Uber Self-drive Fatality

I think they will come. You only need to look at te ‘electric is king’ brigade, and thus insistence of removed combustion engines from the road (cars mainly). The fact that 0.5% of new cars registered last year were electric, or the issue of built up areas and flats and charging points never mind remote living, seems to be conveniently ignored.

Same with AI. States have self driving trucks, where the link to the truck in front. All have a driver now, but they do work like a train of carriages. What happens if a moose runs out? Accident between the gaps? Etc etc aren’t factored in because they can’t be: who sets the moral rules of ‘those in that car die, or I may get injured avoiding and going up the embankment’? Are these set at the factory (China laws apply?) or at local govt level (what if you travel abroad/cross borders?). So much to consider.

 

I think thankfully, it will be a very long time before we see a truly autonomous vehicle for the public to go out and buy. There are just to many circumstances whereby a squillion algorithms would be needed. I saw a video recently whereby a simulation with the latest Google vehicle was given the relatively simple choice of animals on the street or a mother and her pram on the pavement, and the car chose to avoid both by swerving to the other lane in the path of an oncoming truck !! There are just too many circumstances where AI does not have the intelligence to make a decision. I can't really think of any circumstance where there will not be a huge risk, and personally I would never entertain the idea for my own vehicle. My lass has a Golf with radar guided cruise, and even that makes me feel a little uneasy, as it just does not do what I'd do, cause its too dim witted to read the road, and so brakes at times when I would have rolled off some time earlier and avoided the need to brake in the first place. Technology has its place for sure, but putting it in charge of a car ain't one of them IMO.

 

No machine will ever come close.
It isn't just the thinking mind. Its unconscious instinct and reflexes. When you drive a car (or ride a bike) it almost becomes an extension of your body. You work the controls often without making conscious decisions. Its incredibly difficult to maintain that spontaneity and connection - and therefore reaction time - if you're not actually physically connected: if you'r not driving the thing.

Try teaching a complete beginner to drive - someone who doesn't even know how to work the pedals and pull away or change gear. Its really difficult. Teaching someone the rules of the road is simply enough but its very hard to break down and put into words physical actions that for you have become reflexive through muscle memory. If you have to stop and think about the physical inputs needed to drive a car, you can never make those decisions as quickly (or as accurately) as your muscle memory can do it for you. And that's for someone who already knows how to drive. What if people in the future never acquire driving skills in the first place, or their skills are dumbed-downed and their reactions slowed because of a reliance on driverless technology? They won't be capable of over-riding the technology.

 

I think it’s coming, it’s just a matter of time. As Bradders had said, it didn’t take long for the facts to start coming out about the incident. Woman crossing a very dark highway, having come out of a set of trees/bushes and wearing no reflective clothing or lights. It didn’t really matter who or what was in control, sadly the woman made the wrong choice to cross where she did.

One area of good we can all see immediately, is that we have live telemetry and video of the whole incident. In terms of accident analysis, it’s the jackpot. If that were the case with every car life would be much easier. In addition to this, the results will be passed to every self driving car out there so they all are made better by this. That’s not the case with human drivers who repeatedly make the same mistakes as their counterparts.

When the automobile started being used, deaths occurred. I beleive a woman was hit at 7mph and died. Unfortunately it happens but thankfully back then we were a bit more sensible about things and life went on. Now we’re expected to abandon the whole show because we live in a world of Internet outrage. This in one death that was not the fault of the car. One death in many hundreds of thousands of miles.

 

There's also the reliability of the systems to consider, the vast majority of vehicle breakdowns are computer or censor related failings. Electronics have a long way to go before they can be relied on to totally control a driverless vehicle.
Steve

 

True. The most reliable internal combustion-engine vehicle ever made is a side valve BSA M20. Cruder and with less power than the average Briggs and Stratton lawn mower - and barely more self-propelled - they'll do a million miles between oil changes and only wear out because of the entropy and the arrow of time. And rust.
Add an ECU and a computerised management system and it's catastrophic failure and certain bankruptcy within six months.

 

i dont see any issues with reliability, most of the tech is already incorporated into modern cars, adaptive cruse control,, fully electric steering, hill holder, its only really the software thats new.

 

I cannot remember where but I saw a film of a test of a self driving car being presented with a choice 1. crash head on into a wall of foam blocks or 2. swerve into the path of an oncoming motorcyclist.
Guess what the AI car did?

 

Yes, early ones made poor choices but those were early test models in the same way we made poor choices as teens. Like us, they learn from those deliberate test scenarios.

In the early stages, they were basically told not to crash. It would therefore see the wall as it’s primary danger and avoid it whatever. We’ve moved way beyond this now.

You don’t really think a pilot flies you on holiday do you? That’s been computer for a while (obviously less risk up there but it’s trusted to behave fine and seems reliable enough to be trusted)

 

The idustry assumption is that the motor manufacturers will pick up insurance costs potentially for 3rd party claims, as the system should be crash-proof, so an owner only looks at theft cost. But, equally as ridiculous (!) the industry also suggests no one will own cars, we will summon one that will be there in a few minutes, take us where we want and be owned by a group of people or companies.

Sounds like taxis

 

What hasn't been discussed, is the number of fatalities/accidents per driven distance. If you have statistics that self driving cars are having 10 times less fatal accidents than humans per driven mile, then the roads will be safer and we should embrace the self driving cars.

The difficult question remains, how to deal with fatalities that _will_ occur anyway due to self driving cars. There are always fatalities, be it human or computer driven. It's just more difficult to accept the ones caused by the computers.

In the case in hand, I read somewhere that there might have been a lidar swiched off or something. There is no reason that Volvo hit the person, as even some regular volvos have already the emergency braking if they detect an object in front. There is something that we are not told about the system/testing/circumstance.

 

Sounds like a hideous world where free will has been removed and we make our choices in life from a list of pre-prepared options compiled and controlled by billion dollar corporations with the collusion of authoritarian big government, and because we're doing it on our phones or at home on our computers with a coffee in our hands, we kid ourselves that we're in control and we're making choices we've arrived at by ourselves.

 

it's all about "economics" at the end of the day without these new ideas the world would crash,so you have to keep on reinventing the wheel,,but i still don't like it