Author Topic: Self driving liability  (Read 4230 times)

0 Members and 1 Guest are viewing this topic.

Online nctnico

  • Super Contributor
  • ***
  • Posts: 28429
  • Country: nl
    • NCT Developments
Re: Self driving liability
« Reply #25 on: April 22, 2022, 04:18:19 pm »
This thread illustrates the dominance of emotion over logic in discussions of this topic.  Specifically the arguments about liability for speeding tickets when in self driving mode. 

In my mind there is only a very minor argument here.  In general, in self driving mode the car should never speed.  There are only two reasons I can think of for this not being the case.  First, if the vendor of the software set it that way.  In which case liability is obvious and indisputable.  The second case is if the speed limit has changed and the database available to the self driving software is not updated.  Liability here could be disputable, but this is relatively uncommon (and also likely to trip up a human driver).
You forget the third case: speed limits can be changed quickly and regulary (think about road works which can also move). Also on many roads there are dynamic speed limits depending on congestion levels. All in all it is very easy to miss a sign that tells the speed limit has changed and a database doesn't help. When using Google maps it happens very often that it indicates the wrong speed limit while there are fixed signs on a highway.

Road works are particulary nasty because the signs aren't always clear. In France for example there isn't always sign that indicates the end of the road works. You just have to deduce that from the section marked with orange cones ending. Another example are parallel roads with different speed limits. Even for a human it is easy to get confused.

The bottom line is: a self driving car will be getting speeding tickets even though the intention is not to exceed the speed limit.

BTW: I'm not against self driving cars at all. On the contrary; I think that they will give me freedom when I'm too old to drive by myself so I don't have to depend on family members or crappy bus service. And hopefully before that an opportunity to do something usefull or just enjoy the landscape during a journey instead of operating a machine that -from a basic operating point of view- hasn't really evolved for over 100 years. I do find the legal implications interesting though.
« Last Edit: April 22, 2022, 06:14:31 pm by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 
The following users thanked this post: Someone

Offline Someone

  • Super Contributor
  • ***
  • Posts: 5155
  • Country: au
    • send complaints here
Re: Self driving liability
« Reply #26 on: April 22, 2022, 10:44:30 pm »
Isn't there a similar problem for airline pilots on long flights when the autopilot is flying?

I am only familiar with US FAA rules.  We have 3 categories: Scheduled airlines, Private and charter (i.e., "General Aviation"), and Military.

Large scheduled airliners are highly integrated.  I learned recently, for example, that in landing, the pilot cannot control some functions like braking and reverse thrust if the radar altimeter and/or some other electronic functions do not work.  There was a recent example on a transcontinental flight to Paris.  Plane was in good shape, but something was wrong with the automatic landing system(s), and the pilots had to declare an emergency.  Add, of course the 737 MAX disasters of a couple of years ago.  Boeing has that liability.  My take was that I don't want to fly in any airplane where having the pilots in control is an emergency. ;)

In smaller GA aircraft, the pilot is responsible.  Sure, the manufacturers get sued too, but the FAA assigns blame on the pilot(s) regardless of the autopilot.  (Some large aircraft used for scheduled service are also flown under GA rules.)   It is not uncommon to see an accident report with the conclusion that the pilot failed to maintain a safe altitude and/or airspeed.  As just one example, there was an incident in Southern California many years ago in which the ground radar controller vectored a small GA Beechcraft into a mountain.  The pilot was still blamed for not maintaining awareness of the terrain.
You have it back to front:
"International Fatality Rates, A Comparison of Australian Civil Aviation Fatality Rates with International Data"
https://www.atsb.gov.au/media/32897/b20060002.pdf
or TLDR/easier to digest:
https://skybrary.aero/articles/general-aviation-ga

Flying on non-scheduled air transport, particularly private (rather than charter), is much much more dangerous by 1-2 orders of magnitude. The measure of fatalities per 100,000 hours is easy to measure for that industry, but the gap is even wider if you use the more sensible fatalities per billion passenger km. I'm sticking to the pilots forced to be conservative by the commercial airline industry, its a system that works and has an excellent track record to back it up.
 

Offline Someone

  • Super Contributor
  • ***
  • Posts: 5155
  • Country: au
    • send complaints here
Re: Self driving liability
« Reply #27 on: April 22, 2022, 10:48:23 pm »
Yes for me the whole point in owning a car, especially a relatively high performance car is that I get to drive it myself. I drove a Tesla Y for a bit and it was a blast. Fastest car I've ever driven.
... I completely agree that being able DRIVE is the complete point of owning any car. I would absolutely HATE living in any big city where car ownership and/or driving was impossible.
The purpose/reason/delivery of public roads is not for your recreation. So you're still free to rent out a race track or private road (they exist in many countries) for all your driving pleasure.
 

Online CatalinaWOW

  • Super Contributor
  • ***
  • Posts: 5570
  • Country: us
Re: Self driving liability
« Reply #28 on: April 23, 2022, 12:29:27 am »
This thread illustrates the dominance of emotion over logic in discussions of this topic.  Specifically the arguments about liability for speeding tickets when in self driving mode. 

In my mind there is only a very minor argument here.  In general, in self driving mode the car should never speed.  There are only two reasons I can think of for this not being the case.  First, if the vendor of the software set it that way.  In which case liability is obvious and indisputable.  The second case is if the speed limit has changed and the database available to the self driving software is not updated.  Liability here could be disputable, but this is relatively uncommon (and also likely to trip up a human driver).
You forget the third case: speed limits can be changed quickly and regulary (think about road works which can also move). Also on many roads there are dynamic speed limits depending on congestion levels. All in all it is very easy to miss a sign that tells the speed limit has changed and a database doesn't help. When using Google maps it happens very often that it indicates the wrong speed limit while there are fixed signs on a highway.

Road works are particulary nasty because the signs aren't always clear. In France for example there isn't always sign that indicates the end of the road works. You just have to deduce that from the section marked with orange cones ending. Another example are parallel roads with different speed limits. Even for a human it is easy to get confused.

The bottom line is: a self driving car will be getting speeding tickets even though the intention is not to exceed the speed limit.

BTW: I'm not against self driving cars at all. On the contrary; I think that they will give me freedom when I'm too old to drive by myself so I don't have to depend on family members or crappy bus service. And hopefully before that an opportunity to do something usefull or just enjoy the landscape during a journey instead of operating a machine that -from a basic operating point of view- hasn't really evolved for over 100 years. I do find the legal implications interesting though.

You case is exactly case two.  Liability in that case is debatable.  Even the facts are debatable.  What if the variable limit sign changes after you pass?
 

Online coppice

  • Super Contributor
  • ***
  • Posts: 10034
  • Country: gb
Re: Self driving liability
« Reply #29 on: April 23, 2022, 12:34:28 am »
You forget the third case: speed limits can be changed quickly and regulary (think about road works which can also move). Also on many roads there are dynamic speed limits depending on congestion levels. All in all it is very easy to miss a sign that tells the speed limit has changed and a database doesn't help. When using Google maps it happens very often that it indicates the wrong speed limit while there are fixed signs on a highway.

Road works are particulary nasty because the signs aren't always clear. In France for example there isn't always sign that indicates the end of the road works. You just have to deduce that from the section marked with orange cones ending. Another example are parallel roads with different speed limits. Even for a human it is easy to get confused.

The bottom line is: a self driving car will be getting speeding tickets even though the intention is not to exceed the speed limit.
Most new cars now read traffic signs. They aren't always that reliable, but if a car is going to self drive, rather than just assist the driver, it had better be able to reliably read those signs. If it can't even read something as predicable as traffic signs, how is it going to detect and react properly to less well structured things happening around it?
 

Online nctnico

  • Super Contributor
  • ***
  • Posts: 28429
  • Country: nl
    • NCT Developments
Re: Self driving liability
« Reply #30 on: April 23, 2022, 02:24:56 am »
You forget the third case: speed limits can be changed quickly and regulary (think about road works which can also move). Also on many roads there are dynamic speed limits depending on congestion levels. All in all it is very easy to miss a sign that tells the speed limit has changed and a database doesn't help. When using Google maps it happens very often that it indicates the wrong speed limit while there are fixed signs on a highway.

Road works are particulary nasty because the signs aren't always clear. In France for example there isn't always sign that indicates the end of the road works. You just have to deduce that from the section marked with orange cones ending. Another example are parallel roads with different speed limits. Even for a human it is easy to get confused.

The bottom line is: a self driving car will be getting speeding tickets even though the intention is not to exceed the speed limit.
Most new cars now read traffic signs. They aren't always that reliable, but if a car is going to self drive, rather than just assist the driver, it had better be able to reliably read those signs. If it can't even read something as predicable as traffic signs, how is it going to detect and react properly to less well structured things happening around it?
You didn't get my point: speed limit signs aren't always obvious which means that a self driving car will get a speeding ticket every now and then. With the 'driver' not being liable while the self driving system is on, who is going to pay for the ticket given the proposed laws in the first post?

Debating how well or not a self driving car can read signs is not relevant at all in this discussion because the underlying assumption is that self driving cars are certified / road legal.
« Last Edit: April 23, 2022, 02:26:49 am by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline chickenHeadKnob

  • Super Contributor
  • ***
  • Posts: 1062
  • Country: ca
Re: Self driving liability
« Reply #31 on: April 23, 2022, 06:03:44 am »
OK, what insurance company pays for this one:

Tesla which is being "smart summoned" crashes into private jet on Earth day. LoL
« Last Edit: April 23, 2022, 06:06:24 am by chickenHeadKnob »
 

Offline wilfred

  • Super Contributor
  • ***
  • Posts: 1388
  • Country: au
Re: Self driving liability
« Reply #32 on: April 23, 2022, 06:28:00 am »
OK, what insurance company pays for this one:

No insurance company will ever be paying. It will be in the premium car owners pay. It isn't simple semantics. That is the way to think about it. Similarly with arguments that suggest anyone (other than the car owner and/or driver) or any corporation will ever be paying speeding tickets or any other driving related liability arising from a self driving system. The world cannot afford the cost of the lawsuits.


Someone will have to accept the responsibility as the "driver" just like we do now when we turn the key. There is no way the legal implications will be resolved before the automation firmware is perfected (which I claim will never happen). Self driving cars on public roads is a Utopian dream that will never materialise. I can accept automated trains, mining trucks and farm tractors and maybe even excavators but none of those operate on public roads.

If self driving cars ever truly materialise they will not have a steering wheel and brake pedals. If they do then that is an admission they are imperfect and require human override. 

I claim it is a fallacy, a myth, to think self driving cars are a solution to crowded roads or bad driving or poorly designed cities over dependence on cars.
 
The following users thanked this post: Someone, MrMobodies

Offline chickenHeadKnob

  • Super Contributor
  • ***
  • Posts: 1062
  • Country: ca
Re: Self driving liability
« Reply #33 on: April 23, 2022, 08:53:07 am »
OK, what insurance company pays for this one:

No insurance company will ever be paying. It will be in the premium car owners pay. It isn't simple semantics. That is the way to think about it. Similarly with arguments that suggest anyone (other than the car owner and/or driver) or any corporation will ever be paying speeding tickets or any other driving related liability arising from a self driving system. The world cannot afford the cost of the lawsuits.
--snip--

Good point. I have always looked at property crime as an act of indirect enslavement, but never extended that logic to the moral hazard of deferred corporate negligence.

By enslavement I mean that a car thief who steals a car across  town is enslaving me through the mechanism of increased premiums.
 

Offline Ed.Kloonk

  • Super Contributor
  • ***
  • Posts: 4000
  • Country: au
  • Cat video aficionado
Re: Self driving liability
« Reply #34 on: April 23, 2022, 09:17:17 am »
Have a look how many times the nose landing gear system fails in the same aircraft verses how many times and the reasons they try and blame the pilots.

iratus parum formica
 

Online jpanhalt

  • Super Contributor
  • ***
  • Posts: 4002
  • Country: us
Re: Self driving liability
« Reply #35 on: April 23, 2022, 12:07:19 pm »
@Someone Post #27

My ISP was down for almost 12 hours yesterday.  Sorry for the late response.

I find accident data interesting as choice of denominator has a huge influence.  For aircraft operations, is it operations, passenger miles, total flight time, etc.? The Australian report uses ATM's (aircraft transport movements).  If accurate, that is probably a decent measure, but how accurate is it?  For example, a small GA aircraft may make multiple approaches with or without landing and taking off.  I don't believe those are logged by the tower, if there is one*, yet they involve the most dangerous phases of flight.

Rates per passenger mile is deceptive as scheduled carriers log far more passenger miles than GA for a single airplane.  That is, most GA flights are over shorter distances than scheduled carrier flights and have far fewer passengers per flight. 

*Unless there is an accident.  In the US, we have lots of uncontrolled airports.  The "pattern" at those airports used to be quite crowded on weekends and holidays.  Today, they are far less crowded.  One can still practice approaches with or without a tower.
 
 

Offline TerraHertz

  • Super Contributor
  • ***
  • Posts: 3958
  • Country: au
  • Why shouldn't we question everything?
    • It's not really a Blog
Re: Self driving liability
« Reply #36 on: April 23, 2022, 01:58:24 pm »
The best thing about self driving cars, is how they are an excellent intelligence test for people. Ask anyone if they think self driving cars are a good idea. In a few moments you'll know whether they are sane or a delusional, amoral lunatic. Whether they understand the nature of consciousness and the importance of having a comprehensive model of the world around them. Whether they understand the dilema of 'strong AI' at all, or think a world with AI slaves would be just peachy.

Anyway, it's all very entertaining (except for side effects like car insurance rates going up.)
Here's an example of the kind of expensive comedy that self-driving cars (with less than strong AI) will keep on giving:



Tesla on Autopilot Crashes Into $3 Million Jet


I particularly enjoyed how the car seemed to think "Wait, did I just hit something? Nah, it's all good, I'll keep going. Uh, no, something is definitely wrong."

Then there was that other hilarious case (last year?) where a Tesla car on highway autopilot decapitated it's occupant by trying to drive under a semi that was pulling out of a side road.

If cars end up being given strong AI, then there will be a whole different world of problems.
And they _would_ be given strong AI as soon as it's developed. Except for other more immediate world factors that will make all such future issues moot.
« Last Edit: April 23, 2022, 02:59:48 pm by TerraHertz »
Collecting old scopes, logic analyzers, and unfinished projects. http://everist.org
 

Online CatalinaWOW

  • Super Contributor
  • ***
  • Posts: 5570
  • Country: us
Re: Self driving liability
« Reply #37 on: April 23, 2022, 02:28:45 pm »
There are certainly many examples of self driven cars doing stunningly stupid things.  Unfortunately the same is true for human drivers.  I know two people who drove at high speed into objects (not suicides).  That is within a group of people I know well enough to have gone to dinner or attended the same party.  A group that totals at most a few thousand people.

Any claim for superiority of one method to another is very weakly supported.
 

Offline TerraHertz

  • Super Contributor
  • ***
  • Posts: 3958
  • Country: au
  • Why shouldn't we question everything?
    • It's not really a Blog
Re: Self driving liability
« Reply #38 on: April 23, 2022, 02:57:01 pm »
Did I say human drivers don't do stupid things? Nope. Just today I was a few meters away (on foot) from a near accident where a human driver did something very stupid, and the two cars involved screeched to a stop about a foot apart.

But it's a logical fallacy to argue that because human drivers screw up, we should allow 'AI' self-driving cars.
Even if somehow self-driving cars managed to have fewer accidents per mile driven, less injuries, etc, it would still be a very bad idea.
Collecting old scopes, logic analyzers, and unfinished projects. http://everist.org
 

Offline eugene

  • Frequent Contributor
  • **
  • Posts: 497
  • Country: us
Re: Self driving liability
« Reply #39 on: April 23, 2022, 09:42:12 pm »
Did I say human drivers don't do stupid things? Nope. Just today I was a few meters away (on foot) from a near accident where a human driver did something very stupid, and the two cars involved screeched to a stop about a foot apart.

But it's a logical fallacy to argue that because human drivers screw up, we should allow 'AI' self-driving cars.
Even if somehow self-driving cars managed to have fewer accidents per mile driven, less injuries, etc, it would still be a very bad idea.

Don't mistake me for an AI fanboi, but I don't understand how anything that makes traveling safer can be a "very bad idea."

In any case, I'm certain that I see the entire topic differently than most of the contributors to this thread. (I find it comical that a bunch of off-duty engineers think they understand the issues better than the on-duty engineers working in the auto industry.) Only nutjobs like Elon Musk think it's a good idea to make a car that is completely driverless and sell it today. On the other hand, ALL of the major auto manufacturers are investing big money and effort on the topic in a more general way. None of GM, VAG or Toyota are going to be offering cars that drive around a tarmac and crash into anything in the way. The reason is obvious and the major topic of this thread: liability.

But, they are doing all of the research that their enormous resources allow. And in the course of their research they will develop all sorts of ways to build cars that assist the driver, possibly in surprising ways. The consequence can only be that the cars will be MORE safe. No major auto manufacturer (Telsla not withstanding) is going to introduce features that make traveling less safe. It's just not part of their business plan.

Anyway, all of this assisted driving will eventually result in auto-driving. In the meantime (i.e. today) we can have fun talking about the what ifs, but it's all essentially hypothetical because unless Google starts making cars, Tesla will be the only example we have for discussion, and they aren't even close to typical.
90% of quoted statistics are fictional
 

Online nctnico

  • Super Contributor
  • ***
  • Posts: 28429
  • Country: nl
    • NCT Developments
Re: Self driving liability
« Reply #40 on: April 23, 2022, 10:15:03 pm »
Did I say human drivers don't do stupid things? Nope. Just today I was a few meters away (on foot) from a near accident where a human driver did something very stupid, and the two cars involved screeched to a stop about a foot apart.

But it's a logical fallacy to argue that because human drivers screw up, we should allow 'AI' self-driving cars.
Even if somehow self-driving cars managed to have fewer accidents per mile driven, less injuries, etc, it would still be a very bad idea.

Don't mistake me for an AI fanboi, but I don't understand how anything that makes traveling safer can be a "very bad idea."

In any case, I'm certain that I see the entire topic differently than most of the contributors to this thread. (I find it comical that a bunch of off-duty engineers think they understand the issues better than the on-duty engineers working in the auto industry.) Only nutjobs like Elon Musk think it's a good idea to make a car that is completely driverless and sell it today. On the other hand, ALL of the major auto manufacturers are investing big money and effort on the topic in a more general way. None of GM, VAG or Toyota are going to be offering cars that drive around a tarmac and crash into anything in the way. The reason is obvious and the major topic of this thread: liability.

But, they are doing all of the research that their enormous resources allow. And in the course of their research they will develop all sorts of ways to build cars that assist the driver, possibly in surprising ways. The consequence can only be that the cars will be MORE safe. No major auto manufacturer (Telsla not withstanding) is going to introduce features that make traveling less safe. It's just not part of their business plan.

Anyway, all of this assisted driving will eventually result in auto-driving. In the meantime (i.e. today) we can have fun talking about the what ifs, but it's all essentially hypothetical because unless Google starts making cars, Tesla will be the only example we have for discussion, and they aren't even close to typical.
The latter is not quite right. Tesla drums up a lot of hype about their brand but the reality is that their assisted driving system (currently SAE level 2) is severly lagging behind on the competition. For example: Mercedes offers 'Drive pilot' which is an SAE level 3 automated driving system on their S class models for use in Germany. BMW is close to introducing a SAE level 3 automated driving system but seems to struggle with the regulatory framework in the EU: https://electricvehicleweb.com/bmw-ix-level-3-autonomous-driving/

Getting the legal framework in order soon is important because there will be more and more cars on the road with automated driving systems.
« Last Edit: April 23, 2022, 10:21:30 pm by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline eugene

  • Frequent Contributor
  • **
  • Posts: 497
  • Country: us
Re: Self driving liability
« Reply #41 on: April 23, 2022, 11:32:04 pm »
Thank you. I stand corrected.

I still contend that, at least in the next decade, assisted driving will play a much larger part than fully autonomous driving. In some case, like an imminent collision, the car might take over completely from the driver, but that's still a long way from the sci-fi notion of putting your kids in the car and telling it to take them to school, then come right back.

I guess that I am ironically responding emotionally to other's emotions.
90% of quoted statistics are fictional
 
The following users thanked this post: nctnico

Offline Someone

  • Super Contributor
  • ***
  • Posts: 5155
  • Country: au
    • send complaints here
Re: Self driving liability
« Reply #42 on: April 24, 2022, 12:30:33 am »
Rates per passenger mile is deceptive as scheduled carriers log far more passenger miles than GA for a single airplane.  That is, most GA flights are over shorter distances than scheduled carrier flights and have far fewer passengers per flight.
Yes, typically shorter distances for GA does somewhat bias their safety figures in the negative, as does hours of aircraft operation. But that alone still doesn't explain the enormous differences in fatality rates between the different modes (having already removed the very very risky agricultural operations from the figures). I suggest most of it are the additional "restrictive" rules that the large airlines enforce, which will be a similar situation when self driving cars arrive.

Big businesses with large exposure will be much more risk averse than an "insured" individual who doesn't see significant loss from their (in)action. So expect self driving cars that accept the legal liability/responsibility to be much more cautious than human drivers who are exposed to little liability. Offloading externalities of cars to government backed/mandatory insurance has been a huge win for private motoring (similar to the nuclear industry).
 

Offline sleemanj

  • Super Contributor
  • ***
  • Posts: 3051
  • Country: nz
  • Professional tightwad.
    • The electronics hobby components I sell.
Re: Self driving liability
« Reply #43 on: April 24, 2022, 02:09:00 am »

   This times 100,000!  IMO if the "rider" wants to go somewhere and doesn't want to "drive" then they should take a cab or a bus!


Which may also be driverless...



~~~
EEVBlog Members - get yourself 10% discount off all my electronic components for sale just use the Buy Direct links and use Coupon Code "eevblog" during checkout.  Shipping from New Zealand, international orders welcome :-)
 

Offline TerraHertz

  • Super Contributor
  • ***
  • Posts: 3958
  • Country: au
  • Why shouldn't we question everything?
    • It's not really a Blog
Re: Self driving liability
« Reply #44 on: April 24, 2022, 09:25:15 am »
On the topic of 'AI' astonishing advances:



Adding- this too: https://this-person-does-not-exist.com/en
« Last Edit: April 24, 2022, 11:06:10 am by TerraHertz »
Collecting old scopes, logic analyzers, and unfinished projects. http://everist.org
 
The following users thanked this post: Ed.Kloonk


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf