Author Topic: EEVblog #1066 - Uber Autonomous Car Fatality - How?  (Read 59605 times)

0 Members and 1 Guest are viewing this topic.

Offline itdontgo

  • Regular Contributor
  • *
  • Posts: 167
  • Country: gb
    • IoT Manufacturer
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #25 on: March 22, 2018, 07:27:08 am »
How?  Seriously? Cars are the world's greatest violent killer. War is responsible for about 3% of violent deaths, homicide about 11% and the products of the automobile industry; 24%.

Every day the products of that industry and their users violently kill 3500 people. Remember that when you're crying over the next tragic news event.  What you do will kill more people this century than all the wars of the previous century managed.

If you have massive fast cars moving at speed by where there are people you will kill people. You can have as many sensors and laws as you like. People will die. When it's someone like you doing the killing as it is with the 3500 daily fatalities and 50,000 severe injuries you won't give a crap. It will be a tragic accident which could not possibly be avoided. You won't be able to imagine a way in which that life could have been saved apart from perhaps the moron who got killed should not have been walking around by a road.  This case is something out of the ordinary so all of a sudden there is a story.

We've become accustomed to the speed cars go and have settled on it. It's not been decided upon by research or evidence it's what people will accept so they can have the cars they want. Our society lives in a fantasy world when it comes to cars. They poison everyone, they violently injure people, they wreck the planet and you don't care because you're in love with them. You're as much a True Believer in cars as people are in religion. I'm sure you're objecting to what I'm saying despite the evidence in the same way a religious nut cannot hear criticism of their beloved religion. But these are facts. The cars we accept as a society being used by people will kill in vast numbers. And we put having a bit of go in our cars above people's lives.

This rant is not aimed at any one individual by the way.

Offline EEVblogTopic starter

  • Administrator
  • *****
  • Posts: 37661
  • Country: au
    • EEVblog
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #26 on: March 22, 2018, 07:36:06 am »
How?  Seriously? Cars are the world's greatest violent killer. War is responsible for about 3% of violent deaths, homicide about 11% and the products of the automobile industry; 24%.

Every day the products of that industry and their users violently kill 3500 people. Remember that when you're crying over the next tragic news event.  What you do will kill more people this century than all the wars of the previous century managed.

If you have massive fast cars moving at speed by where there are people you will kill people. You can have as many sensors and laws as you like. People will die. When it's someone like you doing the killing as it is with the 3500 daily fatalities and 50,000 severe injuries you won't give a crap. It will be a tragic accident which could not possibly be avoided. You won't be able to imagine a way in which that life could have been saved apart from perhaps the moron who got killed should not have been walking around by a road.  This case is something out of the ordinary so all of a sudden there is a story.

We've become accustomed to the speed cars go and have settled on it. It's not been decided upon by research or evidence it's what people will accept so they can have the cars they want. Our society lives in a fantasy world when it comes to cars. They poison everyone, they violently injure people, they wreck the planet and you don't care because you're in love with them. You're as much a True Believer in cars as people are in religion. I'm sure you're objecting to what I'm saying despite the evidence in the same way a religious nut cannot hear criticism of their beloved religion. But these are facts. The cars we accept as a society being used by people will kill in vast numbers. And we put having a bit of go in our cars above people's lives.

This rant is not aimed at any one individual by the way.

You missed the entire point of this video.
 

Offline EEVblogTopic starter

  • Administrator
  • *****
  • Posts: 37661
  • Country: au
    • EEVblog
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #27 on: March 22, 2018, 07:39:04 am »
That view of stepping out from the shadows may be misleading. A human may have had better vision than the vision shown by the camera footage.

Which is why I give no value judgement on that and do not hold him responsible for the impact in that respect.

Quote
This could easily have been a non fatal accident. I doubt it could have been avoided entirely by a human driver but I also don't think the camera footage is clear enough to be sure of that.

And that's why it's not the point of the video, and really shouldn't be discussed here.
This is a video about autonymous car safety and sensor tech etc, it is not about a car accident that happens a thousand times a day.
 

Offline Brumby

  • Supporter
  • ****
  • Posts: 12288
  • Country: au
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #28 on: March 22, 2018, 07:46:01 am »
This is a video about autonymous car safety and sensor tech etc, it is not about a car accident that happens a thousand times a day.

Exactly.

Our interest here is understanding why the tech didn't respond.  That's it.

And further discussion will be about the tech ... nothing else.
 

Offline EEVblogTopic starter

  • Administrator
  • *****
  • Posts: 37661
  • Country: au
    • EEVblog
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #29 on: March 22, 2018, 07:46:26 am »
We expect better than humans when we take humans out of the loop. All this extra tech means that it should perform better than a human in the areas that the sensor excels against a human. e.g. at night with LIDAR.
Even if accident rate is exactly the same, but we have gained ability to do other stuff while commuting, it is a win overall.

Unless you are the person that gets hit because of a LIDAR system that didn't work properly when it could have worked properly. And that that same issue could effect other people in future because it's a fixable problem. In that case you might start to re-think your position.
 

Offline Brumby

  • Supporter
  • ****
  • Posts: 12288
  • Country: au
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #30 on: March 22, 2018, 07:53:47 am »
Indeed.  How absurd would it be to accept a compromised technology simply because of something fixable?

The range of what is "fixable" is a broad one, true, and some things might be rather challenging - but what if the fix is a simple solder joint?
« Last Edit: March 22, 2018, 07:55:30 am by Brumby »
 

Offline itdontgo

  • Regular Contributor
  • *
  • Posts: 167
  • Country: gb
    • IoT Manufacturer
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #31 on: March 22, 2018, 08:17:50 am »
"Uber Autonomous Car Fatality - How"

I answered it. Cars are too fast to be used by people. You cannot help but kill people if the cars go that speed. There is no solution but slower cars.

Engineers think they can engineer away road fatalities without slowing down cars. But you can't because you're engineering the car and the road and not the people who are fixed. They're there and they're easily killable at over 20mph.

This I admit is a diversion from your video but I think what I am saying is entirely relevant
« Last Edit: March 22, 2018, 08:30:28 am by itdontgo »
 

Offline CNe7532294

  • Regular Contributor
  • *
  • Posts: 108
  • Country: us
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #32 on: March 22, 2018, 08:19:59 am »
Honestly, technology no matter how good isn't a substitute for brains and actions. |O :palm: It a tool. A tool to be used to enhance human activity. Doesn't matter if things are visually terrible. A smart and alert human being would slow down and be cautious in darken areas. If you need guidance on self driving cars then look no further than what is required of commercial airlines. Making the driver a passenger is a terrible idea. :palm: Hopefully lessons will be learned. Sad its going to be the hard way.

 

Offline Brumby

  • Supporter
  • ****
  • Posts: 12288
  • Country: au
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #33 on: March 22, 2018, 08:31:29 am »
Doesn't matter if things are visually terrible. A smart and alert human being would slow down and be cautious in darken areas.
That's a rather interesting statement.  More holes than a cheese grater, but interesting (or should I say naive?).

Quote
If you need guidance on self driving cars then look no further than what is required of commercial airlines.
Really?  TCAS would go apeshit!!
 

Offline EEVblogTopic starter

  • Administrator
  • *****
  • Posts: 37661
  • Country: au
    • EEVblog
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #34 on: March 22, 2018, 08:31:58 am »
"Uber Autonomous Car Fatality - How"
I answered it.

No you didn't, because you refuse to acknowledge what the video and question is actually about.

Quote
This I admit is a diversion from your video but I think what I am saying is entirely relevant

And this is not the place to discuss that stuff.
 

Offline HKJ

  • Super Contributor
  • ***
  • Posts: 2899
  • Country: dk
    • Tests
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #35 on: March 22, 2018, 09:05:25 am »
It do not look like she goes out in front of the car, more like she already is there, but because the head light is very limited in range, she is first seen to late.
This may mean the car is driving too fast compare to the headlight range (I know that is very common), but I do not see how lidar or radar could miss her.
 
The following users thanked this post: JXL

Offline CNe7532294

  • Regular Contributor
  • *
  • Posts: 108
  • Country: us
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #36 on: March 22, 2018, 09:27:36 am »
Doesn't matter if things are visually terrible. A smart and alert human being would slow down and be cautious in darken areas.
That's a rather interesting statement.  More holes than a cheese grater, but interesting (or should I say naive?).

Quote
If you need guidance on self driving cars then look no further than what is required of commercial airlines.
Really?  TCAS would go apeshit!!

Please do point out the cheese holes.

TCAS is to avoid plane collisions. Not detect other objects in the way (ie. birds, people, truck, cars, etc.). Its not even turned on yet while taxiing from the gate to the runaway.

Overall you misunderstood my post. I'm not for shutting down the tech. I laugh at anyone thinking this is the end of automated driving. However, I'm against shutting down the extra layer of protection called human action. A quick search on piloting will turn up pilots work in tandem with the autopilot. Tech is not there to make people lazy. As I said, only there to make things easier. There is a difference between those things. Also if you notice, planes try to have 2 or more things for a reason (ie. two engines, two pilots along with the autopilot, two sets of gauges, 2 radio sets, 2-3 hydraulic systems, 2-3 fuel lines with crossfeed valves, etc.). Its called being redundant. It works. :palm:

I have no proof in saying this exact part here but there is no doubt in my mind yet uber reps told the driver everything is fine. This car drives itself. Causes her to lower her guard. Pays attention to the phone. We all know the rest. :palm: If you're by yourself and need to sleep, drunk, or pay more attention to your phone, take a real taxi or bus please. Basically saying err on the side of caution. This is yet another case to build on this wise saying. Certainly won't be the last. :rant:
 

Offline firewalker

  • Super Contributor
  • ***
  • Posts: 2450
  • Country: gr
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #37 on: March 22, 2018, 10:00:53 am »
Companies on the autonomous car industry exchange infos for their systems? Should they forced to do so?

Alexander.
Become a realist, stay a dreamer.

 

Offline Brumby

  • Supporter
  • ****
  • Posts: 12288
  • Country: au
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #38 on: March 22, 2018, 10:53:49 am »
Overall you misunderstood my post. I'm not for shutting down the tech. I laugh at anyone thinking this is the end of automated driving. However, I'm against shutting down the extra layer of protection called human action. A quick search on piloting will turn up pilots work in tandem with the autopilot. Tech is not there to make people lazy. As I said, only there to make things easier. There is a difference between those things. Also if you notice, planes try to have 2 or more things for a reason (ie. two engines, two pilots along with the autopilot, two sets of gauges, 2 radio sets, 2-3 hydraulic systems, 2-3 fuel lines with crossfeed valves, etc.). Its called being redundant. It works. :palm:

I have no proof in saying this exact part here but there is no doubt in my mind yet uber reps told the driver everything is fine. This car drives itself. Causes her to lower her guard. Pays attention to the phone. We all know the rest. :palm: If you're by yourself and need to sleep, drunk, or pay more attention to your phone, take a real taxi or bus please. Basically saying err on the side of caution. This is yet another case to build on this wise saying. Certainly won't be the last. :rant:

To put it simply, your implication is impractical.  Put anyone in a car that drives itself and after a while, EVERYbody is going to drop their attention at times.  AS IT IS, we can't even be assured that people who have full driving responsibilities will pay attention.  Just look at the issues arising from mobile phones.

Anyway, that's not the topic of this thread.
 

Offline G7PSK

  • Super Contributor
  • ***
  • Posts: 3859
  • Country: gb
  • It is hot until proved not.
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #39 on: March 22, 2018, 10:56:40 am »
What I would like to know is for how long the car had been driving autonomously, was the driver in the act of handing over to the car or taking over from it. Also how does roadside clutter affect the lidar and radar, it was not so long ago that military airborne radar systems were confusing things like helicopters with trucks and even trees. That cyclist was pushing the bike across the road so would have been on the open road for some seconds, certainly should have been enough time for the lidar and or the radar to pick up.
 

Offline AF6LJ

  • Supporter
  • ****
  • Posts: 2902
  • Country: us
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #40 on: March 22, 2018, 11:42:49 am »
The car didn't even slow down.
This technology has much further to go before it is ready for prime time.
Sue AF6LJ
 
The following users thanked this post: JXL

Offline Fungus

  • Super Contributor
  • ***
  • Posts: 16561
  • Country: 00
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #41 on: March 22, 2018, 11:56:28 am »
I think it's very complicated, I don't think we can say "It should have seen that on the LIDAR!", or "Shadows don't affect robots!".

LIDAR can see shapes but it can't see things like white lines painted on the road, you need visible light for that. The vision system has to be a complex composition of all the inputs. Shadows will definitely cause problems at night.

It's a two lane road, and an object in the other lane as a car is driving along isn't unusual. I'm sure the LIDAR saw it but how is the car supposed to know the object is about to move in front of an approaching car? It makes no sense. Somewhere in the software there has to be some assumptions that other road users simply don't do that. The car would be constantly braking otherwise and could never overtake anything.

The real question is: How could the pedestrian not see the approaching car? They just walked straight across the road without even looking. No human would have expected that either.
 

Offline Fungus

  • Super Contributor
  • ***
  • Posts: 16561
  • Country: 00
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #42 on: March 22, 2018, 12:03:37 pm »
The car didn't even slow down.

Should it have? Should it brake every time there's an object in the other lane?

This technology has much further to go before it is ready for prime time.

I disagree. The time to switch over is when cars are safer than humans, not when cars are 100% perfect (which they can never be).

The question to ask is therefore: Is it likely a human have done better in that situation? I say "no".

https://en.wikipedia.org/wiki/Nirvana_fallacy
 

Offline AF6LJ

  • Supporter
  • ****
  • Posts: 2902
  • Country: us
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #43 on: March 22, 2018, 12:08:31 pm »
I agree on the LIDAR issue....
I would also add there is so much input a human driver receives under the same circumstances, I find myself wondering if there is enough parallel processing taking place for the car to react as fast as a human.
Sue AF6LJ
 

Offline AF6LJ

  • Supporter
  • ****
  • Posts: 2902
  • Country: us
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #44 on: March 22, 2018, 12:12:18 pm »
The car didn't even slow down.

Should it have? Should it brake every time there's an object in the other lane?

This technology has much further to go before it is ready for prime time.

I disagree. The time to switch over is when cars are safer than humans, not when cars are 100% perfect (which they can never be).

The question to ask is therefore: Is it likely a human have done better in that situation? I say "no".

https://en.wikipedia.org/wiki/Nirvana_fallacy

The car should have seen the pedestrian in the shadows.
The car should have calculated the path the pedestrian was taking.
The car should have  slowed down.

Granted...
The Pedestrian was jay walking...
In The People's Republic of Kalifornia you must be within 30' of an intersection (about 9 meters) to be considered to be making a lawful street crossing. The pedestrian should have walked to the intersection, even though it was 500' away.
Sue AF6LJ
 

Offline Fungus

  • Super Contributor
  • ***
  • Posts: 16561
  • Country: 00
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #45 on: March 22, 2018, 12:15:32 pm »
The pedestrian should have walked to the intersection, even though it was 500' away.

If only they had a bicycle...

I find myself wondering if there is enough parallel processing taking place for the car to react as fast as a human.

Humans are fast?
« Last Edit: March 22, 2018, 12:20:10 pm by Fungus »
 

Offline Fungus

  • Super Contributor
  • ***
  • Posts: 16561
  • Country: 00
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #46 on: March 22, 2018, 12:26:56 pm »
The car should have seen the pedestrian in the shadows.

I'm sure it did see the pedestrian.

(and that a human driver wouldn't have)

The car should have calculated the path the pedestrian was taking.
The car should have  slowed down.

That part I'm not so sure about.

It's easy to say in when you only look at this one data point in isolation.

In general: Not so much. I bet you can find examples where this exact set of inputs is perfectly Ok (right up until it's too late).

eg. How do you propose cars should deal with motorcyclists in the other lane? Hitting the brakes every time they see one move slightly sideways? That would be dangerous.
« Last Edit: March 22, 2018, 12:38:01 pm by Fungus »
 

Offline orion242

  • Supporter
  • ****
  • Posts: 746
  • Country: us
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #47 on: March 22, 2018, 12:35:08 pm »
So does Uber have a case to go back on the judgement over stealing Google's AV tech?  Certainly it wasn't as good and valuable as G made it out to be LOL.
 

Offline AF6LJ

  • Supporter
  • ****
  • Posts: 2902
  • Country: us
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #48 on: March 22, 2018, 12:41:29 pm »
The car should have seen the pedestrian in the shadows.

I'm sure it did see the pedestrian.

(and that a human driver wouldn't have)

The car should have calculated the path the pedestrian was taking.
The car should have  slowed down.

That part I'm not so sure about.

It's easy to say in when you only look at this one data point in isolation.

In general: Not so much. I bet you can find examples where this exact set of inputs is perfectly Ok (right up until it's too late).

eg. How do you propose cars should deal with wobbly motorcyclists in the other lane? Hitting the brakes every time they see one? That would be dangerous.

That video is far from one data point in isolation; that video is evidence the autonomous driving system is far from ready. This proves to me that UBER has no idea what the hell they are doing, no software QC, no dynamic testing with actual moving targets that need to be tracked.
While that video may not be all you need to know the system is not anywhere near ready for prime time, the video is proof Uber along with what I have seen come out of Google are nowhere near ready to deploy these cars on the street.

Don't even get me started on how stupidly foolish it is to solely rely on GPS as a means of navigation.
Sue AF6LJ
 

Offline Brumby

  • Supporter
  • ****
  • Posts: 12288
  • Country: au
Re: EEVblog #1066 - Uber Autonomous Car Fatality - How?
« Reply #49 on: March 22, 2018, 01:10:11 pm »
The car should have seen the pedestrian in the shadows.

I'm sure it did see the pedestrian.
One would hope the technology should have - but that's not been demonstrated as yet.

Quote
(and that a human driver wouldn't have)
I'm not so sure about that.

Quote
The car should have calculated the path the pedestrian was taking.
The car should have  slowed down.

That part I'm not so sure about.
I am.

Quote
It's easy to say in when you only look at this one data point in isolation.
Careful how you use the term "data point".  I interpret it as you meaning "this one example".

Quote
In general: Not so much. I bet you can find examples where this exact set of inputs is perfectly Ok (right up until it's too late).
This is where I would ask for examples - but I cannot see you succeeding.

The car knows its speed and direction.  It detects the location of the potential hazard at a specific point in time.  It should also detect the approximate size.  A fraction of a second later, these measurements are repeated.  The change in position can be used to calculate a movement vector, the paths of the potential hazard and the vehicle can be assessed and collision risk determined.  Subsequent measurements can be used to confirm this - or show changes to that risk.

These - and a lot of other measurements - are captured and processed in real time ... and if you think computing power is a problem, then you really need to update your understanding.

Bottom line - the intersecting paths of the pedestrian and the vehicle should have been easily determined.

Quote
eg. How do you propose cars should deal with motorcyclists in the other lane? Hitting the brakes every time they see one move slightly sideways? That would be dangerous.
Using exactly the same process as I outlined above.  A bike wandering towards a lane is not a risk.  Crossing into might be.

The same for a bike zipping through traffic, ducking in and out of lanes.  As long as the bike's vector does not indicate a collision course with the vehicle, no action needs to be taken.
« Last Edit: March 22, 2018, 01:12:08 pm by Brumby »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf