[Closed] Self-Driving Cars

Apart from the collision avoidance aspect - which driverless cars will easily be better at than humans - for me, the greatest benefit will be improved traffic flow.
No more cutting in at the last minute, no more lane hogging (mostly because lane usage will change) no more speeding, no more distracted drivers.

The ownly downside I can see is that driverless vehicles will, out of necessity, be following Sat. In the case of lorries, this will inevitibly lead to vehicles stuck down narrow roads.

On balance, things can only improve, I believe.

So a driverless car is travelling along a mountain pass, there's a coach approaching too far over the white lines to avoid a collision without taking you and the passenger over the edge of the cliff. The collision will potentially end in a greater number of fatalities, how should it react?

But why would the coach be over the white line in the first place? It's going to be driverless as well, so it will be driving within its limits on the correct side of the road.

Bring it on. I'm hoping that once this technology is adopted, it will be so much safer that human driving will simply be priced off the road by the huge insurance cost.

We are so far behind countries like the Netherlands that I don't think we will ever have effective cycling infrastructure in UK urban areas (let alone on our miles of country roads, which can be even more dangerous for cycling). Driverless cars might just leapfrog this problem by making cycling specific infrastructure unnecessary.

Think how popular cycling might become if it is the only remaining form of road travel where humans will still have full autonomy.

Not onlybselfvdrivibg cars but self driving electric cars. Cant wait. Especially if they are all made nice and boring and are no longer seen/used as status symbols. (As if that will happen)

I don't think anyone has addressed the security aspect of autonomous cars. To function properly they will have to be online and networked and so open to hacking. Its not a question of if but when autonomous vehicles are hacked. Say someone manages to hack into all the "Fiesta self drive Zetecs" and either gets them all to apply brakes or go to 100% throttle at the same time. I'm convinced this is a major barrier to acceptance of autonomous vehicle technology and it seems incredible the likes of Google or any of the car manufacturers working on this tech haven't addressed it.

[quote=globalti ]From my limited experience [b]of old technology which is in no way comparable to what we're discussing[/b] I'd say the technology has a long way to go before it becomes viable.

fixed

[quote=belugabob ]The ownly downside I can see is that driverless vehicles will, out of necessity, be following Sat.

I suspect they might manage to improve that as well.

I like your thinking, kcr. Have pretty much given up on the idea of having decent cycling infrastructure here in my lifetime, but I don't think it's particularly Tomorrow's World optimism to think that driverless cars will not only exist well within my lifetime, but they will have pretty much taken over. I was also wondering what would be the forcer behind adoption - yep insurance will be one significant thing, but they should be a lot cheaper to run in general - as others have said the culture of usage will change.

[quote=uponthedowns ]I don't think anyone has addressed the security aspect of autonomous cars. To function properly they will have to be online and networked and so open to hacking.

I'm not sure that is necessarily true. You should maybe check the definition of "autonomous" 😉

Sure, there appear to be some advantages of networking for travelling in "trains" and stuff like that, but the whole point of the first generation is that they can interact with current traffic and for that they have to be able to make their own decisions without any outside influence.

it seems incredible the likes of Google or any of the car manufacturers working on this tech haven't addressed it.

How do you know they aren't? Also I'm not convinced that even if they are networked in some way that it needs to be via a traditional comms stack - that would certainly be a good way to introduce lots of security vulnerabilities when the data exchange requirements aren't that high and are quite specific. I'd think you should be able to isolate a lot of the comms from hackable stuff and have air gaps to anything critical. Sure it's something which needs to be considered, but I don't think it's as big an issue as you make out.

In industry it's common for numbers like 1:100,000 to be the acceptable chance of killing an individual worker on site, but the chances of killing someone off-site have to be 1:1,000,000 (i.e. 10x less). So you as a driver in the car have made your decision to be there and accept that risk, so the hypothetical situation where the choice is kill the driver or kill the pedestrian, the car would have to kill the driver unless it judged the chance of the pedestrian surviving to be 10x higher (which is unlikley given airbags etc).

So I'd expect them to always default to stuffing it in the hedge or infront of the HGV (killing the driver and injuring no one else).

Numbers don't equate to ethics, if the pedestrian stepping out in to the road made a mistake why should others die for their actions? Even if they are a child, in the natural world - idiots die for their mistakes. Can the car decide whether the occupant has more utilitarian value than the child? Noooope, unless that childs mind and future life has been uploaded to some networked cloud computer.

Imagine the Daily Mail headline "Stephen Hawking Tragically Killed After 7 year Old Illiterate Pikey Runs In Front of Smart Car". Not going to happen is it? Although the power of "someone think of the children" might prove me wrong.

I mean, who is going to get into a car that might be crashed by someone putting a silicone doll in the middle of a blind corner.....

....

No one, they'll buy the car more likely to kill other people.

[url= http://www.bbc.co.uk/news/technology-33650491 ]1.4 million Fiat-Chryslers recalled because their systems could be hacked.[/url] And another group has demonstrated hacking through digital radios. The more the car is in charge, the more hazardous hacking becomes

Just continuing to miss the point. Why would the car have to decide between a child and others and who it should kill?
It would be driving in such a way to minimise the risks to an extent people wouldn't be killed. If it was driving where there were children, parked cars etc,. it would travel at 15mph or even 10mph so risk of actually killing anyone has pretty much gone.

Recent data I saw was some speed testing performed on a road with a school on it in the afternoon when kids would be about. Average speed of 100 drivers was something like 38mph with one driver even managing over 60mph.
The electric car would be doing 15 or 20.
Which is the safer in the above example?

And don't forget it is not just speed it is the lack of awareness or concentration, impatience etc,. that majority of drivers are showing when driving.

Yes some people will still be killed but it won't be 2,000 a year and the roads will be 100 times nicer to cycle on

A lot of these 'it will fail when scenarios' are based on the assumption that every single road and situation will allow for autonomous driving, which is still a very long way off.

The Volvo project in Gothenburg set for next year/2017 has a limited set of roads that will allow for autonomous driving, and none will have pedestrians or cyclists. Once this is proven to be successful then the application of this technology can be increased to other areas/situations.

The majority of the claims made by various car manufacturers are very far fetched regarding the scope.

[quote=Tom_W1987 ]Numbers don't equate to ethics, if the pedestrian stepping out in to the road made a mistake why should others die for their actions?

Am I missing something here? Does the ped stepping into the road have a shotgun which they accidentally set off? Or is it somebody else introducing danger to the environment?

[quote=timba ]1.4 million Fiat-Chryslers recalled because their systems could be hacked. And another group has demonstrated hacking through digital radios. The more the car is in charge, the more hazardous hacking becomes

You know what I'd do to fix that (based on experience working in IT security)? I'd put an air gap between the internet connected entertainment system and the car control systems. Autonomous cars are not going to be doing critical stuff over the internet - apart from anything else as I mentioned before that would require a constant reliable internet connection, which simply isn't possible. Of course you could hack the car to car data link, but that's a completely different proposition.

The reason those things got hacked is mainly because nobody put any thought into proper security.

The car will need a connection to the Internet to receive and transmit data, which it will of course use to make decisions. However the majority of the data needed to make the decision will be provided by on car systems.

Hacking is clearly a major risk and is something that is taken every extremely seriously. There are a number of different methods used today to create a secure connection, both to/from the car and to/from the cloud based service it communicates with.

[quote=The Swedish Chef ]The car will need a connection to the Internet to receive and transmit data, which it will of course use to make decisions.

Only for sat-nav, routing type stuff, and I'd suggest that actually an autonomous car doesn't need that at all to function properly - how else would it work when it loses the signal? One very basic security step is to separate the systems, so any internet stuff like that isn't tightly integrated with the main control system as it isn't needed for instant decisions - with a very tightly controlled secure data link between the two.

In the near term not all roads will be certified for autonomous driving so the car will need to enquire if the next segment on the planned or predicted route is valid for autonomous driving or not.

Also certain segments will be dependent on weather conditions, bridges in high wind, and other segments may be turned off due to accidents, so the car will need to periodically request an ok/nok to drive autonomously.

toby1 - Member

So a driverless car is travelling along a mountain pass, there's a coach approaching too far over the white lines to avoid a collision without taking you and the passenger over the edge of the cliff. The collision will potentially end in a greater number of fatalities, how should it react?

[img] [/img]

Am I missing something here? Does the ped stepping into the road have a shotgun which they accidentally set off? Or is it somebody else introducing danger to the environment?

Well lets assume that autonomous cars follow the rules of the road exactly and drives perfectly, for one to enter a scenario where the pedestrian may be killed - then the pedestrian will have had to have made a mistake either misjudging the distance to the oncoming car or having crossed at a blind spot.

Their fault, so they should pay the price.

[quote=Tom_W1987 ]Their fault, so they should pay the price.

Wow!

I'm equally horrified by a machine attempting to make an ill informed utilitarian judgement as to who should live and who shouldn't. So IMO, the best option is to just do what a human would do....apply the brakes or maneuver in a way that doesn't endanger the occupants.

That and it would encourage irresponsibility on the part of pedestrians if they thought that cars would do their very best, to the point of killing their occupants, to avoid a collision.

We already live in a world where the Flynn effect is going into reverse, why would we want autonomous cars potentially adding to the survival of people who make poor decisions over those who haven't, albeit probably in a highly statistically insignificant way.

Should someone who is suicidal endanger the life of an occupant of a car, who is perhaps a functioning member of society?

The idea that this will reduce cars as a status symbol is very unlikely. If anything, they're likely to end up more so.

The plebs will use taxis which will be on patrol based on previous usage patterns so that you rarely have to wait long. These cars will be like public transport ie fairly basic.

The next step up will be much the same as current cars but self driven.

The real status symbols will be like self driven luxury lounges.

Nothing will really change. People will still be judged by their car.

It's still the people in the car who've brought something dangerous into the environment (excluding situations involving peds with shotguns). Who exactly has made the bad decision?

[url= http://worldif.economist.com/article/11/what-if-autonomous-vehicles-rule-the-world-from-horseless-to-driverless ]Self-driving cars[/url]

A really interesting article summarising a lot of the work that is being done towards making this a reality, and also all the (many, many) arguments for doing it...

The safety benefits alone are massively compelling

Today 94% of car accidents are due to human error, according to America’s National Highway Traffic Safety Administration, and the three leading causes are alcohol, speeding or distraction. Accidents kill around 1.2m people a year, reports the World Health Organisation, equivalent to a 9/11 attack every day.

Google’s driverless vehicles have driven 1.8m miles (2.9m kilometres) in the past six years, and have been involved in 12 minor accidents, none of which caused injury and none of which was the car’s fault.

A study by the Eno Centre for Transportation, a non-profit group, estimates that if 90% of cars on American roads were autonomous, the number of accidents would fall from 5.5m a year to 1.3m, and road deaths from 32,400 to 11,300.

And that's not adding in the lives saved from reducing obesity from more people cycling as it will be safer

Autonomous cars are not going to be doing critical stuff over the internet - apart from anything else as I mentioned before that would require a constant reliable internet connection, which simply isn't possible. Of course you could hack the car to car data link, but that's a completely different proposition.

I also doubt there'd need to be a continuous connection to the internet but they will need to download new map data, maybe traffic info, software updates etc. and maybe upload diagnostic data so they will be connected. The map/satnav system will have to communicate with the car controls so I would have thought the opportunity for hackers is there.

but they will need to download new map data, maybe traffic info, software updates etc. and maybe upload diagnostic data so they will be connected. The map/satnav system will have to communicate with the car controls so I would have thought the opportunity for hackers is there.

All that functionality is available today, for example in Volvo on Call, amongst others.

As I said before, the need for connectivity comes from the need, legal or otherwise, to inform the car as to when and where autonomous driving is allowed or not. Whilst not requiring a fixed connection, the ability to check each road segment requires a pretty good level of connectivity.

Consider the following situation; you are riding along in your autonomous vehicle on a busy road with pedestrians on both pavements. There's a truck coming in the opposite direction travelling at the speed limit.

Just as the truck begins to pass you a child runs out in front of your vehicle.

The 'car' has three options; swerve left and kill you (the occupant) in a collision with the truck or swerve right and kill a pedestrian on the other pavement or 'it' can carry straight on and kill the child.

Scenario's like this are giving car manufacturers and their legal teams sleepless nights - I hope

I don't think this is a particularly difficult decision for the car designers to make, because it's effectively a "non-decision" for the software in the car.

It MUST be designed to see pavements (with pedestrians on) as a no-go area under any circumstances. This is, I hope, obvious... It will also see a lane with oncoming traffic as an inadmissible zone (because if you cross the lane the consequences for many people are unpredictable).

So the only option is for the computer to brake as hard as possible and mitigate the impact with the child (pedestrian airbags, for example). From 30 mph the stopping distance is small; from 20 it's pretty much a single car length.

This technology is already here - my Volvo will maintain its position in the road, track pedestrians on the pavement and intervene if it judges a collision is unavoidable.

There is no complicated programming or ethical decision making to be made (IMHO, of course!) so I suspect autonomous driving, on motorways at least, is only a few years away.

so I suspect autonomous driving, on motorways at least, is only a few years away.

Next year hopefully! 😀

It's still the people in the car who've brought something dangerous into the environment (excluding situations involving peds with shotguns). Who exactly has made the bad decision?

If they don't crash and cause accidents unless someone else makes a gross mistake, they aren't inherently dangerous.

Train companies aren't usually sued because someone committed suicide on their train tracks, they aren't responsible for those peoples deaths.

More good news

[url= http://www.cityam.com/221505/audi-bmw-and-daimler-accelerate-driverless-car-plans-28bn-nokia-here-maps-deal ]Audi, BMW, Daimler buy mapping technology for self-driving cars[/url]

European car sales have slowed right up, so investment in self-driving cars (which could replace 100% of the installed base of human-driven cars in time) is a good way to future growth and profits

xkcd.com today.
lol

[quote=Tom_W1987 ]Train companies aren't usually sued because someone committed suicide on their train tracks, they aren't responsible for those peoples deaths.

They don't put train tracks alongside pedestrian footways. There are very few places where logical walking routes involve crossing train tracks.

[img] [/img]