Author Topic: First pedestrian death by autonomous car...  (Read 27231 times)

0 Members and 1 Guest are viewing this topic.

Online nctnico

  • Super Contributor
  • ***
  • Posts: 26752
  • Country: nl
    • NCT Developments
Re: First pedestrian death by autonomous car...
« Reply #75 on: March 22, 2018, 10:59:00 pm »
And to get an idea how far away she is in the crap video....

AZ center line striping appears to be 6" wide strip 10' long with 30' spacing.  So she's got to be around 60' away when her shoes appear in the grainy video.
60 feet is 20 meters. The camera lense makes it look much further than it is in reality. With the car travelling at 40mph=17.8m/s means the time from 'visual' to 'impact' is little over 1 second. A human driver won't be able to do anything in that time anyway.
« Last Edit: March 22, 2018, 11:00:57 pm by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline mdijkens

  • Regular Contributor
  • *
  • Posts: 146
  • Country: nl
Re: First pedestrian death by autonomous car...
« Reply #76 on: March 22, 2018, 11:06:55 pm »
probably haven't seen better video in the other thread of the lighting situation there:

Most dash cameras do a pretty good job at night. This Uber one seems to have done a rather poor job.

Uber must have used a potato instead of a dashcam.

Below a footage decent dashcam at the same spot at night.

https://youtu.be/CRW0q8i3u6E?t=30
 

Online nctnico

  • Super Contributor
  • ***
  • Posts: 26752
  • Country: nl
    • NCT Developments
Re: First pedestrian death by autonomous car...
« Reply #77 on: March 22, 2018, 11:21:37 pm »
probably haven't seen better video in the other thread of the lighting situation there:

Most dash cameras do a pretty good job at night. This Uber one seems to have done a rather poor job.

Uber must have used a potato instead of a dashcam.

Below a footage decent dashcam at the same spot at night.

https://youtu.be/CRW0q8i3u6E?t=30
Even if a better camera would be able to 'see' the pedestrian from twice as far away that still leaves little over 2 seconds to react. Assuming the LIDAR has a similar range. Before anything can be done the direction and speed of the 'object' needs to be established. Meanwhile time is ticking away. Neither human or autonomous driver can do much in these kind of situations where the time to react is extremely short. One thing to keep in mind is that in hindsight it is easy to say the car could have moved left or right. But if you have ever been in a similar situation then you'd know that swarfing to the left or right needs time to check in order not to create an even bigger accident. If you only have a couple of seconds to check AND do something the time simply isn't enough.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline ez24

  • Super Contributor
  • ***
  • Posts: 3082
  • Country: us
  • L.D.A.
Re: First pedestrian death by autonomous car...
« Reply #78 on: March 22, 2018, 11:35:42 pm »
Who is at fault? In my opinion 50% the pedestrian, 25% the human driver, 25% technology.

My guess is since Uber hires felons to test their cars, they put in their contract that the drivers are 100% responsible (the felons need the job and probably will sign anything) and will try and say so.  I think the courts will rake them over the coals and Uber will be 100% at fault. 
YouTube and Website Electronic Resources ------>  https://www.eevblog.com/forum/other-blog-specific/a/msg1341166/#msg1341166
 

Offline mdijkens

  • Regular Contributor
  • *
  • Posts: 146
  • Country: nl
Re: First pedestrian death by autonomous car...
« Reply #79 on: March 22, 2018, 11:40:28 pm »
Even if a better camera would be able to 'see' the pedestrian from twice as far away that still leaves little over 2 seconds to react.

twice?
This looks more like 10x

A slow reacting human could have easily stopped
 

Offline Halcyon

  • Global Moderator
  • *****
  • Posts: 5629
  • Country: au
Re: First pedestrian death by autonomous car...
« Reply #80 on: March 23, 2018, 12:12:12 am »
A slow reacting human could have easily stopped

No, they wouldn't have.
 

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Re: First pedestrian death by autonomous car...
« Reply #81 on: March 23, 2018, 02:36:39 am »
A slow reacting human could have easily stopped

No, they wouldn't have.

Full stop was not needed to avoid collision. Possibly even 0.5 secs of extreme braking with avoidance steering would do the job. Most sane drivers tend to stay on the road during obstacle avoidance - they possibly would steer into left lane. Quick search shows that accident experts use 1.5 sec as average reaction time. Well ok, ok. Olderly ppl reach 3 secs or even worse, young and alert - around 1 sec. I would say plausible to save her with 50% or better chances in case of human driver - if time to impact was 2secs or more.
 

Offline Halcyon

  • Global Moderator
  • *****
  • Posts: 5629
  • Country: au
Re: First pedestrian death by autonomous car...
« Reply #82 on: March 23, 2018, 03:27:05 am »
A slow reacting human could have easily stopped

No, they wouldn't have.

Full stop was not needed to avoid collision. Possibly even 0.5 secs of extreme braking with avoidance steering would do the job. Most sane drivers tend to stay on the road during obstacle avoidance - they possibly would steer into left lane. Quick search shows that accident experts use 1.5 sec as average reaction time. Well ok, ok. Olderly ppl reach 3 secs or even worse, young and alert - around 1 sec. I would say plausible to save her with 50% or better chances in case of human driver - if time to impact was 2secs or more.

Your original comment stated "A slow reacting human could have easily stopped". No one disputes that any amount of breaking would have reduced the speed.
 

Online Marco

  • Super Contributor
  • ***
  • Posts: 6694
  • Country: nl
Re: First pedestrian death by autonomous car...
« Reply #83 on: March 23, 2018, 10:09:06 am »
The level of black crush on that video is quite amazing. It doesn't look like that on a mobile phone camera.

 
The following users thanked this post: wilfred, thm_w, fubar.gr, ez24, MT

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Re: First pedestrian death by autonomous car...
« Reply #84 on: March 23, 2018, 10:35:19 am »
Your original comment stated "A slow reacting human could have easily stopped".

Well... it was not my comment.

Quote
No one disputes that any amount of breaking would have reduced the speed.

Reduction of the speed?  :palm:
I talk about - pedestrian could be saved by human driver compared to robot, or not.
« Last Edit: March 23, 2018, 10:40:38 am by ogden »
 

Offline Brumby

  • Supporter
  • ****
  • Posts: 12288
  • Country: au
Re: First pedestrian death by autonomous car...
« Reply #85 on: March 23, 2018, 11:19:34 am »
I suggest you are probably making an incorrect comparison: Comparing a properly functioning human to an improperly functioning "robot".  I would also think a properly functioning robot would outperform an improperly functioning human.

Be careful the component attributes of your comparison are properly identified.
 

Offline Halcyon

  • Global Moderator
  • *****
  • Posts: 5629
  • Country: au
Re: First pedestrian death by autonomous car...
« Reply #86 on: March 23, 2018, 11:28:26 am »
Well... it was not my comment.

My mistake, you are correct (as in, you weren't the author, the comment itself is still incorrect).

Reduction of the speed?  :palm:

Why *palm* ?

What point are you attempting to prove?

All things being equal, I still am of the opinion that the pedestrian is mostly to blame for the collision, based on what I've read, observed and my own training and expertise.

The level of black crush on that video is quite amazing. It doesn't look like that on a mobile phone camera.

Great, so now morons are going out to the same site, driving illegally with a mobile phone in their hand... what could go wrong?

You can't assume that a camera's sensor is equal to or worse than a human's sight. Eyesight varies. Cameras can be both more or less sensitive to light. My dashcam for example reproduces dark scenes better than I can view them in certain light and my night time vision is above average. I would argue that the original (darker) footage is more representative of an average, middle-aged persons eyesight in those conditions.
« Last Edit: March 23, 2018, 11:34:56 am by Halcyon »
 
The following users thanked this post: thm_w

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Re: First pedestrian death by autonomous car...
« Reply #87 on: March 23, 2018, 11:34:35 am »
Why *palm* ?

What point are you attempting to prove?

I *palmed* because you refused to comprehend that I did not talk just about full stop (or reduction of the speed) but: pedestrian can be saved by human driver or not.

Quote
I still am of the opinion that the pedestrian is mostly to blame for the collision

Well, indeed. Anyway such conclusion does not disqualify discussions about other related to particular accident, topics.
 

Offline Halcyon

  • Global Moderator
  • *****
  • Posts: 5629
  • Country: au
Re: First pedestrian death by autonomous car...
« Reply #88 on: March 23, 2018, 11:41:15 am »
I *palmed* because you refused to comprehend that I did not talk just about full stop (or reduction of the speed) but: pedestrian can be saved by human driver or not.

"Refuse" and "comprehend" together do not form part of my thought process... ever. Everything to me is evidence-based.

Referring back to your earlier comment, you are actually correct regarding reactionary gap. But in this case, from what I can see, if you were to apply brakes at full braking pressure at the moment an average person would have seen that person crossing the street, coming to a stop before collision would be difficult if not impossible. Swerving (in the correct direction) would probably have avoided the collision as those cars handle exceptionally well in those kinds of circumstances, but that does not reduce the pedestrian's liability. It's easy to be an armchair critic when presented with the evidence, but if you were that driver in that situation, you probably have less than a second to identify, consider and react.
 

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Re: First pedestrian death by autonomous car...
« Reply #89 on: March 23, 2018, 11:44:45 am »
I suggest you are probably making an incorrect comparison: Comparing a properly functioning human to an improperly functioning "robot".

Agreed. I was talking about "particular robot that failed", w/o intent of generalization.

Quote
I would also think a properly functioning robot would outperform an improperly functioning human.

Indeed - properly built and functioning robot can outperform human (in situations like this).
 

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Re: First pedestrian death by autonomous car...
« Reply #90 on: March 23, 2018, 12:07:56 pm »
"Refuse" and "comprehend" together do not form part of my thought process... ever.

Right. Don't even try to draw me into discussion about how bad my englisch actually is :)

Quote
But in this case, from what I can see, if you were to apply brakes at full braking pressure at the moment an average person would have seen that person crossing the street, coming to a stop before collision would be difficult if not impossible.

Do you ever read whole posts or just few out of context words that suit your agenda?

Quote
It's easy to be an armchair critic when presented with the evidence, but if you were that driver in that situation, you probably have less than a second to identify, consider and react.

1) I have enough prior knowledge and driving experience to not receive "armchair critic" label. 2) I said "I would say plausible to save her with 50% or better chances in case of human driver". Please pay close attention to fact that I leave possibility of human failure error.
« Last Edit: March 23, 2018, 08:24:59 pm by ogden »
 

Offline fubar.gr

  • Supporter
  • ****
  • Posts: 365
  • Country: gr
    • Fubar.gr
Re: First pedestrian death by autonomous car...
« Reply #91 on: March 23, 2018, 12:15:34 pm »
The level of black crush on that video is quite amazing. It doesn't look like that on a mobile phone camera.



Thanks. I came to post this video, only to see you had already posted it.

The street is perfectly lit. An attentive human driver would have absolutely no problem seeing the woman from a great distance and brake in a safe manner.

There's zero excuse for both the human driver and the autonomous system.

Online BrianHG

  • Super Contributor
  • ***
  • Posts: 7660
  • Country: ca
Re: First pedestrian death by autonomous car...
« Reply #92 on: March 23, 2018, 12:59:30 pm »
I would like to see the source video file from the camera.  This video was edited and re-mpeged by the police to blotch out the stats at the bottom of the screen.  The severe black-crushing and extra low res may have happened here, as some video editors and re-compressors do this, and there might be a lot more detail in the original source.

Or as those who delve into conspiracies, the footage was purposely degraded to hide how visible the pedestrian really was.  We are talking about the type of money here where this may be the case.
 

Offline metrologist

  • Super Contributor
  • ***
  • Posts: 2199
  • Country: 00
Re: First pedestrian death by autonomous car...
« Reply #93 on: March 23, 2018, 02:36:56 pm »
So, I think I heard the annual death rate for autonomous vehicles now is around 1 in 4 million miles traveled and for humans it's around 1 in 120 million miles.

source: armstrong and getty text in
 

Offline MT

  • Super Contributor
  • ***
  • Posts: 1616
  • Country: aq
Re: First pedestrian death by autonomous car...
« Reply #94 on: March 23, 2018, 04:43:44 pm »
1: The deliberately darkened video is why i asked for Swedish Moose avoidance software.
2: Time between pedestrian hit by car and Swedish Moose test shows Moose test is due.
3: Swedish Moostest should be a industry standard even if your country dont have Mooses.
4: Toyouta Hilux and other cars of the like is particularly vulnerable to moosetest as proven.
5: Swedish moose test is because of moose at night move from oneside to the other, at daylight
    moose test is not due since moose is out in the bush feeding.
6: With Swedish moose test avoidance software the pedestrian would have been alive.
7: Fault are at Uber software just like Tesla so called autopilot software that rear ends Firedepartment truck and
    they thought Tesla car in orbit will do the sun but in reality will rear ending earth in future due to faulty autopilot.
8: If the honorable NTSB does as they usually do they are not going to be fooled away by Uber and the local Police
    clownery.

Typical Swedish road sign who Germans enjoy stealing as souvenirs!


However Sweden has other signs as well:


Australia , well,, is just Australia!


Canada is best!


« Last Edit: March 23, 2018, 04:55:14 pm by MT »
 

Offline ez24

  • Super Contributor
  • ***
  • Posts: 3082
  • Country: us
  • L.D.A.
Re: First pedestrian death by autonomous car...
« Reply #95 on: March 23, 2018, 06:22:40 pm »
The level of black crush on that video is quite amazing. It doesn't look like that on a mobile phone camera.

Maybe the first accident recreation.  What I like about this is I can re-create it also, so next time I go out at night I will use my phones to record and see what it looks like.  Because I never thought about this.  Just curious.

Thanks Marco

YouTube and Website Electronic Resources ------>  https://www.eevblog.com/forum/other-blog-specific/a/msg1341166/#msg1341166
 

Online Marco

  • Super Contributor
  • ***
  • Posts: 6694
  • Country: nl
Re: First pedestrian death by autonomous car...
« Reply #96 on: March 23, 2018, 06:35:11 pm »
Cameras can be both more or less sensitive to light.

That's why I only said it didn't look like that, mobile phone cameras have much more competitive pressure for quality due to the importance of reviews than dashcams. They are the state of the art. Regardless, the black crush on the dash video is quite amazing. It's not just how dark it is, also how it's quantized. Part of that is compression, but it also has the look of a faulty range conversion (which is really easy to do with video, because of the soup of standards).
 

Offline fubar.gr

  • Supporter
  • ****
  • Posts: 365
  • Country: gr
    • Fubar.gr
Re: First pedestrian death by autonomous car...
« Reply #97 on: March 23, 2018, 08:09:07 pm »
Here's another video: https://twitter.com/LaurenReimerTV/status/977077647543955458

NTSB and NHTSA are conducting tests at the accident scene. They are using the same vehicle that was involved in the accident, the dent in the bumper is visible.

The street looks adequately lit.

Offline ez24

  • Super Contributor
  • ***
  • Posts: 3082
  • Country: us
  • L.D.A.
Re: First pedestrian death by autonomous car...
« Reply #98 on: March 23, 2018, 08:33:12 pm »
Here's another video: https://twitter.com/LaurenReimerTV/status/977077647543955458

The victim's name -> Elaine Herzberg.  It is reported that her daughter has hired a lawyer.

https://twitter.com/LaurenReimerTV/status/975877179924819968
YouTube and Website Electronic Resources ------>  https://www.eevblog.com/forum/other-blog-specific/a/msg1341166/#msg1341166
 

Online Marco

  • Super Contributor
  • ***
  • Posts: 6694
  • Country: nl
Re: First pedestrian death by autonomous car...
« Reply #99 on: March 23, 2018, 09:02:32 pm »
For all the excuses about "it's impossible to keep paying attention in an autonomous car" they do make it awfully easy for the driver not to even have their eyes on the road.

Hands on wheel, foot on break pedal, eyes on road ... all this can be detected and should be the standard behaviour of a safety driver. Any deviance should be punished, in real time with a loudspeaker and offline with performance metrics. This will make being a safety driver incredibly annoying, but it's a job.

PS. as for all the autopilot systems in existing cars, just ban them for the moment. In the end anything which diverts your vision below dash range for even a moment, or not having your hand/feet on the controls at speed is irresponsible. Their vigilance systems allow a lot more than that before annoying the driver, which is understandable because they want to sell you the ability to not pay attention.

PPS. a huge pay out in the mentioned lawsuit might ban autopilots through economics, but that will take a while.
« Last Edit: March 23, 2018, 09:14:07 pm by Marco »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf