Problem with driverless car.

Mike Sales
Posts: 4202
Joined: 7 Mar 2009, 3:31pm

Problem with driverless car.

Postby Mike Sales » 19 Dec 2015, 11:34am

Interesting problem with Google car. Seems it is not pushy enough to share the road with human drivers. Is it being too careful?

http://road.cc/content/news/173523-google-working-make-its-driverless-cars-%E2%80%98more-aggressive%E2%80%99

irc
Posts: 4654
Joined: 3 Dec 2008, 2:22pm
Location: glasgow

Re: Problem with driverless car.

Postby irc » 19 Dec 2015, 1:02pm

Since it isn' as good at seeing and assessing the road and hazards it has to err on the side of caution. At the moment Google cars need a crew of two to be "driverless."

http://www.livescience.com/50841-future ... -cars.html

I think we are many years away from true driverless cars where non drivers can be taken to a destination without any human input.

User avatar
[XAP]Bob
Posts: 17151
Joined: 26 Sep 2008, 4:12pm

Re: Problem with driverless car.

Postby [XAP]Bob » 19 Dec 2015, 3:38pm

They require one driver - for legal, not technical, reasons during the current testing phase.

The second person is actually a software engineer.

They do the right thing when unsure - they slow/stop

“These vehicles are either stopping in a situation or slowing down when a human driver might not,” he said. “They’re a little faster to react, taking drivers behind them off guard"
A shortcut has to be a challenge, otherwise it would just be the way. No situation is so dire that panic cannot make it worse.
There are two kinds of people in this world: those can extrapolate from incomplete data.

Pete Owens
Posts: 1641
Joined: 7 Jul 2008, 12:52am

Re: Problem with driverless car.

Postby Pete Owens » 19 Dec 2015, 4:31pm

Mike Sales wrote: Is it being too careful?

No, the problem is the drivers hitting them are failing to take care. Note that the cars were not at fault in any of the reported crashes.

kwackers
Posts: 14054
Joined: 4 Jun 2008, 9:29pm
Location: Warrington

Re: Problem with driverless car.

Postby kwackers » 19 Dec 2015, 5:05pm

Pete Owens wrote:
Mike Sales wrote: Is it being too careful?

No, the problem is the drivers hitting them are failing to take care. Note that the cars were not at fault in any of the reported crashes.

I've lost count of the number of times I've heard folk complain that older drivers "get in the way and cause accidents'.
(Or as I like to think: "I'm an impatient git who's a danger to everyone")

Same thing I suspect. Perhaps hitting a driverless car should be an automatic disqualification... ;)

Bmblbzzz
Posts: 3034
Joined: 18 May 2012, 7:56pm
Location: From here to there.

Re: Problem with driverless car.

Postby Bmblbzzz » 19 Dec 2015, 5:53pm

Oh dear. Google cars are programmed to drive correctly, which means within the limits of their own sensory and decision making capacities. Those sensory capacities are superior to humans, because they are not restricted to roughly 180 degrees of vision but use a full 360 of vision and other senses. As for its decision making, well, it's a computer. The trouble is that no human drives (or rides) within their own capacities all the time. So we're "dumbing up" the Google cars.

irc
Posts: 4654
Joined: 3 Dec 2008, 2:22pm
Location: glasgow

Re: Problem with driverless car.

Postby irc » 19 Dec 2015, 6:30pm

Bmblbzzz wrote:Oh dear. Google cars are programmed to drive correctly, which means within the limits of their own sensory and decision making capacities. Those sensory capacities are superior to humans, .


In some ways yes, in other ways no. Humans car drive on roads which haven't been mapped in high detail. Humans can see potholes and avoid them. Humans can drive in rain and snow. Humans can deal with a trackstanding cyclist and 4 way stops etc.

http://www.technologyreview.com/news/53 ... ving-cars/

axel_knutt
Posts: 1511
Joined: 11 Jan 2007, 12:20pm

Re: Problem with driverless car.

Postby axel_knutt » 19 Dec 2015, 6:46pm

Pete Owens wrote:
Mike Sales wrote: Is it being too careful?

No, the problem is the drivers hitting them are failing to take care. Note that the cars were not at fault in any of the reported crashes.

That might be what the law says, but it won't be what people think, any more than they think it's their own fault that they're caught by speed cameras.
“I'm not upset that you lied to me, I'm upset that from now on I can't believe you.”
― Friedrich Nietzsche

beardy
Posts: 3382
Joined: 23 Feb 2010, 4:10pm

Re: Problem with driverless car.

Postby beardy » 19 Dec 2015, 9:32pm

Humans can drive in rain and snow.


Yes but plenty of them come to a crunchy stop when they try. Always lots of extra roadside debris on the first snowy day around here and of course other road users have to keep well clear for their safety.

kwackers
Posts: 14054
Joined: 4 Jun 2008, 9:29pm
Location: Warrington

Re: Problem with driverless car.

Postby kwackers » 19 Dec 2015, 11:00pm

irc wrote:In some ways yes, in other ways no. Humans car drive on roads which haven't been mapped in high detail. Humans can see potholes and avoid them. Humans can drive in rain and snow. Humans can deal with a trackstanding cyclist and 4 way stops etc.

Amazing how good humans suddenly become when self driving cars are mentioned - when the evidence suggests exactly the opposite!
I was driving behind someone today who seemed to have difficulty with almost all aspects of driving, from figuring out what to do at junctions to dropping to nearly 20mph in a 40 whenever there was an oncoming vehicle.

Anyway, self driving cars have enormous potential for improvement and from a pretty decent starting point too. Humans on the other hand can't get better and potentially will get worse as their skills atrophy as they become more dependent on such cars to drive for them.

irc
Posts: 4654
Joined: 3 Dec 2008, 2:22pm
Location: glasgow

Re: Problem with driverless car.

Postby irc » 20 Dec 2015, 12:00am

kwackers wrote:
irc wrote:In some ways yes, in other ways no. Humans car drive on roads which haven't been mapped in high detail. Humans can see potholes and avoid them. Humans can drive in rain and snow. Humans can deal with a trackstanding cyclist and 4 way stops etc.

Amazing how good humans suddenly become when self driving cars are mentioned - when the evidence suggests exactly the opposite!.


The evidence is that self driving cars are a good idea that is many years away from everyday reality on normal roads. Humans for all their flaws are here now. One estimate is that

By 2035, 75 percent of vehicles sold worldwide will have some autonomous capabilities, such as being able to park themselves or drive at least part of a trip on autopilot, Navigant Research predicted in a report this week.


So 20 years from now self driving cars will still be in the future. Sounds about right.

http://www.mercurynews.com/business/ci_ ... bizarre-is

From the comments in that link ....

The more meaningful number (if they'd only give it to us) is how far those cars go, on average, between episodes that require human intervention. That is, how far would they go, on their own, before they got into an accident? From hints here and there in news articles, it looks like the answer so far is, about 20 miles. And that's driving around the mapped-to-the-last-inch streets of Mountain View, California.


A vehicle which drove itself for 99% of the time might be more dangerous than a normal car since the "driver" might have mentally switched off when intervention was required. If self driving cars were here at the same price as conventional cars I'd buy/use one. I won't hold my breath.

Bmblbzzz
Posts: 3034
Joined: 18 May 2012, 7:56pm
Location: From here to there.

Re: Problem with driverless car.

Postby Bmblbzzz » 20 Dec 2015, 10:09am

What on earth makes you think self-driving vehicles (not just cars, I think it's probable that lorries will see mainstream take up first, for economic reasons) cannot 'see' potholes or see through rain and snow? As they don't rely on visible light alone, they can do these things better than humans; much better, in the case of snow and rain.

Learning what to do with a trackstanding cyclist is just that – learning. As it was, the Google car acted sensibly – cautiously – after this reprogramming, it might not be so careful.

That this is still in the future is obvious. What's less obvious is that it's already here, quite likely in your own car. For instance, those parking sensors which buzz when you get too close to an object are a small part of the same system. In a fully autonomous vehicle they won't be buzzing, of course, just telling the controls to brake and/or turn. See also this:
https://youtu.be/vt20UnkmkLI
[youtube]https://youtu.be/vt20UnkmkLI[/youtube]

It's crude. At the moment.

So there are two ways in which these vehicles will arrive: implementation by implementation (parking sensors, windscreen wiper sensors, steering assist, etc,) that provide increasing amounts of help to the driver and become increasingly required by drivers and manufacturers, and a separate path (currently pursued by Google and a few other tech firms rather than motor manufacturers) of bringing a fully-fledged self-driving car to market. I'd suggest the rate of uptake will be something like that of other technologies such as email or mobile phones: something very minor for a couple of decades then rapidly becoming not only mainstream but a default and integral part of almost everybody's lives. The question is how many decades, but we're already at the talking about it stage and possibly even beyond the dismissing it completely bar a few freaks stage, so comparable to the '80s or late '90s for mobile phones.

Bmblbzzz
Posts: 3034
Joined: 18 May 2012, 7:56pm
Location: From here to there.

Re: Problem with driverless car.

Postby Bmblbzzz » 20 Dec 2015, 10:17am

The less-discussed aspects of self-driving vehicles are the social ones. Changes to employment (never mind Uber, cabbies will hate this!), increased attractiveness of road haulage compared to rail freight (because of no driver's hours etc), and increased reliance on cars generally. If you can dial up a self-driving car it will be like a chauffeur-driven limo. Or a mobile office. No need for the interior to resemble a vehicle at all. Businessmen will be able to literally sit at their desks on the M1. Or a travelling hotel room. There might even be a "Mobilodge" perfect for audaxers to reach distant events and get home again afterwards!

But all of these will still be cars. Safe, quiet, efficient, ego-free and non-polluting they may be but they will still cause congestion, isolation, contribute demand for road building, out of town shopping malls, and so on.

kwackers
Posts: 14054
Joined: 4 Jun 2008, 9:29pm
Location: Warrington

Re: Problem with driverless car.

Postby kwackers » 20 Dec 2015, 1:25pm

irc wrote:The evidence is that self driving cars are a good idea that is many years away from everyday reality on normal roads. Humans for all their flaws are here now. One estimate is that
By 2035, 75 percent of vehicles sold worldwide will have some autonomous capabilities, such as being able to park themselves or drive at least part of a trip on autopilot, Navigant Research predicted in a report this week.

So 20 years from now self driving cars will still be in the future. Sounds about right.

I don't know where they get that from but it's obviously nonsense. A decent percentage of cars now have *some* autonomous capabilities. At the moment this is confined to high end models which have braking and self parking but the evidence is this technology trickles down very quickly so rather than 2035 I'd suggest you'd struggle by 2020 to buy a car that didn't have some autonomous capability. Certainly 2035 is a joke.
irc wrote:A vehicle which drove itself for 99% of the time might be more dangerous than a normal car since the "driver" might have mentally switched off when intervention was required. If self driving cars were here at the same price as conventional cars I'd buy/use one. I won't hold my breath.

The assumption here is that you can turn it off. I suspect that once it appears 'self driving' even when 'switched off' will continue in supervisor mode. So I doubt for example you'll be able to turn collision detection off and possibly obstacle avoidance.
At any rate I think the switch will be almost binary, I suspect insurance companies will put hefty premiums on cars that allow manual control and cheaper premiums on cars that don't. As a result I think a lot of folk will simply opt not to drive at all.

irc
Posts: 4654
Joined: 3 Dec 2008, 2:22pm
Location: glasgow

Re: Problem with driverless car.

Postby irc » 20 Dec 2015, 1:37pm

Bmblbzzz wrote:What on earth makes you think self-driving vehicles (not just cars, I think it's probable that lorries will see mainstream take up first, for economic reasons) cannot 'see' potholes or see through rain and snow?


Because Chris Urmson, director of the Google car team.says so. As per the link I posted above.

Urmson says safety concerns preclude testing during heavy rains. Nor has it tackled big, open parking lots or multilevel garages. The car’s video cameras detect the color of a traffic light; Urmson said his team is still working to prevent them from being blinded when the sun is directly behind a light. Despite progress handling road crews, “I could construct a construction zone that could befuddle the car,” Urmson says.

Pedestrians are detected simply as moving, column-shaped blurs of pixels—meaning, Urmson agrees, that the car wouldn’t be able to spot a police officer at the side of the road frantically waving for traffic to stop.

The car’s sensors can’t tell if a road obstacle is a rock or a crumpled piece of paper, so the car will try to drive around either. Urmson also says the car can’t detect potholes or spot an uncovered manhole if it isn’t coned off.


http://www.technologyreview.com/news/53 ... ving-cars/
Last edited by irc on 20 Dec 2015, 1:46pm, edited 1 time in total.