Oxford University's driver-less car

[gilesjuk]

They're asking permission to try it on the roads:

http://www.telegraph.co.uk/motoring/917 ... s-car.html

A car may be able to drive around the roads, but can it handle all the pitfalls that humans are able to deal with?

[thirdcrank]

... A car may be able to drive around the roads, but can it handle all the pitfalls that humans are able to deal with?

There's a questionable assumption in there.

[kwackers]

... A car may be able to drive around the roads, but can it handle all the pitfalls that humans are able to deal with?

There's a questionable assumption in there.

I think if we'd designed robots as good as humans they'd be recalled PDQ!

[meic]

I wonder if it can tell a traffic light from a Dinotte rear light?
My GPS has been known to have some Software anomalies.

It will quite possibly be safer than many brainless drivers but as the deaths will have a different cause to those that currently happen with driven cars, will the Judges still hand out "Get out of jail free" cards to offenders?
Or will they be charged with manslaughter like a gas fitter is if a boiler explodes?

[kwackers]

The good thing about robots is that they'll always get better.
As opposed to drivers who seem to steadily get worse...

With regards the manslaughter issue - that only happens if you can demonstrate incompetence on behalf of the fitter.
It's probably no different from a garage being at fault for incompetently servicing your brakes...

[Si]

I think that even if we do get a driverless car there will be a sizeable group who look down on it and attempt to disengage the autopilot at any opportunity. For instance, suppose the autopilot (Roger?) senses a cyclist in front on a moderately narrow road with blind bends. It decided to follow at a safe distance until it is safe to overtake. The driver, on the other hand, can't understand why his device isn't making reasonable progress and so takes control back to squeeze past the cyclist and help Roger out.

[thirdcrank]

The ultimate version would have sensors of every type to ensure that decisions were taken on the best possible info. Then somebody would have to write the algorithms if that's the right word. If responsibilty was going to end up there, I suspect they would be fail-safe in every way. As soon as the concept of a driver override was accepted, then there'd be a whole range of expediency settings. Just like it is now with human drivers.

[JohnW]

What the hell's the point of a driverless car? Does this mean that motorists will be able to cause pollution, misery and death by proxy - from the comfort of their own armchairs? Will they be able to have web-cams in the damned things, so that they can watch the problems they're causing - anonymously, and without ever having to go out in the rain?

They'll be able to use their satnavs, pull their caravans and block country lanes and nobody will be able to shout at them.

What an invention. I bet someone is really proud of this.

[kwackers]

What the hell's the point of a driverless car? Does this mean that motorists will be able to cause pollution, misery and death by proxy - from the comfort of their own armchairs?

I suspect quite the opposite.
Properly done a driverless car is less likely to kill, will produce much less pollution and causes less misery by virtue of the first two points plus the fact that it's owner can do something more constructive whilst in it than fuming at the car in front.

[Cunobelin]

They're asking permission to try it on the roads:

http://www.telegraph.co.uk/motoring/917 ... s-car.html

A car may be able to drive around the roads, but can it handle all the pitfalls that humans are able to deal with?

No it cant

I can't reference this, but I remember in the 1970's that did the basic safe distance driving.

It was very successful in tests and by altering safe distance to suit speed.

Then it went on the road......

On the first test, someone saw the gap, and dived through it to a slip road.

The automatic car registered the shorter distance and slowed immediately to the very low speed that was safe

Of course this then caused panic braking in the following traffic, and a near pile-up as the others tried to compensate for the rapid change of speed

The tests were stopped as dangerous.

I know computers have advanced, and "decision making" is quicker, but muppets are still the same.

[cycleruk]

The Chinese have already sorted it:-
http://www.telegraph.co.uk/motoring/pic ... ml?image=5

[[XAP]Bob]

The Google version works well too.

[JohnW]

..................A car may be able to drive around the roads, but can it handle all the pitfalls that humans are able to deal with?

No it can't.......................

No Cunobelin, I agree, I don't think it can - or ever could - either. And the worry is how many will it kill whilst they're finding out that it can't?

The academics will put it down to 'teething troubles' no doubt - and for as long as it's not their own loved ones that it kills, why should they worry anyway?

Talk about 'mad professors'............

[snibgo]

A common maxim among computer geeks decades ago was, "Artificial intelligence is better than none at all."

I'm out of touch with AI, and don't know how far it has advanced, as opposed to a by-product from hardware being thousands of times more powerful. The problems of 1970s AI described by Cunobelin can be seen everyday on today's roads, courtesy of human intelligence.

I wouldn't be surprised if today's AI drivers are safer than human drivers.

They are still in development, of course, but they raise some far-reaching questions. Driving (compared to cyling) already divorces the traveller from the environment. This is an attraction of driving, but creates its own problems -- drivers are immune from the problems they cause.

If driverless cars become readily available so driving becomes the trivial action of sitting in a car and pressing a single button so the person is merely a passenger, then travelling everywhere by car really will be a "no-brainer", in all senses. If we think today's society is car-centric, we ain't seen nothing yet.

[Cunobelin]

A common maxim among computer geeks decades ago was, "Artificial intelligence is better than none at all."

Far too many drivers fall into the latter category!