'Safety will always come first,' insist Arizona biz org in response to Uber self-driving car death "My greatest fear is that we will hit the winter of AV (autonomous vehicles)," says Jill Sciarappo in response to an unwelcome question at the Intel Capital conference in Arizona – the US state that saw the first self-driving car death last year. Sciarappo is the marketing director of Intel-owned Mobileye, one of the main …
"Arizona is flat, its roads are straight, and new, and well marked, and the climate is dry and sunny. It's the perfect test-bed for getting machines to learn how to drive cars."
Except the rest of the world isn't flat, with straight new, well marked roads where it never rains or snows.
If self driving cars are developed in only these conditions, they'll have to go all the way back to the drawing board to teach them how to cope with curved roads with inclines and less than perfect road markings. Where rain, hail, sleet or snow is a part of evryday life. Heck there may even be salt on the roads or even potholes.
Or something totally random that fell off another vehicle or blew on to the road from elsewhere.
Or something totally random that fell off another vehicle or blew on to the road from elsewhere.
Dogs, cats, duck+chicks, pheasant, hay bale falling off a trailer, sheep, ponies, cow, herd of cows crossing, walkers, cyclists etc.
That's my list from just a minute going over my experiences on British country roads. How on earth would an AI brought up in ideal conditions cope?
That was pretty much my list driving to the feed store and back this morning. There were also a couple goats in with the sheep, a doe and her fawns, a small pack of feral neon-clad bicycleistotards, and a flock of wild turkeys.
(The hay bale fell off in Brocco's parking lot, not the road, but I still had to swerve in the mud to avoid it. Lost traction, which I expected, so recovery was easy. For a human.)
Even when its the same type of hazard you react differently. I live in a rural area and whilst I always slow down for horses If a horse looks nervous, the rider is young or leading a second horse I'll pass at walking pace almost on the other verge or even come to a stop and let the horse pass me. Similarly one local road passes an open meadow if there are any sheep in the field I'll reduce my speed to under 30 mph but if there are lambs in the field or sheep close to the road I'll be going much slower.
Whilst this is a rural problem there are similar issues around passing schools, playgrounds and even driveways. It snot just the visual hazard, it can be data and time specific, around event times or school term and opening/ closing times.
These are very specifc & local hazards and I suspect that AI is a long way from being that granular.
I'll add in
* Cambridge cyclists (who I swear are a different breed to cyclists in the rest of the country)
* country lanes where you have to guess where the road ends and the verge begins
* the A14. Any of it. All of it would be too much of a challenge
* a motorway toll booth. The M25 is much less exciting since they ditched the Dartford tolls
* a ford. Or any road with flowing water on it.
Actually, I'd love to know how any AI autodrive would handle the average British roundabout...
That leaves out all the fatalities that will be caused ONLY by the autonomous driving "AI". What I've always said is that to be accepted by the public autonomous vehicles have to have a 10x lower accident and fatality rate than human driven vehicles. Because humans like a sense of control, and will only be willing to give it up if there is a very large and very definite safety improvement. And because 90% of people believe they are better drivers than 90% of everyone else, so just beating the "average" won't do anything for them.
And they have to be REAL statistics across all conditions, not the cherry picked conditions Tesla's autopilot drives that leads to Musk's buttboys talking about how it is better than human driven fatality rate already. If I only drove your car on freeways in light traffic on sunny days, and left you to drive on foggy snowy nights over mountain passes, my fatalities per mile would be better than yours too even if you are the best driver in the world and I was driving drunk.
"What I've always said is that to be accepted by the public autonomous vehicles have to have a 10x lower accident and fatality rate than human driven vehicles."
Agreed, although I suspect it needs to be zero unless there is an acceptable form of blame and where it can be attributed. If a human driver kills someone, there can be a sense of closure when said human is proven to be the cause. Who is to blame when an AV kills someone? The owner? The manufacture? Someone in the entire chain who produced the software? Some anonymous SourceForge "donor" where the relevant bit of vital code was copied from?
"Who is to blame when an AV kills someone?"
That there is the proverbial hundred dollar question.
Consider this ... What would happen if I were to outfit an automobile with miscellaneous servos, tie them into a computer, and send it for a spin around the block sans human at the wheel. The basics of doing this are pretty easy. But the devil is in the details ... After it kills someone, I guarandamntee that I'd be arrested for murder. Even if I was legally DBA "jake Automotive Intelligence with Blockchain, LLC"[0].
And yet somehow Uber is immune from prosecution? What the fuck?
[0] Note gratuitous use of Blockchain to attract suckersinvestors.
@macjules
"94 per cent of serious vehicle crashes are due to human error. "
So, you mean, that all the bullshit from Road Transport/Safety whatever departments, are feeding us wrong information when they tell us that "speed kills"?
Mind you, everyone knows that speed does not kill at all. It's the sudden stop that does it every time!
Oh, and I've just noticed below someone mentioning "mobility as a service"
Yes, can just imagine it. As you settle back in your seat you will be shown targeted ads.
I notice that only Uber was called out. What about the Tesla "accidents"? Right, not full blown AI-Self Driven yet but the hype has lots of folks believing it is.
And this new "Institute" is already on the side of the cars companies and some of their statements. Smells like their goals will be pre-determined by the industry. . Frankly, be afraid, very afraid.
>>>We may not see cars with synthetic noses any time soon but microphones? It's very possible. Anything that tips the balance when it comes to safety. Aeye's system has also started looking for, and making sense of color, in order to give it an extra sense<<<
Microphones - not sure thats needed, visual processing of a direct image must be far easier than processing a sound thats subject to Doppler shift, echos & every irrelevant noise being made off road. On the other hand it's a great way for parked cars to provide steet level snooping on the public for marketing research purposes.
Colour seem sensible - do these thing spot emergency vehicles approaching and move over to let them pass? then again I'll now wait for the first story of a car swerving to avoid pedestrian of colour A and hitting one of colour B instead.
It will be neither.
We're entering the era of mobility as a service. This can have pro's and con's for service users, but sadly I fear the con's win out long term. Every journey will be charged the maximum the market will bear, varying by time of day, week, month, year, local traffic conditions, demand an a gazillion other factors..
Competition won't help as even without cartels, pricing will naturally tend to the local bearable maxima. Sure there might be some initial loss-leader effects as newcomers try to undercut and kill off the established players (e.g. bus, taxi), but once those are dealt with watch the prices jump.
Fleet sizes will cover the 80-90% average use requirement at best. Service providers certainly will not have extra vehicles sitting unused just in case they're needed. That's poor resource management. They will just surge-price during peak demand, discouraging enough customers to keep service within acceptable wait times for those still willing/able to pay the exta.
"We're entering the era of mobility as a service."
It sounds great until you think about a few scenarios such as getting to work and back at the same time as everybody else. Getting a car that just transported somebody with the flu or plague or something. I like to keep a few things in the car such as a fleece, jumper and basic shirt, just in case, an umbrella, some bottled water, etc etc ... I really should sort out all of the junk in the car this weekend. Won't it be nice to have to (by law) remove and replace the child seat a half dozen times a day in all sorts of different makes of cars? It will also be great to have done your weekly shop only to find out that the mobility service is "experiencing technical difficulties at this time, please try back later" or your card that's billed for service is declined as you stand outside ready to be picked up for the trip home (with an ice cream cake for the kid's party tomorrow).
Public transport is a good thing. I'd rather take the train to visit my mum than to drive, but sometimes schedules don't line up or I need to pickup/drop off stuff that would be problematic to drag around on various busses, the Tube and trains. I do find it odd that many times it's more expensive to take the train than what it costs in petrol. I'm in the US right now, so that makes a big difference. The trains take longer and cost more than paying for petrol and doing the driving myself. Forget a "roomette". I can fly for far less and get a free feel up in the process. If they'd just pull my hair a bit...........
"How would we all change our habits if travel was readily available and almost free?"
[Looks down at bare feet, glances over to hiking shoes by the door, scratches head]
Oh, he said travel. I don't see airline, train, or cruise ship tickets being anywhere near free in my lifetime. These kids today sure dream big, don't they.
"How would we all change our habits if travel was readily available and almost free?"
We could for example, keep all our cars out driving on the street the whole time, hoping to capture a new fare. While using fuel and blocking traffic. (As uber is already doing in some city centers).
We could slow down all our cars, slowing down traffic, to optimize profits by optimizing fuel economy.
We could have unknown effects on urban density and urban sprawl, as all other transport revolutions have done..
What else? Dunno, but no reason to expect it all to be good.
Quick review: how does nearly every so-called tech firm design and market products these days? They knock together (or copy) a widget, then race it to market as quickly as possible. QA and security take a back seat, if they were even invited for the ride. They PLAN to let the customers find problems and fix them later. It's a crappy business model, and gives predictable results.
So here we are! Self-driving tech was rushed out with known problems and limitations, no security, and insufficient testing. But where a dodgy e-lock will only inconvenience you, the AI cars have killed people. OOPS that's harder to ignore, whoa, backlash!
Self-driving tech NEEDS to be frozen for a while until it has been proven safe and reliable. Public roadways are no place to experiment with 3500-pound kinetic energy weapons.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
