The legal issues might not be a game-stopper. Any lawyer worth their legal degree should be able to argue that the fault resides with someone other than their client.
A new report has come to the rather startling conclusion that there will be nearly 100 million autonomous cars sold per year globally by 2035. Navigant Research, a clean-energy consultancy and research group with offices in the US, UK, and South Korea, forecasts that self-driving cars will comprise 75 per cent of all light- …
Obviously, if the car is genuinely self-driving, the fault lies with the car if it makes a mistake.
Equally obviously, a range of punishments will need to be defined, varying from periods of clamping for minor offences right up to having its OS replaced by Windows ME (for causing death by dangerous driving, for example). I realise that last one might be declared to be "cruel and unusual" by the Americans.
(Lapsing back into serious mode, I'm pretty certain it'll be the manufacturer.)
As far as culpability is concerned I made a prediction to my Russian wife today. Very many Russian drivers for a number of reasons including dodgy cops, have dashboard cameras mounted in their cars (witness dozens of YouTube videos from Russia), I think as soon as insurance companies get a whiff of this fashion they will do all they can (after buying shares in dash camera manufacturing companies) to make it law that all cars are built with dash cameras (as well as rear facing, think Google glass with wheels) and that they are functioning as part of the MOT test. Ten years max and every car will have them.
Being an old fart I doubt I'llbe driving when 'autonomous cars' are the norm, as it is I don't trust any other drivers on the road but at least I can toot the horn and shout obscenities at them. It won't have the same impact yelling at a small black box buried somewhere inside the car.
I have to say though that a figure of 100 million a year is a little on the unrealistic side short of an economic miracle for the whole planet.
>>I think as soon as insurance companies get a whiff of this fashion
Um, I think something which is an internet phenomenon is probably going to have attracted their attention already, and that the revolutionary idea of a camera on your dashboard probably wasn't hard to come with in the first place.
It doesn't really gain the insurers much anyway. On average each insurer will still have to pay out as often so while this might make the system more accurate, most of the errors in the current setup probably balance out :)
Insurers generally pay on a "knock for knock" basis, regardless of "he said, she said" issues.
Cameras are of major use when the other party has NO insurance or traffic charges are involved (and having seen some of the crashes caught on cam in russia, I have no doubts that charges apply in a lot of them.)
As for rear-facing, that's useful, but sidefacing would help as well. I've had a couple of recent issues where reviewing recorded footage doesn't show stuff happening nearby (such as the twat in the next lane not noticing there's a car beside him before changing lanes) that could have been useful.
I've been holding off buying a dedicated recorder until there are more than 2 cams on 'em.
"If I'm able to have the car drop me off, and then send it away to a car park, or back home; and summon it back to me when I want to go home myself, then I will definitely buy one."
Which still means that for a huge amount of time, the car will not be doing anything apart from taking up space.
The sensible and efficient model would be a self-driving taxi that you hire to get you to where you want to go and then which can go and be used by someone else meaning less vehicles on the roads, less traffic and less congestion, not to mention less fuel being burned.
Yes, I can really see it come to pass that instead of Joe Normal buying a car, he just signs up with a company that has a fleet of cars that make rounds, picking up and delivering passengers.
A bit like a cross between a taxi and a bus. Hit the button on your phone's ap, vehicle picks you up at the door, drops you off at your destination. All carefully plotted to minimize distance and travel time of its load of passengers.
This could end up hitting traditional car companies hard!
"A bit like a cross between a taxi and a bus. Hit the button on your phone's ap, vehicle picks you up at the door, drops you off at your destination. All carefully plotted to minimize distance and travel time of its load of passengers."
Indeed, the big cost of a taxi fleet are the drivers.
Given maybe 80% of journeys are home-work-home a vehicle that gave you your personal bubble, all the features you wanted and was crash safe could be quite feasible. And let's be real here. All that tech is going to have a substantial impact on price.
But a 100 million of them in 25 years? I don't think so.
In addition there are journeys from home to the station to get a quicker train...
As to the robo-taxi app, as long as it lets me specify I don't want to share I'll be happy: I prefer my choice of music.
Thinking about the liability angle - I presume, as long as manual controls remain, the occupant would be held responsible for drink driving etc?
"The sensible and efficient model would be a self-driving taxi that you hire to get you to where you want to go and then which can go and be used by someone else meaning less vehicles on the roads, less traffic and less congestion, not to mention less fuel being burned."
'Johhnycab is here to serve you'
"What a load of utter bollocks."
Yep, I think that about covers it.
The article (and seemingly the study) touches only briefly on liability. And there is a fatal stumbling block; artifical semi-intelligent systems such as those that drive the current autonomous cars aren't known to be fool proof. We've got a pretty good idea that they work reasonably well, but that's not proof. This is reflected in current law: places where they are legal still require a competent driver to be behind the wheel in a sober condition and paying attention, to take control just in case.
So that leads to three possible future situations.
First, the law doesn't trust the tech and requires the 'passenger' to be able to become the 'driver' at a moments notice. In which case, what's the damned point of the whole thing anyway? If I've got an autonomous car I want it to be able to drive me home pissed as a newt from any watering hole that I choose, but I can't; I have to remain sober and with it. That's the current situation AFAIK. Sooner or later there will be a case where such a car is involved in a bump and all sorts of legal arguments will ensue.
Second, the tech advances to a point where the law and can reasonably trust the tech and pass all liability on to the manufacturers. Clearly we're a long way from that, and I don't think that we'll ever really get there. Not even the aviation industry has managed to wean itself off having two pilots in the front. And their operating environment is much more controlled (i.e. far simpler from an automatic software point of view) than the roads; their attempts so far have been far from reliable in UAVs.
Third, and most distastefully, the manufacturers do a large amount of lobbying and get autonomous vehicles mandated by law, but with the liability for their malfunctions residing with the 'passenger'. The old "you have to have it, but its your fault if it goes wrong" problem. In some countries (the US?) where the legislative system is completely broken and at the mercy of the powerful lobbyists I don't think that you can rule this situation out.
Boredom Threshold, and the Human Inability to Cross it Quickly
Regardless, there is a real danger that the public will fall for the marketing and the blurb and will start trusting the tech. Ok, so we trust car design and manufacturing now, but even though cars are mostly very mechanical at a fundamental level (so no room for complicated software to break) we still can't make Toyotas drive along the motorway at a speed of our choosing all of the time.
Put most people in an autonomous vehicle and they will stop paying attention; it will be just too damned boring. It's bad enough at the moment in normal cars. Expecting someone to intervene at a moments notice when something is going badly wrong quickly after they've gotten used to months of trouble free operation is unrealistic, but failing to do so will (currently) result in the liability resting with them.
Security? Is there any?
And none of that even begins to address the opportunities for the maliciously minded hacker. Google's car is no doubt wirelessly connected via the internet to The Chocolate Factory. How long before someone spots a crazy simple security weakness in that? I mean, has anyone done any penetration testing on these things at all? For all the current drivers know it could be dead easy for some script kiddie on the other side of the planet to hack into their car and send it haywire and cause an accident, just for the kicks. Would you want that happening to your car with you in it?
Driverless cars may not be known to be foolproof, but humans ARE known to NOT be foolproof. What's more, they're known to be not be foolproof at a pretty well defined rate, as measured by deaths, injuries and accidents. You know, the stuff insurance companies use to set your rates.
Insurance companies will be able to do the same with a car/software combination, and once they can perform better than the worst drivers, they'll get letters like the following:
Dear bazza, based on your driving history we are currently charging you $2000/yr for liability insurance, but if you switch to a car running GM DriveSense version 4.2 or higher your rates will drop to $1200/yr, conditional on your car indicating manual driving only on non-public roads.
You won't have to take them up on it, but at some point it'll be as hard to get auto insurance as a manual driver as it is today if you've had 6 OWIs and 3 fatal accidents. i.e., pretty much impossible unless you're absurdly wealthy.
Driverless cars will never be foolproof, and people will die at their "hands". Big deal, guess how many people die on the roads worldwide each year? Over a million! Computerized driving is a much lower bar than computerized flying, since commercial jet pilots killed less than a thousand people last year (and that gives them the blame for all deaths, even those that weren't pilot error) There are other factors at work (such as pilot egos) for why autopilot isn't used more heavily, especially in low visibility situations where autopilot is clearly superior to human pilots.
"Driverless cars may not be known to be foolproof, but humans ARE known to NOT be foolproof. "
"Driverless cars will never be foolproof, and people will die at their "hands". Big deal, guess how many people die on the roads worldwide each year?"
Yeah right. Fools have always and will continue to find imaginative ways to kill themselves and possibly others. The problem with autonomous cars is that you are placing your life entirely in the hands of other people - you have no control, no choice whatsoever. So then, how many of those people are fools? How many of them are malicious? Inevitably, a proportion of people involved in your safety are idiots, yet none of them will be involved in the car crash they'll end up causing. Personally speaking I'd rather choose to take responsibility for my own safety as far as is possible, and I definitely wouldn't want to be bored witless behind the 'wheel' of a car I'm not allowed to drive but am somehow required to supervise.
Your statistic of a million a year glosses over many regional differences. The roads in Germany for instance are amongst the safest on the planet, yet they have no automation and impressively high speeds. Go figure.
The problem with automatic cars is that they may reduce the accident rate in the short term, but they're inevitably just one unfortunate software bug away from causing a few billion car crashes in a single day (assuming that there's that many in use). Does that really sound like a good idea? Arguably it's unlikely, but no one would ever consider the outcome to be acceptable under any circumstances. Google can't even get a calendar right on a mobile phone; who says they can get a car right?
Also I note you didn't consider the opportunity for malicious hacking attack. Want to crash someone else's car? Deploy an exploit. Internet connecting these things sounds like a sure fire recipe for trouble on the roads, and you know that given the opportunity someone out there is guaranteed to give it a go. I just hope they're not internet connected, though knowing the US's / Googles propensity for connecting literally everything to the net, I fear the worst.
"Computerized driving is a much lower bar than computerized flying"
Totally and completely wrong. Computerised driving is far harder than computerised flying. An aircraft has a very simple navigational problem to solve (fly from here to here), and obstacle avoidance is easy (fly at this height, pay attention to the TCAS). Whereas the obstacle avoidance part of an automatic car is a really difficult problem. I notice that current auto-cars are mainly currently used in dry sunny places. I'd like to see them work reliably on a horrible stormy, rainy night with lumps of tree and rubbish flying all over the place on a narrow and twisty road in the civilised world. What if a fly squishes over a sensor? Is that obstacle ahead a genuine problem, or is it just a piece of paper blowing in the wind? And a car doesn't even have the luxury of being able to go where it wants; there's a road to identify, follow, and keep to the correct side of to within a couple of feet or so. Planes don't even have to be that precise when landing on a nice and straight runway. And, apart from landing and take off, there's generally loads of time in an airliner to sort out problems. In a car you've got perhaps half a second to respond to a system failure on a busy fast road.
There is a growing feeling in the aviation industry that the reason pilots are making mistakes is because the automatics are doing too much. Pilots these days (depending on which airline) are really just system supervisors, and only rarely do they actually do any flying. It's hardly surprising that when the automatics fail or are unavailable that they make mistakes. Even Airbus acknowledge this, and apparently the upcoming A350 will be less 'automatic' and will require the pilots to actually do some flying.
Sure, you could remove the pilots altogether and go fully automatic, but the crash rate for UAVs is appalling in comparison to manned aircraft. Making that change is, at the moment, guaranteed to lead to a significant increase in fatalities.
Why do you think you can react more quickly and correctly to something like a deer or another driver swerving directly into your path than a computer could? You sound like one of the fools who disables his ABS believing he is better at braking than it is.
A software bug that causes billions of car crashes in a single day? Hacking into Internet connected cars? You really have no idea of the type of engineering that goes into life critical software, such as that which runs airplanes. This isn't the open source "release early, release often" strategy followed by Google but slow and methodical and passing through many layers of change control.
I've never heard of a bug in autopilot software that led to any more than one crash at a time, nor have I ever heard of someone hacking into a plane and causing it to crash. But feel free to worry about stupid stuff because you think there is only one way to develop software. It won't be easy and there will be fatal mistakes, but people need to worry about things that are actually a potential problem, not go all Luddite and just throw up your hands and say it'll never work and we shouldn't even try.
If planes carrying hundreds of passengers can take off and land on autopilot and fly from place to place, and fighter planes can engage other planes at Mach 3 before landing on an aircraft carrier, I reckon driving you to Asda is not exactly outside the realms of possibility.
There is absolutely no reason on the technical side that cars can't drive at least as well as humans, even amongst other human driven cars. It may well require a lot more work to figure out, but there's nothing magical about your brain that can't be reproduced.
Once upon a time the idea that a phone could beat a grandmaster at chess would have been laughable.
Driverless cars, are the future.
For most people, most of the time, driving is a chore, a dangerous task which they barely master and which can cause deadly accidents.
Currently, only private cars can take you from any point A to point B, without changing conveyances. Trains and other public transport are constrained to specific routes on account of practical limitations, chiefly by virtue of their communal nature. Taxis are too expensive to be a regular or long-distance option owing to the manpower cost involved.
If driverless cars became the norm, firstly, road travel would become safer and more pleasant, and secondly, automatic hire cars (like taxis, but without drivers) would become a viable possibility. As a result, there would not even be a need to own a car.
"Currently, only private cars can take you from any point A to point B, without changing conveyances."
Really - my pedal cycle does the same.
In fact a friends motorbike does the same as well...
As a cyclist I'd far rather see automated cars than driven ones - machines tend to be far more patient and predictable, as well as getting tired/distracted far less often.
As a motorist I'd rather see automated cars than driven ones - it would make life cheaper and far less stressful on the roads of today.
As a motorists who quite enjoys driving I'd therefore also want a "driver" option on my vehicle - which is available for use in appropriate locations (i.e. tracks).
Of course an automated vehicle has no inherent objection to an electric drive (and the low noise level would be appreciated by many) allowing redistribution of emissions to more appropriate locations (and preferably to nuclear/geothermal/??? plants). Infrastructure involving on-the-move charging on motorways would become much easier to justify as well.
There are many, many benefits to this technology, and as for objections about driving at night in the rain. They don't exactly use HD cameras - they use all sorts of different sensors, which allow the software to have a good, all round, picture of the surrounding area - and more importantly to know where it doesn't have a view and drive accordingly....
I guess im not one of the" most people".I just love driving, wether its my car, or my truck or my tractor.
My interest is to try and get the highest milage from every gallon of fuel I buy., as well as operating whats in fact a transportation machine in the most accurate way from here to where ever im planning on going.
The truck is an interesting case.
I needed a small commercial vehicle back in 1985 to carrry a heavy weight.all 2.5 tons in fact.
So built it. Needed to get special vehicle classification for it. DVLA wer most helpful.
Used it all over europe in the 1990's.
So when I drive Im not beholden to any one else like a coach driver or train operator.
It just me and my decisions at play.
same I guess for pilots who fly for pleasure.
Tho im sure commercial pilots enjoy getting it right everytime as well.
Anyone else drive for pleasure? Cant be the only one?.
I love driving too.
When the roads are windy, uninterrupted and clear of traffic. Or have no speed limit, and loop back on themselves. Generally, I enjoy driving a whole lot more if I don't have somewhere to be.
I dislike driving in traffic, in built-up areas, and on freeways.
The latter represents 95% of my total distance travelled by car. I would happily have a self-driving car if it could take care of that 95%, as long as it left the other 5% to me.
You aren't the only one, ted. Driving is one of life's pleasures to me. I don't mind driving in traffic (it is still a skill), or on long motorways - there is the knowledge that I am doing perhaps the most difficult thing I have in my skill-set. When I'm competing, it gets better!
I recently rented a car with cruise-control etc. For economy, I set the speed at a reasonable compromise - not dawdling, but not as fast as I would if I had full control of the throttle. Most boring journey (350 miles) I've ever done, and it was really hard to keep concentration.
For me it is definitely a chore, I gamify it trying to spend as little fuel as possible and I have been known to look all smug when the attendants look surprised at the mileage on the on-board computer when I take the car for revisions, but really, driving is as boring as it gets, no wonder so many people die because they fall asleep at the wheel.
I can see how the illusion of control can make some people happy about it, but it doesn't do anything for me.
Lots of people like riding horses but they were replaced by cars. They are now exclusively used for recreation (OK 99% to account for people on farms, etc but you get the point).
The same could happen to self-drive cars but more likely is that you'll always be allowed to take control, except perhaps on motorways where it makes sense to force everyone to be automated.
Driving in the Scottish countryside is not going to disappear.
Add me to the list of recreational drivers.
I do the odd bit of hypermiling (45mpg at 70mph in a ratty old 1.7 Puma - awsume) and more regularly, drive point to point as quickly as possibly, safely 40 miles of mixed DC/A/B roads in just over 35mins...rather less than 45mpg ;-) ) and I am also one of those who finds that unless they are engaged in driving - moving through traffic, overtaking, setting a good line through a corner, then I get bored and distracted. Driving in 45mph traffic is a massive chore as there is nothing fun to do (beyond the obvious keeping a safe distance, reading traffic ten cars ahead, etc).
If I can, I overtake, and get back on the boil, frankly.
I'm very much of the opinion that if people were taught to enjoy driving as a skill to master to make the most efficient progress (in any manner) from point a to b and get satisfaction out of the 'to' part, accident rates would drop rather a lot. It's the people who are on autopilot (without computer assistance, natch) all the time who are the problem. How many times a day do you see people changing lanes or pulling into traffic or exiting a roundabout without indicating. Not even indicating correctly, just indicating at all?
At least computers might make indicators less of an optional extra.
People who don't take the skill of moving over a ton of metal and glass at speed with deadly seriousness are a fucking liability IMHO.
I could rant on this subject for hours...
As for automated cars - we can't even get automated trains in the UK. If we get driverless cars before driverless trains, then something has gone badly wrong.
Unlikely. Self driving cars still don't allow you to spark up a fag indoors.
However a quick trip to the local off license, and you can do what you like in your own front room with a few friends.
And there is the problem facing the publicans, in a nutshell.
If the car can drive you then distance becomes unimportant and you're more likely to see stuff like 1970s-era New Zealand Booze Barns (essentially: 400-2000 slot carparks with a building in the middle containing 2-3 bars) at the edge of town than a revival of the neighbourhood pub.
Your forget that other manufacturers were not having that widespread problem. So, Toyota couldn't even produce a floor mat that wasn't defective. A floor mat is a pretty simple device as compared to a car that can drives itself. If the engineers cannot produce a simplistic floor mat, how are they going to handle something very complex?
Driverless cars isn't going to happen in any meaningful way any time in the next 100+ years (during which time the car itself may be replaced) - apart from possibly on motorways.
I refer you to the speech recognition debacle, always 10 years away, always looks likes it works, never works well enough for people to actually use it. If it worked smartphones wouldn't have screens would they?
If driverless cars worked, cars wouldn't have steering wheels. Ain't gonna happen.
"I refer you to the speech recognition debacle, always 10 years away, always looks likes it works, never works well enough for people to actually use it. If it worked smartphones wouldn't have screens would they?"
Angry birds doesn't have the same effect if you just yell at them to kill the damn pigs.....
> Driverless cars isn't going to happen in any meaningful way any time in the next 100+ years
You guys ARE aware that there are driverless cars in, in real traffic, *today* right? They've been on California roads for several years, driving hundreds of thousands of miles in real traffic with a perfect zero accidents record.
See e.g. http://www.pewstates.org/projects/stateline/headlines/states-take-the-wheel-on-driverless-cars-85899493304
The ONLY thing presently holding back driverless cars is legislation. They are presently expensive, but weigh that against 24/7 operation and zero accidents for e.g. taxi and trucking companies, and that cost is easily recoverable.
If driverless cars worked, cars wouldn't have steering wheels.
That's right up there with "if God meant us to fly, He would have given us wings" in terms of silliness. Besides, speech recognition is a done deal unless you're a drunk Glaswegian, trying to shout at the phone while in the stands at the FA cup final, or using Siri.
The one reason why a self-driving car sounds tempting that you could get plastered at the pub and be driven home afterwards without being a danger to yourself, others, and your driving licence.
The problem is that this won't be allowed. The official reason will be the liability question: somebody with the power to override the automatic settings must be legally responsible for the vehicle. The real reason is that once the government has managed to take away people's liberty to have an enjoyable evening at the expense of tomorrow's productivity, it will never give it back.
In many countries apart from the USA
Personally, I think the lawyers will have a field day over there. Autonomous Car has an accident. Then sue the occupants and the car maker into the ground
Rinse, repeat until no driverless cars are usable in the USofA. The country where you can be sued you for just about any reason someone could dream up.
Driverless cars are an open invitation for endless lawsuits. I'll bet that the lawyers are ordering new legal pads as we speak.
I would have thought comparing the black boxes of the cars involved would I identify the cause whether it be user error, manufacturing error, third party error or just "bad luck". There will always be a need for insurance but based on car reliability not age, shoe size etc
Any non-automated vehicle should be retro fitted with warning lights and squawk box to let the "intelligent" road users there is a Muppet behind the wheel. Picture the Ferrari with the flashing tit on top!
I cannot think of any secure comms protocol that has not had security holes in the past. I doubt very much that any secure comms protocol in existence is completely without security holes that can be exploited in the future.
If ever this comes about, terrorists will pump a lot of time and effort into figuring out how to hack into this. Not only will they be able to cause a single car to swerve over to the wrong side of the road, it will first pass this message to other cars in the vicinity. Instant carnage.
Actually, given some of the stuff reported recently in El Reg, perhaps it will take very little time and effort to hack such a system.
and the speed camera technicians, and the tow truck drivers, and the driving instructors, etc.
Seriously, has anyone thought of the impact on the "motorist economy"? (Those who make bucks out of training drivers or cleaning up after their mistakes).
(And yes I did not include Insurers - you really think less people having less accidents will mean they will lower your premiums?)
Biting the hand that feeds IT © 1998–2019