'm married to one of these, and it can be scary).