Waymo got a ticket and there was confusion about who or what was being ticketed(twitter.com)
twitter.com
Waymo got a ticket and there was confusion about who or what was being ticketed
https://twitter.com/antoniogm/status/1822141923538161776
65 comments
This situation is not any different from a ticket issued by a speed camera. Since the camera can’t tell who is driving, the fine is issued to the owner of the vehicle (determined by the license plate). The owner is legally responsible for the fine. The only difference between a speed camera ticket and a ticket given by an officer in person is that the speed camera does not result in penalties for driving infraction (eg points on a license or license suspension) only fines.
A ticket to a driverless taxi should work the same way; the citation goes to the owner of the vehicle.
Is this really how it works in America? In the UK the notice of infraction is sent to the registered keeper of the car, obviously specifying where and then the offense occurred, and you then have the choice of telling the police who was driving the car or going to court. And, since it's your car and you're responsible for it, you'd better have a good reason for being unable or unwilling to say who was driving it at that time. There is no question whatsoever of a camera being unable to deliver points on the license.
Other than the fact that there are 50+ states here, and that the rules are different in each one, this is mostly correct. Ultimately corporations have a lot of privilege and deference compared to individuals. And traffic cameras are run by companies who often provide some sort of fee to the city.
This answer makes sense. But this is the law we're talking about, so I'm not sure about it.
this is incorrect (for us law). The operator of the vehicle is legally responsible for the ticket. An owner can show up in court and say “I don’t know who was driving my car at that time and place”. That’s why traffic cameras take a very clear photograph of the driver.
Assign one safety driver per vehicle, the remote safety driver is responsible (like a driving instructor and its driving student, except the AI is the student).
This way, the internal safety drivers will do everything they can to have the company act responsibly, and refuse the work if the system is not reasonably safe.
It can also solve other problems:
In Estonia we have delivery "robots". On the paper it is "AI", but in reality they are actually remotely controlled by students aka "Robot operators".
The law makes the other drivers guilty if you hit one of these "AI-robots", which really sucks, because the law was changed due to intense lobbying.
I'd rather have people designated as drivers to be responsible.
This way, the internal safety drivers will do everything they can to have the company act responsibly, and refuse the work if the system is not reasonably safe.
It can also solve other problems:
In Estonia we have delivery "robots". On the paper it is "AI", but in reality they are actually remotely controlled by students aka "Robot operators".
The law makes the other drivers guilty if you hit one of these "AI-robots", which really sucks, because the law was changed due to intense lobbying.
I'd rather have people designated as drivers to be responsible.
The point of Waymo is that human intervention is only used in exceptional circumstances. There isn't a one-to-one mapping on purpose, because it's not needed in a system generally controlled by AI.
Yes I agree, Waymo is real, and very nice tech, but if there is such high confidence, couldn't we find drivers who can remotely monitor the rides for safety ?
It could potentially solve the double-speech: "oh no no no, never willing to be the legal driver it's not a safe system", and at the same time: "oh but this system is very safe, we can put it in real traffic with real lives on the line".
It could potentially solve the double-speech: "oh no no no, never willing to be the legal driver it's not a safe system", and at the same time: "oh but this system is very safe, we can put it in real traffic with real lives on the line".
Even if the system is safer than human drivers (I can’t speak to this), why would the lowest paid workers at the company voluntarily sign up for _any_ increased liability? If one of the programmers pushes out a bug that kills someone, why should an employee who had nothing to do with this be charged with a crime?
You raise a good point - if the system is so safe, the people who make such claims should accept some liability - but those are not the same people as the ones who sometimes intervene in the driving.
You raise a good point - if the system is so safe, the people who make such claims should accept some liability - but those are not the same people as the ones who sometimes intervene in the driving.
A separate job, remote safety drivers, who are responsible for monitoring the rides and acting (aka braking) in case of danger or mistakes.
Except from home / office.
Except from home / office.
You're asking for... what? For the secretary to be liable because a random car they have never seen or interacted with broke a law? How is that in any way shape or form reasonable? And why would anybody ever sign that contract except under duress or exceptional circumstances when they can get a job... literally anywhere else?
And even if that was a viable approach right now, the point of AI/self-driving technology is that the number of cars will vastly outnumber the number of employees once fully deployed and not in limited pilot programs.
And even if that was a viable approach right now, the point of AI/self-driving technology is that the number of cars will vastly outnumber the number of employees once fully deployed and not in limited pilot programs.
The responsibility of this safety driver is to monitor the ride and be responsible in case of accident.
It's his main task, like any driver, just he has to pay attention to less things than a standard ride, since he has a very smart car (like when you sit in a Tesla with FSD).
Just he is not sitting in the car, instead in his office or home.
It's his main task, like any driver, just he has to pay attention to less things than a standard ride, since he has a very smart car (like when you sit in a Tesla with FSD).
Just he is not sitting in the car, instead in his office or home.
I basically covered this already in my previous comment, so I'm just going to restate it.
> why would anybody ever sign that contract except under duress or exceptional circumstances when they can get a job... literally anywhere else?
> And even if that was a viable approach right now, the point of AI/self-driving technology is that the number of cars will vastly outnumber the number of employees once fully deployed and not in limited pilot programs.
> why would anybody ever sign that contract except under duress or exceptional circumstances when they can get a job... literally anywhere else?
> And even if that was a viable approach right now, the point of AI/self-driving technology is that the number of cars will vastly outnumber the number of employees once fully deployed and not in limited pilot programs.
I understand what you mean, though drivers already take personal risks on the behalf of corporations.
If you are a Uber-driver, and enable the FSD on your Tesla, it's not Uber nor Tesla who will get in troubles if you are having an accident.
For now the situation with Waymo, can be the same as a driving instructor. The remote driver is the driving instructor, and Waymo AI is the student getting monitored and doing the driving.
The driver is responsible and, then they can potentially sue (or have agreement with Waymo) if it caused damages beyond their own responsibility (for example, if driver hits the brakes, and Waymo has overridden his manual commands).
Once this solution is safe, drivers will get more comfortable babysitting more than one car at once; essentially solving the scale issue that you mention.
If you are a Uber-driver, and enable the FSD on your Tesla, it's not Uber nor Tesla who will get in troubles if you are having an accident.
For now the situation with Waymo, can be the same as a driving instructor. The remote driver is the driving instructor, and Waymo AI is the student getting monitored and doing the driving.
The driver is responsible and, then they can potentially sue (or have agreement with Waymo) if it caused damages beyond their own responsibility (for example, if driver hits the brakes, and Waymo has overridden his manual commands).
Once this solution is safe, drivers will get more comfortable babysitting more than one car at once; essentially solving the scale issue that you mention.
Twitter op responded with this, idk if the cop was just having a laugh or what but kinda lost interest due to this.
>The whole situation was a societal Turing test: the car was waiting in a bike lane for me.
A parking enforcement officer rolls up and thinks someone parked there. But it wasn’t parked: software was driving and waiting for me. A human wouldn’t be ticketed, so why would AI be?
>The whole situation was a societal Turing test: the car was waiting in a bike lane for me.
A parking enforcement officer rolls up and thinks someone parked there. But it wasn’t parked: software was driving and waiting for me. A human wouldn’t be ticketed, so why would AI be?
Wouldn’t human can be legally ticketed for stopping in the bike lane?
Technically they should be. In practice they aren’t. Really, American cities just do bike lanes so wrong that everyone involved just throws up their hands and anything goes.
Strongly beg to differ. Blocking a bike lane is $100 here, ticketing the lanes of cars parked in them can and should provide a nice revenue stream to the city
I get it, I really do. But they often throw bike lanes into formal waiting lanes and then try to have both at the same time. It’s just crappy road design, and traffic officers have trouble figuring put what law should apply when they contradict each other. If the cops start fining drivers for reasonable things, then lots of people complain and the cops are told to back off.
If you want bike lanes, get rid of parking and waiting lanes all together and physically protect or grade the lanes. Don’t half ass it and try to throw car traffic and bikes together in constant conflict. It doesn’t work out, and no amount of righteous fury will make it work out.
If you want bike lanes, get rid of parking and waiting lanes all together and physically protect or grade the lanes. Don’t half ass it and try to throw car traffic and bikes together in constant conflict. It doesn’t work out, and no amount of righteous fury will make it work out.
American law enforcement is highly variable and defers to the police officers’ individual judgments to write a ticket.
In some American cities like NYC a lot of the moving and parking violations are actually cops, who don’t ticket each other out of solidarity even if not actually on the job, and they often don’t ticket people exhibiting similar behaviors that they do.
The practice of illegally parking on a sidewalk by cops in NYC is so endemic the DOJ warned it might sue NYPD for a systematic violation of the ADA. https://www.nbcnewyork.com/news/local/doj-threatens-nypd-wit...
In some American cities like NYC a lot of the moving and parking violations are actually cops, who don’t ticket each other out of solidarity even if not actually on the job, and they often don’t ticket people exhibiting similar behaviors that they do.
The practice of illegally parking on a sidewalk by cops in NYC is so endemic the DOJ warned it might sue NYPD for a systematic violation of the ADA. https://www.nbcnewyork.com/news/local/doj-threatens-nypd-wit...
That reminds me of why Beijing traffic behavior is so much worse than Shanghai: Beijing is full up of powerful people or people connected to powerful people, so traffic rule enforcement is really risky unless done by machine or by the party police. Shanghai, in comparison, is just rich.
I know zero about Shanghai but the first time in visited in August 2008, my friend that lived there said if I had visited just 2 months earlier it would have been a totally different experience because I'd have heard incessant car honking. He claimed the government banned honking your horn except in an emergency and then actually enforced it so that everyone got the message quick. Honk, get a fine.
I was impressed since I didn't hear any honking. I can't imagine that being enforced pretty much anywhere else in the world.
I was impressed since I didn't hear any honking. I can't imagine that being enforced pretty much anywhere else in the world.
Some cities physically removed car horns, I think Chongqing did that.
I don’t remember hearing much honking in Beijing either during my 9 years there.
I don’t remember hearing much honking in Beijing either during my 9 years there.
There is pushback against automated camera enforcement in the US, partially because of worries of reducing union membership and power. This isn’t likely because there already is a police hiring shortfall in major urban areas despite lackluster traffic enforcement, and it would be impossible to find enough cops to police every single motor violation.
But also, it would remove police discretionary power to turn a blind eye to friends, family and colleagues.
But also, it would remove police discretionary power to turn a blind eye to friends, family and colleagues.
There is a lot of pushback on camera enforcement because people don’t like getting tickets and US government doesn’t have that much power compared to say Australia or Canada. That’s not even considering police unions, they don’t have to lift a finger here since most Americans already won’t let it happen. Maybe if there is less public resistance, the police unions will have to chime in on the issue.
We obviously can’t afford to pay cops $200k+ a year without a huge increase in productivity, and only tech can provide that. It’s only a matter of time for the Americans to come around to it.
We obviously can’t afford to pay cops $200k+ a year without a huge increase in productivity, and only tech can provide that. It’s only a matter of time for the Americans to come around to it.
> There is pushback against automated camera enforcement in the US, partially because of worries of reducing union membership and power.
There is pushback in the US because the cameras got installed as a profit making exercise instead of a safety improvement.
Yellow-light cameras, for example, magically got installed on busy intersections instead of dangerous ones. And then the yellow light timings got trimmed down when they weren't making enough money.
There is pushback in the US because the cameras got installed as a profit making exercise instead of a safety improvement.
Yellow-light cameras, for example, magically got installed on busy intersections instead of dangerous ones. And then the yellow light timings got trimmed down when they weren't making enough money.
People repeat this but this really depends on where you were, and then people take these blanket arguments that don’t apply to other states’ traffic regulations.
For example this would not be an issue in a state that enshrines the minimum time of a yellow light in state law, and these issues occurred in jurisdictions where that is not true.
There are also two kinds of red light cameras; those to catch dangerous red light running, and those to catch violations of the “block the box” variety. The latter isn’t necessarily dangerous but does gum up traffic flow unnecessarily.
For example this would not be an issue in a state that enshrines the minimum time of a yellow light in state law, and these issues occurred in jurisdictions where that is not true.
There are also two kinds of red light cameras; those to catch dangerous red light running, and those to catch violations of the “block the box” variety. The latter isn’t necessarily dangerous but does gum up traffic flow unnecessarily.
Do a lot of states enshrine minimum yellow (and I assume green and red too) times? I figured that wouldn't be part of law but at best a DMV regulation type thing.
As far as I know, no state legislature in this country has completely given up traffic code rulemaking, so legislatively passed rules would always take primacy, and if you wanted that enshrined you could get it done. After all, DMVs don’t have the power to set criminal penalty standards for things like DUI, seatbelt laws, etc. And traffic cameras usually have to be enabled by state legislation since they result in penalties, so the same law could also involve minimum standards on yellow lights.
It normally is a state DOT (department of transportation) regulation rather than a legislative law.
The timings didn't really need much regulation as they were generally set by Professional Engineers(PEs) who had to sign off that they analyzed the situation and it was safe. Those PEs would generally take into account the speed limit as well as actual car speeds, the general road conditions, average human reaction times, and average car braking characteristics so that you had sufficient time to stop when the light turned yellow. Normally they got it right, but occasionally they missed a bit (actual traffic speeding more than expected) and you got a "short yellow" which generally would get corrected if enough people complained.
And then some bright bulb got the idea to install enforcement cameras on the intersections--which doesn't sound like a bad idea. Except that the issue is the revenue goes to the police department AND gets split with the company maintaining the cameras--yeah, you can see where this is going a MILE away.
The cameras initially got installed on a couple of the most dangerous intersections and genuinely helped--accidents went way down on those. People were genuinely pleased about this. But, see, that's an issue as the revenue is now below projections. So, then the cameras started getting installed on super busy intersections instead of dangerous ones that were literally blocks away. At this point people started getting pissed off but could still be deflected with "Don't do illegal crap at the intersection". But, you see, revenue was still below projections because people adjust their behaviors--they either go a different route or adjust to avoid the camera firing.
It was at this point that they shortened the yellow lights to bump the revenue and where the shit hit the fan. The yellow lights were sufficiently short that you couldn't stop legally, accidents at the intersections increased because of rear-ending, people could document all this, and discovery pulled up the profit sharing agreements between the camera vendors and the police. At this point, everybody was pissed off and the cameras all got yanked because it was now an electoral issue.
I got to watch this occur in real time in San Diego, but they were far from the only municipality to pull this garbage.
The timings didn't really need much regulation as they were generally set by Professional Engineers(PEs) who had to sign off that they analyzed the situation and it was safe. Those PEs would generally take into account the speed limit as well as actual car speeds, the general road conditions, average human reaction times, and average car braking characteristics so that you had sufficient time to stop when the light turned yellow. Normally they got it right, but occasionally they missed a bit (actual traffic speeding more than expected) and you got a "short yellow" which generally would get corrected if enough people complained.
And then some bright bulb got the idea to install enforcement cameras on the intersections--which doesn't sound like a bad idea. Except that the issue is the revenue goes to the police department AND gets split with the company maintaining the cameras--yeah, you can see where this is going a MILE away.
The cameras initially got installed on a couple of the most dangerous intersections and genuinely helped--accidents went way down on those. People were genuinely pleased about this. But, see, that's an issue as the revenue is now below projections. So, then the cameras started getting installed on super busy intersections instead of dangerous ones that were literally blocks away. At this point people started getting pissed off but could still be deflected with "Don't do illegal crap at the intersection". But, you see, revenue was still below projections because people adjust their behaviors--they either go a different route or adjust to avoid the camera firing.
It was at this point that they shortened the yellow lights to bump the revenue and where the shit hit the fan. The yellow lights were sufficiently short that you couldn't stop legally, accidents at the intersections increased because of rear-ending, people could document all this, and discovery pulled up the profit sharing agreements between the camera vendors and the police. At this point, everybody was pissed off and the cameras all got yanked because it was now an electoral issue.
I got to watch this occur in real time in San Diego, but they were far from the only municipality to pull this garbage.
Just install them on all intersections and then people won’t be able to insinuate bias anymore. The tech is cheap enough at this point, the problem is that private companies are involved at all when they really don’t need to be. The same with speed cameras, you could just blanket them everywhere at this point, and if not, in ten years you definitely can.
This isn’t really the main issue though. People want to get away with speeding and running red lights, and these people vote so…
This isn’t really the main issue though. People want to get away with speeding and running red lights, and these people vote so…
> Blocking a bike lane is $100 here, ticketing the lanes of cars parked in them can and should provide a nice revenue stream to the city
But do they?
But do they?
double parking and stopping in bike lanes has become the norm in many American cities. It's illegal and it often causes accidents since it forces people to make dangerous lane changes and backs up traffic into intersections. But, the police are not enforcing it so here we are.
Maybe we'll get AI driven police cars that film and ticket double parkers? I doubt it.
What's extremely frustrating though is that double parking and stopping in bike lanes is so normalized that a driver will do it, even though there's free space they could have actually used with 1 or 2 car lengths or even right at the stop they're stopping. They just don't give a fuck and just stop in the lane to deliver a package etc...
Maybe we'll get AI driven police cars that film and ticket double parkers? I doubt it.
What's extremely frustrating though is that double parking and stopping in bike lanes is so normalized that a driver will do it, even though there's free space they could have actually used with 1 or 2 car lengths or even right at the stop they're stopping. They just don't give a fuck and just stop in the lane to deliver a package etc...
>Maybe we'll get AI driven police cars that film and ticket
You don't really need that. London is covered by cameras that will pick that stuff up and send you a ticket in the post with a link to a photo of the offence.
You don't really need that. London is covered by cameras that will pick that stuff up and send you a ticket in the post with a link to a photo of the offence.
Why wouldn't a human driver be ticketed for blocking a bike lane?
In SF licensed Taxicabs can pick up / drop off in bike lanes.
I learned this by being overly aggro to a taxi driver when I was a cyclist in SF. Credit to the driver for calmly talking me through the rules.
http://www.sfbike.org/images/taxi_truck_flyers/taxi_flyer_FI...
I learned this by being overly aggro to a taxi driver when I was a cyclist in SF. Credit to the driver for calmly talking me through the rules.
http://www.sfbike.org/images/taxi_truck_flyers/taxi_flyer_FI...
The human could discuss with the officer and say they were waiting for someone or be asked to move on out of the bike lane. The difference being the human interaction.
There's a difference between parking and standing.
They both block the bike lane
OK? I have no idea about the particular municipal laws about temporarily blocking bike lanes, but that's exactly the difference the OP was describing.
For different amounts of time. You'd want to add more specific timelines for a driverless car, but there's a real difference and there's a reason lots of places allow standing but not parking.
There's a full bodycam video here[1] of an officer in Phoenix pulling over a Waymo. The driver's window gets rolled down and he gets connected via voice to a Waymo agent and they're able to discuss what was going on. Waymo then kept the car parked there and sent someone out.
I'm really curious what the full interaction was like for this SF Waymo. The QR code on the window just links to a couple phone numbers[2] which seems insufficient to me in the case where a car is merely ticketed but not formally pulled over.
[1] https://www.reddit.com/r/Damnthatsinteresting/comments/1dvso... [2] https://waymo.com/emergency-response/
I'm really curious what the full interaction was like for this SF Waymo. The QR code on the window just links to a couple phone numbers[2] which seems insufficient to me in the case where a car is merely ticketed but not formally pulled over.
[1] https://www.reddit.com/r/Damnthatsinteresting/comments/1dvso... [2] https://waymo.com/emergency-response/
I read a blog post years ago about a guy who fought a traffic ticket and ran into the undefined legal area of whether they were laying a charge against him or his vehicle, since they couldn't prove it was him driving. I imagine it may be a similar scenario here, but unfortunately I cannot open Twitter links on mobile.
I don't know in the US but in the UK the legals are fairly defined. For parking the owner is liable. For driving the driver is and you have the option of paying a fixed penalty, saying who did it or going to court and saying it wasn't me. The latter is dangerous to game as if they can show you were lying it's a worse offence than just paying the fine. Although I've got a friend of Pakistani origin who's gotten away with saying it was various similar looking Pakistani cousins who were speeding.
Actually what is supposed to happen here and how the fuck are they allowed on streets without simple rules like this in place?
It's entirely possible the rules are indeed established and the officer just doesn't know.
Nope. They are out there with no laws in place. https://amp.theguardian.com/technology/2024/jan/04/self-driv...
That article makes it seem as if there are policies in place. They may well not be fully thought out but they seem to be there from the SFPD.
And: why the hell did the "FSD" vehicle get a ticket?
If it can't complete a ride without getting tickets, then how do we know it's safe?
If it can't complete a ride without getting tickets, then how do we know it's safe?
It seems obvious it's the car--or more precisely the operator of the car, i.e. Waymo. Certainly, as a passenger, I'm not responsible for anything my taxi or rideshare does--absent specific behavior on my part that causes an accident or other violation of course.
It doesn’t seem that obvious to me.
Let’s say you sit in the driver’s seat of your friend’s Tesla and turn on autopilot. Are you the driver or passenger?
Now let’s say you sit in the driver’s seat of a Waymo. Are you the driver or passenger?
Let’s say you sit in the driver’s seat of your friend’s Tesla and turn on autopilot. Are you the driver or passenger?
Now let’s say you sit in the driver’s seat of a Waymo. Are you the driver or passenger?
In a taxi, the passenger is not the liable party.
Teslas aren't approved to operate as taxis, whereas Waymos explicitly are. I think that's the difference here.
Teslas aren't approved to operate as taxis, whereas Waymos explicitly are. I think that's the difference here.
Yes. If you have a "FSD" Tesla, you're still supposed to be in the driver's seat and are the responsible driver--whatever claims the company makes to the contrary.
Your hypothetical is about an entirely different situation
One of them forces you to engage with the controls, and the other will not let you engage with the controls.
I can imagine ambiguous situations, but I don't see how either of these two are ambiguous. As long as you can prove it.
Though the easy answer is to not sit in the driver's seat when not driving.
I can imagine ambiguous situations, but I don't see how either of these two are ambiguous. As long as you can prove it.
Though the easy answer is to not sit in the driver's seat when not driving.
Makes me wonder what happens if a self driving car kills somebody because of a fault on the vehicle software or hardware. Can the big corporation behind it just pay a fine and nobody else will get prosecuted? How do you prove that a corporation was negligent?
This isn't really a new thing in our society, just because of driverless cars.
Big corporations (or even small ones) kill people accidentally all the time.
Like when planes or trains malfunction. Or there is a construction site accident. Oil spills. Factory fire. Mine collapse.
People rarely go to jail (and I don't think they should in most cases), and a civil suit pays out victims depending on the exactly how negligent the corporation was. You can see Boeing going through this process now
Big corporations (or even small ones) kill people accidentally all the time.
Like when planes or trains malfunction. Or there is a construction site accident. Oil spills. Factory fire. Mine collapse.
People rarely go to jail (and I don't think they should in most cases), and a civil suit pays out victims depending on the exactly how negligent the corporation was. You can see Boeing going through this process now
Well, that's how it works when when humans kills others by accident. You don't go to jail, you might not even get a fine. It's just a fact of life we live with in our car-centric world.
It's when negligence gets involved is whwre I have questions.
It's when negligence gets involved is whwre I have questions.
True, but only if you are a large corporation, because if you are an individual you are in deep trouble.
It is not as serious as you think despite the seriousness of the topic. If it was a true accident, no negligence, etc it would be very common to get away with just a fine or a ticket, not even going to jail. We (USA) are a culture very accustomed to cars and what they impose on society. A friends relative got killed after a car lost control on an embankment turn and crashed into their living room - driver walked away unscathed and without even a ticket.
Probably. Otherwise, absent a paper trail that senior executives knew the product was unsafe and deliberately concealed it, you're chasing after junior people who made a mistake somewhere--perhaps unknowable. Drug side effects kill people all the time but, absent effects that should have been foreseen, they're considered just part of how things are.
I imagine waymo has a government relations team, including traffic control relations.
So do cops receive training on these situations? Its only going to get worse (probably).
So do cops receive training on these situations? Its only going to get worse (probably).
So much for AV never makes mistakes.