[RM7]

Originally Posted by glennQNYC

Geez- I don't agree with this statement at all! I can think of countless ways machines can't come close to human ability.

Well, good for you, but for situations that require a human to repeatedly perform a certain task, the number of errors is vastly greater than a machine.

[MightyMouseTech]

Originally Posted by RABAUKE

Shouldn't a "self driving car" have sensors to detect objects moving into it's path rather than waiting for the object to be directly in front of it?

This tech is still new and unproven, the police chief has said the pedestrian is at fault which is often the case, but I suspect this investigation is far from over, since the driver appearently wasn't watching the road, what about the manufacturer for not having enough sensors or the programmer for not having the sufficient safeguards built into the software.

I know what the existing laws say about pedestrians etc as have been earlier posted but we're entering new territory here and I suspect that lawyers will be having a field day with this regardless of what the law says. The rules and law are black and white, until lawyers and courts start to interpret them.

Just my two cents folks.

After seeing the video, I am surprised the LIDAR did not pick up the pedestrian.

[GuidoK]

Originally Posted by nick3753

That's who is truly at fault and paid the ultimate price for that decision. Not to mention she would certainly see the cars headlights before any car or person would see her.

Thats the fault in Arizona. But the implications are far greater.
Uber is doing this milliondollar research (probably 100+million dollar research, as they ordered 24000 XC90's at volvo alone, not counting the tech, personell and companies they bought to develop this tech...) not for using autonomous driving in Arizona, and not even for the US alone. They want to become world leader in autonomous driving transport. And in a lot of places a pedestrian has more legal protection. (where I live, when hitting a pedestrian or cyclist while driving a car is at the start always 50% your fault as pedestrians and cyclists are considered weak traffic participants and motorists arent. (the difference between a motorist and a cyclist or pedestrian is also that the motorist is a schooled traffic participant), so that gives extra responsibility to motorists. Who's 'truely' at fault is only important for this incident, but not for the globalized process of implementing autonomous driving imho.

Also who is really/truely at fault is not clear to me. I mean the car WAS speeding from what I understand of all the info around this. (how that car could be speeding is beyond my comprehention, its dead easy to program a car to follow the maximum speed limit within 0,1mph based on gps data alone, so if that car was speeding, it would be a deliberate programming decision)
And the pedestrian was hit on the front right of the car, so she was almost on the other side of the road. Depending on how fast she was moving with her bike and stuff, so how long she had been crossing that road, it could be the case that if the car wasnt speeding, she would have reached the other side of the road in time. You can wonder if someone can estimate or calculate a crossing of a road on that level of accuracy, but thats not the point here if the issue is who's 'truely' at fault. If there's a lawsuit, there will be experts that will be pointing out this issue.

What also is important is that the car didnt stop where experts say that it should have stopped based on the sensorarrays and hardware that the car is equipped with. That would mean its likely a software error. Software is very much an intellectual product, so its very difficult to measure software performance as in this case software has to interpret something. If hardware cant detect something, its due to physical parameters, like the range has to be extended or light sensitivity has to be increased to a certain lumen or other physical parameter. But interpretation performance is a more subjective scale.

[Dog Face Pony Soldier]

Originally Posted by JamesNoBrakes

Well, good for you, but for situations that require a human to repeatedly perform a certain task, the number of errors is vastly greater than a machine.

I don't think anyone would define driving as "repeatedly performing a task." Too many variables. In any case- It sounds like you're backing away from humans being bad "at pretty much everything." Thanks for agreeing.

[RABAUKE]

Originally Posted by nick3753

Should a self driving car be able to avoid this? Yes, but it's still in development. Should a human driver be able to avoid? Maybe. Should you cross 3 (presumably 6 counting the opposite direction) lanes of traffic at 10 at night while wearing dark clothes, pushing a bike on what appears to be a highway? No. That's the one person/thing that could have avoided this with certainty. That's who is truly at fault and paid the ultimate price for that decision. Not to mention she would certainly see the cars headlights before any car or person would see her.

Of course the pedestrian owns some of the responsibility , but your premise falls short if the pedestrian was a child running out for a ball and the tech didn't intervene.

[MalibuBimmer]

Originally Posted by glennQNYC

Actually the report I linked to above addressed this exact question. Their findings are different than your opinion. They found that consumer expectations are that autonomous vehicles are far safer than a human driver. You can even see that flawed viewpoint expressed right in this thread! The report warns that manufacturers need to set realistic expectations for autonomous driving systems to help curb lawsuits.

From the report: [car makers promise accident-free vehicles]

I don't think the report addresses the philosophical question I raised which is the level of proficiency leading to acceptance.

What's interesting in the quotes you've provided from manufacturers (for some reason I can't add them to my post) is that manufacturers see the trap as well. Autonomy will never be able to prevent all accidents, but in the future when we have true autonomy when there are accidents or collisions automobile manufacturers rather than drivers will be on the chopping block. So manufacturers will want to push the myth that their cars are 100% accident free. If an accident occurs, it's an Act of God.

[IK6SPEED]

Originally Posted by glennQNYC

You're just full of compassion and empathy aren't you?

Reread.

That is the Drivers mouth

Replying to poster who claimed driver was on phone.

[IK6SPEED]

Originally Posted by och

You can see THREE street lights in that shot in the immediate proximity, and looks like there are two more right up ahead. She was crossing in a very well lit area, there were no shadow. There are no larger objects there blocking the light and casting shadows whatsoever. Uber used a terrible dashcam that doesn't have good contrast, but human eye would easily see her. You better believe they will have camera experts in court that will testify to this, and they will shot footage with better cameras on the same road as well as have experts recreate the scene of the accident and test visibility.

So you are calling the Sherriff investigating a liar?

I believe both officers have a better idea of the accident scene than you or I.

You continue to fail to acknowledge victim dressing in dark clothes hide her even more.

There will never be a trial.

Officials will not charge and homeless victim has no one that can claim a loss for civil court, so your statements are false speculation.

[wdb]

I just watched the video. The car's automation was primarily at fault; the backup driver, secondarily at fault. The pedestrian had a bike and was in the street for many many seconds before the impact. There were no obstacles, no visual blockages, no nothing. A human with a bike crossing a road.

Bad robot.

[aozer]

Originally Posted by 1MOREMOD

That would teach you even quicker lesson.

what lesson do you learn when someone you love is killed by a distracted driver?

[izzyM2]

Originally Posted by wdb

I just watched the video. The car's automation was primarily at fault; the backup driver, secondarily at fault. The pedestrian had a bike and was in the street for many many seconds before the impact. There were no obstacles, no visual blockages, no nothing. A human with a bike crossing a road.

Bad robot.

The bicycle and rider emerged from a dark shadowed area until it was too late. I bet that’s why the car did not identify it. The driver can be deemed at fault for not keeping her eye on the road but isn’t that the whole point of having a driverless car? We can also argue that the pedestrian was also faulty for not being in a pedestrian lane but that is why we are suppose to drive defensively to try not to worsen other people’s mistakes.

[IK6SPEED]

Originally Posted by GuidoK

Thats the fault in Arizona. But the implications are far greater.
Uber is doing this milliondollar research (probably 100+million dollar research, as they ordered 24000 XC90's at volvo alone, not counting the tech, personell and companies they bought to develop this tech...) not for using autonomous driving in Arizona, and not even for the US alone. They want to become world leader in autonomous driving transport. And in a lot of places a pedestrian has more legal protection. (where I live, when hitting a pedestrian or cyclist while driving a car is at the start always 50% your fault as pedestrians and cyclists are considered weak traffic participants and motorists arent. (the difference between a motorist and a cyclist or pedestrian is also that the motorist is a schooled traffic participant), so that gives extra responsibility to motorists. Who's 'truely' at fault is only important for this incident, but not for the globalized process of implementing autonomous driving imho.

Also who is really/truely at fault is not clear to me. I mean the car WAS speeding from what I understand of all the info around this. (how that car could be speeding is beyond my comprehention, its dead easy to program a car to follow the maximum speed limit within 0,1mph based on gps data alone, so if that car was speeding, it would be a deliberate programming decision)
And the pedestrian was hit on the front right of the car, so she was almost on the other side of the road. Depending on how fast she was moving with her bike and stuff, so how long she had been crossing that road, it could be the case that if the car wasnt speeding, she would have reached the other side of the road in time. You can wonder if someone can estimate or calculate a crossing of a road on that level of accuracy, but thats not the point here if the issue is who's 'truely' at fault. If there's a lawsuit, there will be experts that will be pointing out this issue.

What also is important is that the car didnt stop where experts say that it should have stopped based on the sensorarrays and hardware that the car is equipped with. That would mean its likely a software error. Software is very much an intellectual product, so its very difficult to measure software performance as in this case software has to interpret something. If hardware cant detect something, its due to physical parameters, like the range has to be extended or light sensitivity has to be increased to a certain lumen or other physical parameter. But interpretation performance is a more subjective scale.

I have read more on this last night than most here I suspect. Not comment blogs but LOCAL news accounts from the local papers and TV stations etc.

The Phoenix New Times States there is a 45 MPH about 1/4 mile away on Mills Avenue. Personally, I do not see how a Dual Lane limited Access road such as that with a median is 35 MPH either.

Considering the Phoenix New Times actually went out and found this speed limit sign and states as such, perhaps a reduced speed limit sign to 35 MPH is missing, if that is indeed the Speed Limit.

The car most likely read the last speed limit sign on the road (45 MPH) and was driving at 38 MPH accordingly or 2) was slowing from a 45 to MPH zone to 35.

In most States, 5 MPH isn’t chargeable to err on the side of caution.

But regardless, 3 MPH would have made no difference here.

[IK6SPEED]

Originally Posted by aozer

what lesson do you learn when someone you love is killed by a distracted driver?

As she was homeless, not sure how much she was someone’s loved one, but I get your point.

1) cross in crosswalks

2) don’t expect a car to stop for you as you don’t have the right a way.

3) wear bright clothing at night. Reflective vest as well

4) have a light and reflector on your bike as required by law at night.

[mkoesel]

Irrespective of who is at fault here, I think (hope) we can agree that to be considered ready for mass consumption, the technology should mature to a point where an incident like this can be avoided. As noted by the gentleman from EEVblog, the sensors that the vehicle is equipped with should theoretically have been able to detect the pedestrian even though she was not visible to the human eye until an instant before the collision. It is an unfortunate situation, but it will be studied and result in a safer product.

[IK6SPEED]

Originally Posted by RABAUKE

Of course the pedestrian owns some of the responsibility , but your premise falls short if the pedestrian was a child running out for a ball and the tech didn't intervene.

Parents would bear responsibility there for letting kid play with ball or simply playing outside at 10PM.

Of course, the kid would most likely have been in residential area at 10PM as well

[RickFLM4]

Originally Posted by MalibuBimmer

I don't think the report addresses the philosophical question I raised which is the level of proficiency leading to acceptance.

What's interesting in the quotes you've provided from manufacturers (for some reason I can't add them to my post) is that manufacturers see the trap as well. Autonomy will never be able to prevent all accidents, but in the future when we have true autonomy when there are accidents or collisions automobile manufacturers rather than drivers will be on the chopping block. So manufacturers will want to push the myth that their cars are 100% accident free. If an accident occurs, it's an Act of God.

I think the debate in this thread is about whether the technology malfunctioned in a way that a human might not. Some think there was no avoiding a collision regardless of who or what was driving. Others aren't so sure and suggest perhaps an attentive human driver could have avoided the crash. That point is debatable. Unless I missed it, I don't think anyone suggested that there is clearly no way a human could have avoided the crash, but the autonomous car clearly should have avoided it nonetheless.

With that said, I think we would all agree that if the safety standard is equal to that of a human driver, it needs to be set to an attentive human driver following traffic laws, not the average driver. Even then, to your point, that isn't much of a "selling point" to the public. Most logically expect an improvement in safety based on hundreds of billions in R&D and infrastructure needs to support safe autonomous driving, as well as societal changes that will result from autonomous vehicles. There needs to be a reason to support change and human nature is to avoid loss of control unless it is pretty clearly safer.

[IK6SPEED]

Originally Posted by och

Did you watch the video showing the driver staring down at her phone and not watching the road?

You have no evidence Driver was looking at phone and not at Vehicle Diagnostic screen in the self driving car, but continue to post this.

Regardless, without the video, what the driver was doing would be unknown.

In a normal situation, she could just claim she didn’t see victim as victim was dressed in dark clothes and darted out in front of her in an area one would not expect a pedestrian.

[IK6SPEED]

Originally Posted by wdb

I just watched the video. The car's automation was primarily at fault; the backup driver, secondarily at fault. The pedestrian had a bike and was in the street for many many seconds before the impact. There were no obstacles, no visual blockages, no nothing. A human with a bike crossing a road.

Bad robot.

On why the car did not slow down and try to avoid yes, but that’s not a crime. Not even a “fault” in legal terms.

The primary fault was the victims violation of the Arizona law.

[RickFLM4]

Originally Posted by IK6SPEED

You have no evidence Driver was looking at phone and not at Vehicle Diagnostic screen in the self driving car, but continue to post this.

Regardless, without the video, what the driver was doing would be unknown.

In a normal situation, she could just claim she didnÂ’t see victim as victim was dressed in dark clothes and darted out in front of her in an area one would not expect a pedestrian.

Do you know if Uber's protocol allows drivers to view diagnostics on a laptop, tablet or screen instead of viewing the road? (I don't know the answer but seems a little odd to have a policy that permits drivers to take their eyes off the road for that long if intent is the be a safety net as technology is tested.)

[RickFLM4]

Originally Posted by IK6SPEED

On why the car did not slow down and try to avoid yes, but that’s not a crime. Not even a “fault” in legal terms.

The primary fault was the victims violation of the Arizona law.

Nevertheless...

[Dog Face Pony Soldier]

Originally Posted by RickFLM4

....seems a little odd to have a policy that permits drivers to take their eyes off the road for that long if intent is the be a safety net as technology is tested.

That is a point that the fanboys are willfully ignoring. THE major part of agreeing to let automated vehicles test on public roads was that there be a human driver in place to intercede in case the system fails. Now we have a fatal example of the system failing and we find that the human was not paying attention. The driver admitted that they were looking down; and hitting the pedestrian was what brought the entire situation to attention. Why wouldn't the driver be held responsible for not paying attention? That was the agreement. A pedestrian jaywalking and/or not dressing like somebody wishes they had, doesn't absolve a driver from blame when their vehicle kills someone.

[IK6SPEED]

Originally Posted by RickFLM4

Do you know if Uber's protocol allows drivers to view diagnostics on a laptop, tablet or screen instead of viewing the road? (I don't know the answer but seems a little odd to have a policy that permits drivers to take their eyes off the road for that long if intent is the be a safety net as technology is tested.)

I do not know Policy.

I suspect they should be able to look at the diagnostic screens as human driver is only there for backup in autonomous mode.

Originally Posted by RickFLM4

Nevertheless...

Reread the statue you posted.

Then read statements from investigators and Sherrif.

The driver could not have prevented.

Originally Posted by glennQNYC

That is a point that the fanboys are willfully ignoring. THE major part of agreeing to let automated vehicles test on public roads was that there be a human driver in place to intercede in case the system fails. Now we have a fatal example of the system failing and we find that the human was not paying attention. The driver admitted that they were looking down; and hitting the pedestrian was what brought the entire situation to attention. Why wouldn't the driver be held responsible for not paying attention? That was the agreement. A pedestrian jaywalking and/or not dressing like somebody wishes they had, doesn't absolve a driver from blame when their vehicle kills someone.

No you are willfully ignoring Arizona Laws concerning Automated Car Testing.

Please research what Arizona’s law permitting Driverless Car testing actually states, especially the Governor’s Executive Order on March 1st of this year.

The way it is written, car is driver, so the Company would need to be charged as they were/are driver.

Also, Human has to give car time to see how it is reacting. Otherwise, no point to tests.

Human is backup.