Forum menu
Wonder if they will hold the software programmer liable for the death.
In reality it will be the first piece of much evidence used to ensure that in the not too far off future no pedestrians or cyclists will be allowed on roads with autonomous vehicles 'for their own safety' and that people will only be able to cross on crossings by law even in the uk.
surely it should be the driver in the car who should be charged? He was there as a safeguard.
Wonder if they will hold the software programmer liable for the death.
This is why the Trolley Problem is a factor in autonomous car design and legislation.
surely it should be the driver in the car who should be charged? He was there as a safeguard.
In theory yes but in practice no. The "driver" is only really of use if they have some warning. For any oh shit we need a response now they arent going to be much use.
Will be watching this with great interest. Having read about Arizona relaxing the rules on self-driving cars much further than other states, I wonder if the victim's family might sue the state government.
I do get the impression that "jaywalking" carries much more stigma over there though.
Well jaywalking is an offence in some jurisdictions. Don’t know about that one though. And of course it shouldn’t affect the software design (there are many reasons why a pedestrian or other obstacle might be in the road).
Has there been any context around the nature of the incident though?
All I've seen is 'there's been a death and an automated car is involved'.
Nope, no context that I’ve seen. For all we know she tripped and landed just in front of it
I'm somehow wholly unsurprised it's an uber.
The liability stuff is way more complicated than sue the programmer as there are a lot of options depending on where they are and what might be considered reasonable for the vehicle to have anticipated
e.g. https://en.wikipedia.org/wiki/Autonomous_car_liability
The references section there gives you an idea of just how much work is going in to it.
On that note it did say there was a driver in the car to take control, so assuming the human in charge of the car was paying attention, it may have been an unavoidable accident.. But without more details it's all just conjecture.
16 pedestrians are killed in the states every day by drivered cars, I think we should ban all cars until we can reduce that to zero.
Robot car was busy twitter-botting the Putin election campaign. Unlike its undistractable meatsack counterparts, robot car should've been paying more attention.
Is it just me who thinks these things shouldn't be on the road yet?
Apparently a cyclist..
https://news.sky.com/story/uber-suspends-self-driving-car-testing-after-cyclist-is-killed-11297320
I work in the industry and IMO people have got wwaayy more confidence in this technology than they should have at the moment. Governments all over the world seem to be rushing to hand out licenses to test autonomous cars on the roads to attract the R&D centres, way before the technology is ready. The civil service should be doing work to make sure the legislation is up to scratch and everything is tested properly, but they are hopelessly outmatched in terms of salaries, skills and the sheer pace of change of the companies and technology they are trying to regulate. It seems like they are unable to tell how mature a particular companies technology is. Personally, I wouldn't trust anyone other than Google.
Spot on northwind if you want to keep killing people until you hit perfection then you are missing the point.
in the not too far off future no pedestrians or cyclists will be allowed on roads with autonomous vehicles ‘for their own safety’ and that people will only be able to cross on crossings by law even in the uk.
The UK has these already, we call them motorways and in places dual carriageway, see also tunnels and some bridges. It is a logical step to create higher speed close uninterrupted transit lanes so that cars eta can move quickly without having to deal with merging and pedestrians etc
Is it just me who thinks these things shouldn’t be on the road yet?
In the 2 months since I've been back in the UK I reckon I'm, have seen a couple of thousand drivers that should be off the road before any driverless car.
Is it just me who thinks these things shouldn’t be on the road yet?
With a human driver overseeing things with complete manual override under strict conditions? I don't see a problem with it. To take issue would be the same as blaming ABS or cruise control for having an accident.
It wouldn't surprise me if this was driver error, or pedestrian error, or even a mechanical failure, but I'm reserving judgment.
It's a bit sensationalist just to assume it's the automation aspect of the cars fault when there are so many variables and no specific information has been released.
Google cars could have killed dozens, you don't know about it because they'd remove it from their search results. (said in jest but when you hold the keys to the eyeball castle it's possible to guide opinions)
It's a statistical certainty autonomous cars will kill given enough road time, though it might be a lot less than meatbags do. What level is acceptable to society? If I was investing a billion or two in this tech it's a question I would have already asked.
TBH in autonomous mode you can't really expect the "driver" to take over in an emergency- in some slower emerging thing like a breakdown or contention or something, then sure but in a collision it's just not going to happen, the driver will be switched off and unready.
The bottom line is that real world testing is required to make these things better, because it's only connecting with the countless idiot things that happen in the real world that you'll get a result that works there. In exactly the same way as we instruct learners on the street then let them out when they're only barely capable. Self-driving clearly isn't there yet- but if we don't let it out in the wild then it probably never will be.
TBH in autonomous mode you can’t really expect the “driver” to take over in an emergency
A good point, but in a controlled test, or proof of concept test, which this hopefully was as it was on a real road with real hazards, you'd expect the human in the car to be ready to brake or take evasive action at any second if things didn't look right.
We really need more information on what exactly happened.
Picture in the link further up is pretty heartbreaking.
That picture says nothing, we already know that there was a collision between the car and a pedestrian pushing a bike across the road.
Presumably the car was fully camera'd up and that footage will reveal more of what actually happened.
Interesting angle from the independent article
Until recently, they have required a real person to be sat in the front of the car and ready to take over – but recently California officials approved the testing of such vehicles without humans in the front seats.
I wonder if uber were trying to run before they could walk, considering the first company to market with a viable autonomous vehicle stands to make a (forgive the pun) killing.
It will be interesting to see if the human in the car was actually ready to intervene, or if there even was a human in the car.
Either way it looks very bad for uber, and california I guess.. there's some criminal negligence there for sure.
I wonder whether any of those calling for such cars to be banned as a result of this will manage the logical connection that exactly the same arguments could be used for banning cars with human drivers.
Sad news.
Unfortunately I think the pedestrian will be found to be at fault here, seeing as the police are quoted as saying she wasn't using a pedestrian crossing. This is the US after all where the car is king.
I'm not 100% happy with autonomous cars driving around urban areas until they have done many miles on easier roads like Google has done. Uber have been doing this a lot less so should be restricted to main routes. Not saying it would have prevented this crash but it seems like some tech companies are running way before they can walk.
Listening to someone on Radio 4 the other days and it is clear that laws, governance, insurance policies etc,. have not kept up and all need to change to cater for the questions raised here.
The key one is that the insurance companies need to treat the autonomous car in same was as a person for liability.
Policy could also be void if you have applied latest patch which could catch a lot of people out in future...
in the not too far off future no pedestrians or cyclists will be allowed on roads with autonomous vehicles ‘for their own safety’ and that people will only be able to cross on crossings by law even in the uk.
I recon you will only be allowed to use the road system with some kind of "transponder" which can be pinged for a response of who where what and how fast and what direction. The transponder box will also control the speed of driven powered vehicles so there a no surprises for self driven ones. Having this would be useful helping with things like "blind" overtakes... who gets to go at box junctions, congestion reduction, forming conveys on motorways etc.
Bez's piece seems highly relevant here:
https://singletrackworld.com/wp-content/uploads/2017/09/the-law-will-be-fixed/
forget the law and technology..
who is actually mad enough to get into a coffin on wheels with a bit of software in charge?
Bez’s piece seems highly relevant here
Not really
As per comments under the article. No one is using V2X in any real sense for actual driving, as it's a logistical and technical nightmare. Plus the biggest markets in the world have patchy network coverage and large numbers of people with no access to that level of tech. Not to mention things like wild animals that you don't tag. (not a problem in the UK where the biggest thing most people hit is usually a malnourished badger, but hitting a kangaroo or Elk/Moose is a serious issue.)
Its more in use for things like improving traffic flow (Vehicle to traffic lights is something that's been talked about.)
who is actually mad enough to get into a coffin on wheels with a bit of software in charge?
About 3.5 billion a year as far as i can tell. ;o) And most of them are several thousand metres up.
The thing that interested me about the news coverage was that the victim's prior criminal misdemeanours were being highlighted within hours of her death. Marijuana convictions, complete with mugshots of her looking a bit ropey.
Must be a lot of people with significant sums invested in the 'transport of the future' to need the poor woman to be trashed so comprehensively, so quickly. In truth we have no idea who was to blame for her death.
People will have to get used to the fact, that in some circumstances, even the best autonomous vehicle will run people over and kill them. You would hope that they return better figures than the present arrangement though.
Money is driving the development, not some need for them. Regulators wont be able to regulate fast enough and we will end up with them being pushed onto the roads, ready or not. Money is the driving force here, not a need for them.
We don't live in a perfect world, no software will be perfect and no human will be. In both situations we still don't have a need for software driven cars.
There's some more context here: Tempe police chief says early probe shows no fault by Uber:. Seems like she stepped out in front of the car.
Edit: to remove random garbage put in by the forum software.
16 pedestrians are killed in the states every day by drivered cars, I think we should ban all cars until we can reduce that to zero.
And how many "drivered cars" are there in comparison to driverless ones?
People are far too trusting in this technology IMO.
In reality it will be the first piece of much evidence used to ensure that in the not too far off future no pedestrians or cyclists will be allowed on roads with autonomous vehicles ‘for their own safety’ and that people will only be able to cross on crossings by law even in the uk.
Unlikely, unless you're somehow also going to legislate against kids, animals, fallen trees, and any other type of obstacle that can turn up unannounced on a road. Which is the same reason Bez's article doesn't make much sense.
not a problem in the UK where the biggest thing most people hit is usually a malnourished badger
Deer, horses, cows and sheep cause a lot of damage even potential fatal for those in the car.
This is like the 1986 film Maximum Overdrive all over again.
From the article linked above:
Traveling at 38 mph in a 35 mph zone on Sunday night,
Regardless of whether the death was preventable, if confirmed this seems like a massive and very basic fail by autonomous software.
16 pedestrians are killed in the states every day by drivered cars, I think we should ban all cars until we can reduce that to zero.
Fair point, except autonomous vehicles aren't just being touted as a replacement for cars. https://www.wired.com/story/las-vegas-shuttle-crash-self-driving-autonomous/
you’d expect the human in the car to be ready to brake or take evasive action at any second if things didn’t look right.
I dont think you realise just how many miles are being racked up by autonomous cars in the USA. In May last year Google had done 3 million miles and bearing in mind it took them from 2009 to 2016 to reach 2 million the current total will be far higher and thats just Google. These "tests" are becoming really routine operations now.
In theory the person should remain alert but what do you reckon the chances are of that in reality.
But how many of those miles are trundling up and down the freeway between Silicon Valley and Vegas, easy as pie, and how many are nighttime passes along Mill St, Tempe past the college bars?
In theory the person should remain alert but what do you reckon the chances are of that in reality.
Google were letting staff test the cars - they had to stop the tests because people were falling asleep. Remaining alert just wasn't possible.