Uber driverless car in fatal accident

Uber driverless car in fatal accident

Author
Discussion

captainaverage

596 posts

88 months

Thursday 22nd March 2018
quotequote all
BugLebowski said:
Also that's a completely unlit section of road and the vehicle appears to be driving faster...
I'm sure the techies will come and tell you how that's just a one off and those computers can do it again and again.

The Selfish Gene

5,516 posts

211 months

Thursday 22nd March 2018
quotequote all
akirk said:
We need to stop this apologist explanation saying that the human driver could have done very little about it... they absolutely could have avoided it...
Someone died - it is a pretty major negative outcome - there is lots that could have been done to prevent it...
- the driver is still meant to be ultimately responsible - they should have been paying attention
- the driver should have picked up that the lights were not showing enough road for the speed, and taken back control
- the driver should then have braked and avoided killing the pedestrian

instead, the driver let the car make all the choices, including bad choices...
they should be preventing the car from entering a situation where they no longer have time to react... otherwise their role is invalid / pointless

i.e. the driver should have avoided the situation starting - which would have avoided the outcome
we should not be accepting media and others saying - they could do nothing about it because they didn't have time to react - it was their job to only drive in such a way that they would have time to react!
bingo

herewego

8,814 posts

214 months

Thursday 22nd March 2018
quotequote all
BugLebowski said:
Fish said:
Heres a human having a go for real... makes a much better effort than the Uber..

https://www.youtube.com/watch?v=wsN3jchvZ1s
Also that's a completely unlit section of road and the vehicle appears to be driving faster...
And s/he seems to be on dipped beams too.

The Selfish Gene

5,516 posts

211 months

Thursday 22nd March 2018
quotequote all
captainaverage said:
BugLebowski said:
Also that's a completely unlit section of road and the vehicle appears to be driving faster...
I'm sure the techies will come and tell you how that's just a one off and those computers can do it again and again.
ha I'm a techie and I say absolutely not. It clearly failed in this case spectacularly and killed someone. It is literally outrageous that more isn't being made of this.



oyster

12,609 posts

249 months

Thursday 22nd March 2018
quotequote all
The Selfish Gene said:
100% this - hopefully it'll cripple Uber and get them binned off the planet.
I'm no fan of Uber at all, but from the moment you posted this I couldn't read any more of your posts, as you come across as having a vested interest.

CzechItOut

2,154 posts

192 months

Thursday 22nd March 2018
quotequote all
Could it be that the car is programmed not to suddenly swerve or change lanes in order to protect the occupants?

captainaverage

596 posts

88 months

Thursday 22nd March 2018
quotequote all
The Selfish Gene said:
ha I'm a techie and I say absolutely not. It clearly failed in this case spectacularly and killed someone. It is literally outrageous that more isn't being made of this.
Ok I'm sorry. I really just meant the disconnected from reality crowd.

Guybrush

4,355 posts

207 months

Thursday 22nd March 2018
quotequote all
It's not practical from a human nature point of view to expect a person sitting in the driver's seat of a 'driverless' car to pay as much attention to what is going on, as a driver who is having to pay attention to managing a non-driverless car. What is the 'driver' of the 'driverless' car to do? Monitor the progress of the 'driverless' car at all times, while wondering if it's going to react appropriately to every situation and take over just in time if it's decided the car won't react in time? It's a completely unreasonable expectation. Surely the main selling point of a 'driverless' car is to enable all occupants to relax while the car does the work of driving. Otherwise, what's the point? It's likely to be even less relaxing than just driving yourself..

Scootersp

3,197 posts

189 months

Thursday 22nd March 2018
quotequote all
akirk said:
We need to stop this apologist explanation saying that the human driver could have done very little about it... they absolutely could have avoided it...
Someone died - it is a pretty major negative outcome - there is lots that could have been done to prevent it...
- the driver is still meant to be ultimately responsible - they should have been paying attention
- the driver should have picked up that the lights were not showing enough road for the speed, and taken back control
- the driver should then have braked and avoided killing the pedestrian

instead, the driver let the car make all the choices, including bad choices...
they should be preventing the car from entering a situation where they no longer have time to react... otherwise their role is invalid / pointless

i.e. the driver should have avoided the situation starting - which would have avoided the outcome
we should not be accepting media and others saying - they could do nothing about it because they didn't have time to react - it was their job to only drive in such a way that they would have time to react!
I think a human driver would have done better but I don't necessarily blame this specific 'driver'. I don't know their terms and conditions or what they are instructed whilst in the car but I 100% agree that the less controls inputs you give a person the less focus they have. I don't even really like cruise control as it's a little step towards switching off.

IMO it's precisely these situations where the technology is meant to help, I remember reading of a Tesla braking hard because the car in front of the car in front broke hard, so the radar (or whatever system is was) picked up a possible issue a, 'not looking far enough forward' human wouldn't, and was there to ensure even if the car in front rear ended the next car hard and thus reducing the available stopping distance that it would stop in time or at the very least lessen any impact (and also it would massively help those behind it getting a similar early warning)

It has to be the easiest thing in the world to test for this sort of thing off the public highway with a variety of subjects/items and ambient conditions? you'd almost expect this to be routinely done before a car like this hits the road at all wouldn't you? it was someone walking with a bike (a common roadside scenario - especially as the bike part is pretty much irrelevant?) that it didn't see, not some based jumper landing randomly in the freeway, maybe the systems failed ie it would normally pick this up no problem, but then you'd expect if a system was inoperable then it would have a limp mode like we have for mechanical issues.

heebeegeetee

28,782 posts

249 months

Thursday 22nd March 2018
quotequote all
The Selfish Gene said:
Lord - the problem is - when there are deaths with new technology - that isn't innocent non related parties like in this case. If a test pilot dies, he has assumed that risk.

We shouldn't be using humans as test fodder. That is entirely unacceptable.

They shouldn't have any freedom to test these cars in public until they have been proven in controlled conditions.
Similar thinking held back the development of the motor car in the UK back in the day, when for 12 years a man had to walk in front of the car with a red flag.

You may find this interesting: https://en.wikipedia.org/wiki/Bridget_Driscoll "The coroner, Percy Morrison, (Croydon division of Surrey) said he hoped "such a thing would never happen again." The Royal Society for the Prevention of Accidents estimate 550,000 people had been killed on UK roads by 2010."

DonkeyApple

55,444 posts

170 months

Thursday 22nd March 2018
quotequote all
BugLebowski said:
I was wondering how an area with so many street lights could be so poorly lit as seen from the video uber released. Then I seen this video which apparently shows the same stretch. This seems to be more like what a human would see given the amount of street lights.

Doesn't look good for Uber or the driver IMO
Seeing as Uber is a fundamentally dishonest company which routinely displays its low moral standards, is there a suggestion yet that their evidence isn’t a pure as it might otherwise be then?

havoc

30,094 posts

236 months

Thursday 22nd March 2018
quotequote all
The Selfish Gene said:
akirk said:
We need to stop this apologist explanation saying that the human driver could have done very little about it... they absolutely could have avoided it...
Someone died - it is a pretty major negative outcome - there is lots that could have been done to prevent it...
- the driver is still meant to be ultimately responsible - they should have been paying attention
- the driver should have picked up that the lights were not showing enough road for the speed, and taken back control
- the driver should then have braked and avoided killing the pedestrian

instead, the driver let the car make all the choices, including bad choices...
they should be preventing the car from entering a situation where they no longer have time to react... otherwise their role is invalid / pointless

i.e. the driver should have avoided the situation starting - which would have avoided the outcome
we should not be accepting media and others saying - they could do nothing about it because they didn't have time to react - it was their job to only drive in such a way that they would have time to react!
bingo
yes

Blind trust in unproven technology isn't enough. Not by a long shot.

In addition:-
- The car SHOULD have had something like LIDAR/RADAR to pick this up, or even some sort of low-light vision system. BLIND reliance on visible light is piss-poor quite frankly. Be very interested to know whether any of the above was fitted, and why it failed.
- There SHOULD (probably won't be / can't be without true AI) some way of picking up on the subtle clues that an attentive driver would use to highlight potential hazards before they 'jump out' (literally), e.g.
- Signs of kids playing nearby. Toys on the ground / playground across the road / even feet visible under cars
- Indications of animals nearby - parks, unfenced farmland, etc.
- Indications of side-roads joining ahead, e.g. hedgerow/treeling convergence, or even driveways/farm tracks which wouldn't be on any inbuilt map
- Seeing a pub ahead when you're driving around kicking-out time
etc. etc...


Human beings are capable of reacting to all manner of subtle / subconscious inputs that a computer most probably cannot. Not saying every driver does, but even a modestly-competent driver would pick up on environmental cues about the appropriate speed.

The Selfish Gene

5,516 posts

211 months

Thursday 22nd March 2018
quotequote all
oyster said:
The Selfish Gene said:
100% this - hopefully it'll cripple Uber and get them binned off the planet.
I'm no fan of Uber at all, but from the moment you posted this I couldn't read any more of your posts, as you come across as having a vested interest.
I have no vested interested (as in i'm not a black cabbie etc)

I'm a computer scientist - and I've seen what they did with the greyball thing. How despicable that was.

I also have seen the reports on how many sexual assaults there are.

I also ride a motorbike in central London every day and see how unbelievably dangerous their drivers are.

I am also a professional tester as I said earlier. I'm the guy that designs the strategies on programmes not unlike the one Uber appear to be badly running on autonomous cars. The fact they've had this failure that has cost a life offends my professional sensibilities of a industry that i'm very proud to be part of!

So - my hate for Uber as an organisation has built up over time with experience............but I promise no vested interest.

akirk

5,395 posts

115 months

Thursday 22nd March 2018
quotequote all
Scootersp said:
akirk said:
We need to stop this apologist explanation saying that the human driver could have done very little about it... they absolutely could have avoided it...
Someone died - it is a pretty major negative outcome - there is lots that could have been done to prevent it...
- the driver is still meant to be ultimately responsible - they should have been paying attention
- the driver should have picked up that the lights were not showing enough road for the speed, and taken back control
- the driver should then have braked and avoided killing the pedestrian

instead, the driver let the car make all the choices, including bad choices...
they should be preventing the car from entering a situation where they no longer have time to react... otherwise their role is invalid / pointless

i.e. the driver should have avoided the situation starting - which would have avoided the outcome
we should not be accepting media and others saying - they could do nothing about it because they didn't have time to react - it was their job to only drive in such a way that they would have time to react!
I think a human driver would have done better but I don't necessarily blame this specific 'driver'. I don't know their terms and conditions or what they are instructed whilst in the car but I 100% agree that the less controls inputs you give a person the less focus they have. I don't even really like cruise control as it's a little step towards switching off.

IMO it's precisely these situations where the technology is meant to help, I remember reading of a Tesla braking hard because the car in front of the car in front broke hard, so the radar (or whatever system is was) picked up a possible issue a, 'not looking far enough forward' human wouldn't, and was there to ensure even if the car in front rear ended the next car hard and thus reducing the available stopping distance that it would stop in time or at the very least lessen any impact (and also it would massively help those behind it getting a similar early warning)

It has to be the easiest thing in the world to test for this sort of thing off the public highway with a variety of subjects/items and ambient conditions? you'd almost expect this to be routinely done before a car like this hits the road at all wouldn't you? it was someone walking with a bike (a common roadside scenario - especially as the bike part is pretty much irrelevant?) that it didn't see, not some based jumper landing randomly in the freeway, maybe the systems failed ie it would normally pick this up no problem, but then you'd expect if a system was inoperable then it would have a limp mode like we have for mechanical issues.
agree that you can't simplify blame - but you absolutely can not absolve the driver of blame either - the car which they were controlling killed someone - therefore they have had a role to play in that. Unless the authorities have agreed a different liability structure and allowed Uber to test without requiring a human to be overall in control 100% of the time, then the driver is still a driver, even if they do not need their hands on the wheel to control the car - the car is an inanimate object - it doesn't decide to be switched on / it doesn't decide what software is downloaded / it makes no decisions independent of what it is given - so we can not allow businesses to blame the car as a method of distracting blame from where it should sit. If the car is to blame - is it locked up for 20 years - or does it get capital punishment?! biggrin

I suspect that there is no agreement to allow cars to drive with no human control - if there is, then at the current state of technology the decision makers will share responsibility - if there isn't then Uber have arbitrarily made a decision to run their tests in an inappropriate manner.

For those saying you can not expect the human to be in control - of course you can - you give them a set of structured activities, get them to give a spoken commentary to be recorded, get them viewing a number of areas ahead and reporting in their commentary what is happening... even if you are having to invent tasks - you do so precisely to avoid the issue where they switch off - not exactly difficult.

Then you restrict their time behind the wheel - someone above suggested 2 drivers with max 30 minute stints - seems logical. The only reason that Uber don't do this is cost - well their making profits seems to have led to a consequence that society will generally not accept - that then becomes corporate manslaughter. If their average recruit can't cope, then employ those who can - there are plenty of industry sectors where that level of concentration is essential and delivered day in day out by highly qualified people - police drivers in pursuit / fighter pilots / air traffic controllers to give just a couple of examples - the reason that Uber doesn't employ them? cost.

Ultimately if you have a situation where business is big and powerful enough that they can do what they want - and their key driver is profit, then this is a likely consequence

anonymous-user

55 months

Thursday 22nd March 2018
quotequote all
akirk said:
We need to stop this apologist explanation saying that the human driver could have done very little about it... they absolutely could have avoided it...
Someone died - it is a pretty major negative outcome - there is lots that could have been done to prevent it...
- the driver is still meant to be ultimately responsible - they should have been paying attention
- the driver should have picked up that the lights were not showing enough road for the speed, and taken back control
- the driver should then have braked and avoided killing the pedestrian

instead, the driver let the car make all the choices, including bad choices...
they should be preventing the car from entering a situation where they no longer have time to react... otherwise their role is invalid / pointless

i.e. the driver should have avoided the situation starting - which would have avoided the outcome
we should not be accepting media and others saying - they could do nothing about it because they didn't have time to react - it was their job to only drive in such a way that they would have time to react!
I admitted earlier in the thread that if it was me driving, that woman would still have been over the bonnet.

If the camera in the car is showing a realistic picture with regards to light levels etc then I also wouldn't have seen the victim until a second or so before impact, and at 35-40mph that isn't enough to avoid that kind of thing.

You mention that there wasn't enough light on the road. I'm not sure what you mean by that? It looked like normal dipped beam to me, which is what I would be using in any kind of urban area, or any road where there is street lighting or other vehicles present.

It's not pleasant admitting the above, because it makes me think 'Could have been me driving'.

What I will say is that the driver should have been concentrating on the road ahead and not on their phone. Not acceptable.

If ever there was a 'came out out nowhere' video, this was it.

I was shocked watching it. I wouldn't have even managed half a swear word, never mind swerved, by the time the pedestrian impacted.

Horrible.

Edited to add:

I've just watched the video where someone else drives that stretch of road at night and it appears much better lit than the Uber dashcam footage.

So I'm not sure what to think.

If it was as dark as the Uber footage then I'm 99% sure I would have hit the pedestrian.

If it was as light as the other footage, then I would like to think I had more of a chance of seeing the pedestrian crossing into my path.

Edited by anonymous-user on Thursday 22 March 17:01

essayer

9,084 posts

195 months

Thursday 22nd March 2018
quotequote all
I don’t understand how the system failed, it’s a pretty simple collision detection that any production AEB-equipped vehicle would have picked up easily

poo at Paul's

14,153 posts

176 months

Thursday 22nd March 2018
quotequote all
Lord Marylebone said:
I admitted earlier in the thread that if it was me driving, that woman would still have been over the bonnet.
If you genuinely think that, maybe you should hang up the car keys!

Plenty of time to brake or swerve of just yield hundreds of yards away by using the lights properly and seeing her on the other carriageway.


No excuses, something / somebody fked up and some old dear is dead. If the observer had been paying attention, it could have been avoided. If you are saying you drive with a similarly low level of participation and concentration, time for a cab, (maybe not an Uber).lol

AreOut

3,658 posts

162 months

Thursday 22nd March 2018
quotequote all
Dunno what the law is in USA but in Australia you must keep both hands on the wheel at all times unless changing gears or indicating, no matter what you drive.

Digger

14,705 posts

192 months

Thursday 22nd March 2018
quotequote all
How long do these Uber drivers drive for in these sessions?

All assumptions.... on the assumption there was no audible alert inside the vehicle prior to impact she had half a second(ish) for her brain to react, and a similar amount of time to perform some kind of emergency response manoeuvre.... by which time it would have been too late.

Apparently 11/10 PH’ers would have effortlessly avoided this accident...

Only going on what was apparent from the video.

It would have been interesting to have seen the section of the driver-facing camera video when the reaction, response, and collision actually took place. This didn’t seem to be shown from what I could see.

eta- just seen the version where she reacts. Looks like she had just enough time to get her hands back on the wheel...

Edited by Digger on Thursday 22 March 17:26

akirk

5,395 posts

115 months

Thursday 22nd March 2018
quotequote all
Lord Marylebone said:
I admitted earlier in the thread that if it was me driving, that woman would still have been over the bonnet.
If the camera in the car is showing a realistic picture with regards to light levels etc then I also wouldn't have seen the victim until a second or so before impact, and at 35-40mph that isn't enough to avoid that kind of thing.
You mention that there wasn't enough light on the road. I'm not sure what you mean by that? It looked like normal dipped beam to me, which is what I would be using in any kind of urban area, or any road where there is street lighting or other vehicles present.
It's not pleasant admitting the above, because it makes me think 'Could have been me driving'.
What I will say is that the driver should have been concentrating on the road ahead and not on their phone. Not acceptable.
If ever there was a 'came out out nowhere' video, this was it.
I was shocked watching it. I wouldn't have even managed half a swear word, never mind swerved, by the time the pedestrian impacted.
Horrible.

Edited to add:
I've just watched the video where someone else drives that stretch of road at night and it appears much better lit than the Uber dashcam footage.
So I'm not sure what to think.
If it was as dark as the Uber footage then I'm 99% sure I would have hit the pedestrian.
If it was as light as the other footage, then I would like to think I had more of a chance of seeing the pedestrian crossing into my path.
I would be surprised - the majority of drivers in our country would simply not drive at the speed suggested in the lighted area suggested - the video implies that there was only c. 60 foot of lighted road ahead based on when the pedestrian can first be seen - if so, that is under a third of the US requirement for dipped headlights (lowered beam in their terminology) - as per the diagrams I linked to above... hence not enough light to meet the US regulations on lighting if those figures are accurate...
Fundamentally that means that:
- the car had the wrong lights on (DRL instead of dipped)
- the car was going too fast for the driver to be able to stop in the time they have for the area they can see...

lots of drivers get many things wrong - but on a darkish road, most realise if they can't see ahead and are driving into the dark (some don't but they are the exception!), so most humans would have dealt with that without issue...

And sorry - it was not a 'came out of nowhere' video - the pedestrian had crossed several carriageways already - the car was going at a speed where there was ample time to stop if dipped headlights were in use - they should have had c. 1 second to react and c. 5 seconds to stop and only c. 2-3 needed - so should have had double the requirement... the vast majority of drivers would have had no issue with that

I think you do yourself down if you think that you would have hit them - I doubt it... this was basically a car with its eyes closed (no systems picking up / dealing with the situation) and a driver with their eyes effectively closed... If you want to go out in a car blindfolded at c. 40mph then you will hit pedestrians... I would hope that you would be a degree better than that biggrin certainly from how you write on here / your introspection and thinking - I would be very surprised if you were totally unaware of what is around you!