merlinextraligh

There's a distinction between negligence and willfulness. And saying someone is at fault for an accident isn't casting moral dispersion, but it is saying accidents have causes, those causes are identifiable, and the vast majority of collisions occur because of some human error somewhere in the process. To say that virtually every accident is the result of someone, or multiple someone's failure to exercise ordinary care is not to say that every accident was willful.

Give me one example of an accident, not involving a wild animal that was not the result of someone's failure to exercise reasonable care, whether that someone was a driver a pedestrian, a highway designer, a highway worker, a product designer, a manufacturer, a mechanic, etc.

Because humans make errors of course it is impossible to eliminate all accidents. But that does not mean that any particular accident was unavoidable.

We need to stop thinking of accidents as unavoidable, and identify causes, and correct them.

It used to be a defendant in an auto collision case could get an instruction that some accidents are unavoidable; the unavoidable accident doctrine however has now been rejected in many states.

I would suggest that a culture that thinks of auto collisions as unavoidable is part of what leads people to think its not a problem when a driver runs over a cyclist with the I looked but didn't see them excuse.

merlinextraligh

we don't know at this point what the human operator was doing, what they saw, or how they reacted.

This does point to one huge problem as we phase into automation. "Auto Mode" does take away from the driver's ability to react. It is extremely difficult for a human operator to remain attentive to a vehicle being operated autonomously. It's a major issue as airplanes become more autonomous.

So the idea that a human operator can serve as a backup for automation is highly questionable.

I-Like-To-Bike

Has there been any information, video, or data released about the victim's actions at the time of the incident that was not furnished and possibly edited by Uber? Was the so-called police report/statement just a re-write of information that was provided by Uber?

"Uber is assisting and this is still an active investigation," Liliana Duran, a Tempe police spokeswoman, said in an emailed statement.
https://www.bloomberg.com/news/artic...ash-in-arizona

Kontact 

I've investigated multiple aircraft mishaps with two pilots and flight crew. So I think I have a certain perspective on how multiple people interact in ways that produce errors and how it is difficult to assign fault in a definitive way to one party.

And that's all fine for a military plane crash, but it doesn't work in the binary world of insurance driven accident investigation, because someone has to pay the bill.


Most accidents do have clear fault. But plenty do not, and it is a logical trap to say that every one must have a clear faulty party. If I slip on icy pavement before anyone could salt it, am I 100% at fault? Is my boss at fault for having me come in at 5am before anyone could clear the sidewalk? Is someone used to listening for oncoming cars at fault when they step out in front of an electric car? Is the silent car maker at fault?

It is more complicated than distilling every mishap down to a singular cause for the purposes of an insurance settlement.


People are going to be pretty unhappy when the data from this accident shows that Uber's robot did the best it could and the human driver did the best he could, but the pedestrian got hit anyway. It won't fit into the narrative many would prefer.

InOmaha

As I've stated, I believe that autonomous cars are coming. I only disagree with the time frame some are putting out. The tech will have to be very close to 100% right from the start because the second someone gets into a self driving car they'll be on their phone. Because that's the point of a self driving car, not having to drive. And a large percentage are on their phones when they're required to drive; they won't pay more attention if someone else is doing the driving.


As to the question in the OP, if you break the law on a bicycle you may confuse the car, or exist outside it's sensor capability and programing parameters, and get hit by it. You should treat the first iteration or two as unproven tech with inattentive backup drivers. Basically ride your bike as you do now, assuming the car doesn't see you and could hurt you. Partially because the tech is unproven and partially because there will be a significant portion of regular cars on the roads for years or decades.


The lower the median income of an area, the more likely you'll find older and lower tech cars being driven.

avole

1.) There can only be one driver at any one time.
2.) You'd need to hire several fleets of buses to accommodate all the people you describe as drivers in your post.

Mind you, I'd love to go along to the airline in which I have shares and be allowed to drive an airplane (airplane in American) from here to, say, Samoa, especially now in this current cold snap.


Edit: the above sentence was written before I had seen the previous post, Contact. I'm not denying your ability in your work, merely the logic of your reply, which, basically, lacked some.

Kontact 

That's simply not true. The robot was actively driving, the driver was standing by to override the robot. That's two people who are in control of the vehicle. And we all know that the first reaction of anyone who is sitting behind the wheel and sees a person in front of the car is to stamp on the brake pedal, auto mode or not.

Anyone who has been a flight instructor in a dual control aircraft can tell you that more than one person can "drive" at a given time. If that wasn't the case, very few first time student pilots would survive.

Kontact 

Didn't you already insist that there will be a cover up? Where did you get that information?

CliffordK

Exactly.

Auto mode would have slowed the human co-pilot's reaction times.

As I mentioned, the appropriate response to seeing a person next to the road is to give the person a little bigger buffer. But, would the human have taken control of the vehicle and moved over or slowed down as a preventative measure?

How aware was the human driver of the entire situation? Did the driver know where all the other cars were on the road? This would be vital to safely execute an avoidance maneuver.

Kontact 

I think it is a stretch to say that a test driver of a new technology is so convinced of the perfection of the prototype that he was completely lulled into inaction.

Was this how you all recall your driving instructors? Sitting in the other seat and passively hoping you'd stop in time?

avole

Er - condemned by your own words ! The robot was driving, not the overseer.

If people accepted your definition, we could all sue back seat drivers after an accident, using the shared responsibility argument.

merlinextraligh

It's not a binary world. Most states have adopted comparative negligence, so there is ample opportunity to assign percentages of responsibility to multiple parties.

And to say there is a cause of an incident that ties to someone's fault, is not to say someone has to pay for every incident.

In your ice analogy, the fault is highly likely to be your own, in failing to look out for ice, for choice of foot ware, or in making the decision to walk in the conditions in the first place.

But your ice analogy is also a red herring. You chose to walk in icy conditions, you assume that risk, and do little to put others at danger. You operate a two ton vehicle on a public road, you take on a responsibility and a duty of care to others, it's not good enough to say sometimes accidents just happen.

BTW, none of this is an argument against autonomous vehicles. At some point they very likely will reduce accidents, when that is remains in question.

It is an argument for getting to the bottom of what happened here, and not just writing off a percentage of clllisons as unavoidable.

indyfabz

And the computer driver wouldn't be looking down at its phone to text.

Kontact 

Back seat drivers don't have steering wheels and brakes, nor do they get payed to oversee another driver.

The vehicle was a dual control system with a human in overall charge. You're giving the copilot complete responsibility when you wouldn't say the same of a dual control airplane with a pilot in command and a copilot.

Kontact 

And that's the red herring. No one is uninterested in getting to the bottom of what happened here.

merlinextraligh

You've investigated airplane accidents, so I'm assuming you're aware of the problem with pilots being lulled into complacency, by autopilot as planes get more computerized.

Not at all a stretch to believe that Uber cars are getting to the point that the operators are lulled into complacency

If the car operates seamlessly for 8 hours a day, and you never have to intervene, it's just human nature that you may take your eye off the road for a few seconds every now and then.

Again its speculation regarding this incident; but it is a major issue in going to autonomous vehicles, and raises the question whether a level 5 autonomous car, without any human controls may actually be safer than a car with human backup.

Kontact 

Sure, but autopiloted airplanes are decades old, and autopiloted cars are brand new and only being operated by employees of the development company. I'm just pointing out that the test driver of this car is just as likely to be paranoid and overly suspicious of the robot as he is to be complacent, just as the first autopilot testers in aircraft likely were.


AND, this is assuming that the test driver was complacent AND the robot screwed up - which means that not one but two drivers had to have both done the wrong thing.

avole

Sorry mate. You know as well as I do that was game set and match .

Swallow hard, tomorrow is another day.

jefnvk

Neither Uber nor the autonomous technology is likely at fault according to police. The lady stepped out suddenly enough even the human driver didn't catch it until she was already hit: https://www.sfchronicle.com/business...e-12765481.php

Actually, the opposite. Once AVs become somewhat commonplace, it will become more expensive to insure human driven vehicles, the way you get discounts now for having safety features on your car.

And yeah, that is an incredibly deceptive and very likely to be misused median design

Maelochs

jefnvk posted this in another thread---https://www.sfchronicle.com/business/article/Exclusive-Tempe-police-chief-says-early-probe-12765481.php

in case anyone cares about facts ... i know they slow down the ideological debate, so maybe just ignore this.

genec

Suicide by car. Of course this whole situation will be examined down to the nth degree.

What I want to know is why was the car going 38 in a 35mph zone... sheesh.

Maelochs

Yup. Anyone who thinks this is more than a minor hiccough in the roll-out and eventual takeover on at least 75 percent of the roads is flat wrong.

Humans are ridiculously bad at driving cars. So far AVs have what, six million fatality-free miles? And this one fatality, it turns out, is almost certainly entirely the fault of the pedestrian.

I think any driver in an urban or suburban environment who wants to pilot his/her own vehicle is going to be paying a stiff premium not too many years from now. Also, I wouldn't be surprised in penalties for accidents caused by human drivers go way up at about the same time.

jefnvk

Actually, I am curious about that. We generally have a 5kph (3mph) tolerance from vehicle manufacturers as to what indicated speed is v high accuracy GPS, so I also wonder what specific speed signal they were using to measure this (there are a lot in cars). Or, is the AI programmed to drive as a person would, a bit over the limit? Or did it just have the wrong speed in its maps?

InOmaha

What happens the first time 2 AV vehicles are in an accident and there are no driver's to blame? Do the insurance companies go after the vehicle manufacturers?


Statistically speaking, there will be accidents involving 2 AVs, if for no other reason, then due to incompatible software algorithms from competing manufacturers.



Will car owners take on the insurance liability for their car's software?

Kontact 

Your experience speaks for itself.