I live in one of the areas they are actively testing/training in. Their cars consistently behave better and more safely than most human drivers that I’m forced to share the road with.
As semi-autonomous and autonomous cars become the norm, I would adore to see obtaining a drivers license ratchet up in difficulty in order to remove dangerous human drivers from the road.
Traffic enforcement, which used to correct some bad driving, has basically evaporated in many parts of the U.S. This has been a long-term trend.
A friend who's a cop told me that only when their department got specific state grants would they set up stings of drivers driving in a pedestrian walkway while someone was crossing the street. He said he would only pull someone over if they were doing something batshit crazy and they happened to be behind the vehicle where it was easy to pull them over. Minor stuff and speeding they would almost never ticket.
The U.S. and other countries need to use automated methods of detecting and applying penalties. Some busy intersections have cameras for this, but it seems to be very limited, maybe because of cost.
Years ago New York used to calculate if you were speeding the NY State Thruway based on the time between toll booths. They cancelled this program for some reason.
The New York State Police have speed cameras set up in a few highway work zones, which is effective (double fines applicable) but it still requires a person driving a car to set up the gear.
The real issue is all the current bad drivers. A requirement to start re-testing normal people in addition to the elderly would be a large benefit to society.
I'm from the UK, took driving lessons in the UK but then passed my driving test in the USA (in California).
The USA driving test is so much easier than the UK one!
UK: Varied junctions and roundabouts, traffic lights, independent driving (≈20 minutes via sat nav or signs), one reversing manoeuvre (parallel park, bay park, or pull up on the right and reverse), normal stops and move-offs (including from behind a parked car), hill start, emergency stop.
California: Cross three intersections, three right turns, three left turns, lane change, backing up, park in a bay, obey stop signs and traffic lights.
My understanding is that the USA test is so much easier because it's hard to get by in most of the USA without a car, so if the test was harder people would likely just drive without a license instead.
I would support re-testing on some interval like every 5 years. That said, so much could be done to make the environment safer. Lower speeds, more traffic calming, safer intersections, safer alternatives (public transit, walking, bicycle).
I can't help but think about the failures of basic human-oriented infrastructure when I can't safely ride my bike to the grocery store 2 miles from my home. I don't know what it'll take to change this in our cities, and it feels like an uphill battle when seemingly very few people care about problems like these.
Everyone agrees to this, the problem is there needs to be a way for this to be done efficiently so it's not another regressive tax on poor people's time and money.
I think a lot about this (bad drivers) and I’m not really sure how to fix it since I think it’s really a problem of underlying selfishness and perceived-exceptionalism mixed with overestimation of skill.
If you actually ride in one, you do notice some off behaviors that I didn't pick up while just driving alongside them. That said, I agree that the bad human drivers have done things far, far worse than any of the cars.
The biggest gripe with riding in one is that they're slow, both because of super cautious driving and because they won't take freeways yet.
A month ago I saw a Waymo turn left into a tiny alley in Palo Alto and continue at full 25mph speed, which was alarming. I guess the alley is marked as a regular road in the software? Highlights how even if it's safer than humans on average, they need to minimize these weird behaviors in order to get socially accepted and avoid $$$ liability when there is an accident.
I have only taken a couple Waymos but I had the opposite experience. They were much faster and more decisive than I expected. They do apparently learn from surrounding drivers and this was LA so maybe that explains the difference.
I live in the bay and occasionally ride Waymo in SF and I pretty much always have a good time.
I visited NYC a few weeks ago and was instantly reminded of how much the traffic fucking sucks :) While I was there I actually thought of Waymo and how they'd have to turn up the "aggression" slider up to 11 to get anything done there. I mean, could you imagine the audacity of actually not driving into an intersection when the light is yellow and you know you're going to block the crossing traffic?
Semi-related, but just once in my life, I want to hear a mayoral candidate say: “I endorse broken windows theory, but for drivers. You honk when there’s no emergency, block the box, roll through a stop sign — buddy that’s a ticket. Do it enough and we’ll impound your car.”
Who knows, maybe we’ll start taking our cues from our polite new robot driver friends…
This always astounds me about cities who have a reputation for people breaking certain traffic laws. In St. Louis, people run red lights for 5+ seconds after it turns red, and no one seems to care to solve it, but if they'd just station police at some worst-offender lights for a couple months to write tickets, people would catch on pretty quickly that it's not worth the risk. I have similar thoughts on people using their phones at red lights and people running stop signs.
It’s amazing how effective even a slight amount of random law enforcement can be.
Several of the hiking trails I frequent allow dogs but only on leash. Over time the number of dogs running around off leash grows until it’s nearly every dog you see.
When the city starts putting someone at the trailhead at random times to write tickets for people coming down the trail with off-leash dogs suddenly most dogs are back on leash again. Then they stop enforcing it and the number of off-leash dogs starts growing.
Random sampling over time is substantially as effective as having someone enforce the law 100% of the time. It's something like how randomized algorithms can be faster than their purely-deterministic counterparts, or how sampling a population is quite effective at finding population statistics.
It feels less fair though. When everyone is driving x mph over the limit but only you get pulled over, it sucks. So I agree for efficiency of enforcement, but I'd rather see 100% enforcement (automated if possible), with more warnings and lower penalties.
It doesn't just "feel" less fair, it often is -- bc it's not truly random, it's selective enforcement which leads to things like "driving while black".
In Massachusetts, USA, red light cameras were illegal until very recently, due to a 70s era law specifying that a live policeman had to issue a citation for something like that. From well before traffic cameras were common.
We had a pilot program in NJ for them, they were universally hated. People would slam brakes on and be hanging over the edge into intersection and throw their car into reverse panicking to avoid the ticket, ended up causing a ton of new accidents so the program was never continued. In newark people shot at the cameras: https://www.nj.com/news/2012/08/shoot_out_the_red_lights_2_t...
In most the USA, or at least Arizona, you have to serve someone. Just dropping something in a mail box doesn't mean dick. The very people that invented the traffic cameras up in Scottsdale were caught dodging the process servers from triggers from their own camera.
Another words, you have to spend hundreds of dollars chasing someone down, by the time you add that on to how easy it is to jam up the ticket in court by demanding an actual human being accuse you, it's not the easy win some may think. You're basically looking at $500+ to try and prosecute someone for a $300 ticket.
NY is not Arizona. They have the plate and send the fine to whomever the vehicle is registered to. If the fine isn't paid they flag the plate and impound the car if it's driven in their state.
In FL, a speed camera can give a car's owner can a ticket without needing to know he was the driver. Your perspective is not true nation wide.
"The registered owner of the motor vehicle involved in the violation is responsible and liable for paying the uniform traffic citation issued for a violation"
In CO we have automatic traffic cameras, and to my knowledge they just mail you the ticket, which is usually only a fine (and no license points). Its one of those “automatic plea” tickets where if you fight it, you fight (and risk conviction on) the actual offense, while if you just pay the ticket it will automatically get downgraded to a less serious offense (IE parking outside the lines).
Not in New Jersey. I visited my parents and didn’t stop for a full three seconds before making a right on red on a deserted road at night and they fined my dad.
They were rolled out but the mailed tickets are legally meaningless, someone has to actually hunt you down within a short timespan (I think 90 days) to create any binding requirement to address it.
A mailed citation from a photo radar camera is not an official ticket and does not need to be responded to unless it has been formally served to you.
Here in my country they removed the cameras in the second largest city after a trial period. It took too much effort to filter out police colleagues running a red (in police or civilian vehicles).
Sweden: Their locations are public. There is even an official API.
They are mostly located in sane places.
Apps like Waze consume this API and warn drivers if they’re at risk of getting caught. It’s the deterrence/slowdown at known risky spots they’re after, not the fine, I guess.
I heard that apps warning drivers this way are illegal in Germany?
Aside: what's up with the traffic speed cameras in Sweden? It feels like they're not designed to catch anybody. In my recent drive there it seemed like most of the cameras were in an 80 zone just before it switch to 50 for a tiny town. They wouldn't catch a typical driver who does something like 10 over everywhere -- they would likely have already started slowing down for the 50.
In my city in Canada, that camera would be in the 50 zone.
Nobody thinks it's racist to enforce traffic laws. People think it's racist to selectively enforce traffic laws by race, which usually takes the form of police pulling over Black drivers at higher rates. (But it can also mean installing more traffic cameras in minority neighborhoods!)
Try driving anywhere in the world that's not Western Europe or The USA and you'll quickly see how advanced even our worst cities are when it comes to traffic.
Last time I was in China drivers simply go through four way intersections at top speed from all directions simultaneously. If you are a pedestrian I hope you're good at frogger because there is a 0% chance anyone will stop for you. I really wonder how self driving cars work because they must program some kind of insane software that ignores all laws or it wouldn't even be remotely workable.
When I was living in China I got used to crossing large streets one lane at a time. Pedestrians stand on the lane markers with cars whizzing by on either side while they wait for a gap big enough to cross the next lane. It's not great for safety, to put it mildly, but the drivers expect it and it's the only way to get across the road in some places. I was freaked out by it but eventually it became habit.
Then I came back to the US and forgot to switch back to US-style street crossing behavior at first. No physical harm done, but I was very embarrassed when people slammed on their brakes at the sight of me in the middle of the road.
I've never been to SK, but in Japan things are -- unsurprisingly, as one might guess -- very orderly. For the most part (in cities at least) you don't jaywalk, even when there are no cars on the road.
If they run a red light today there is some small chance they will injure/kill someone.
If they run a red light with a camera, there is a 100% chance they will receive a ticket.
The key factor is not the magnitude of the penalty (i.e. whether someone dies or they receive a fine) but the chance that they will encounter the penalty.
That sounds reasonable to me. Everybody makes mistakes, but nobody should be consistently making grievous mistakes capable of causing serious injury or death to other motorists on a regular basis.
I'm less concerned with a little speeding than I am with blowing through lights and stop signs.
> Who knows, maybe we’ll start taking our cues from our polite new robot driver friends…
I think this could be an interesting unintended consequence of the proliferation of Waymos: if everyone gets used to drivers that obey the law to letter, it could slipstream into being a norm by sheer numbers.
Ultimately I wouldn’t support this level of snitching (especially in our current political env) but I’ve had the idea of:
A bounty program to submit dash cam video of egregious driving crimes. It gets reviewed, maybe even by AI initially and then gets escalated to formal ticket if legit. Once ticket is paid, the snitch gets a percentage.
Again, I am fundamentally against something like this though, especially now.
It is in NYC also, except it's entirely unenforced. We need a lot more red light cameras.
The nominal regulations on automotive behavior is pretty sufficient throughout the US, the main problem is that in most parts of the country traffic law may as well be a dead letter.
> It is in NYC also, except it's entirely unenforced.
It's enforced in the worst congested zones, the intersections around tunnel entrances and midtown, but as I said in my other comment usually by parking enforcement not NYPD.
A workaround in the law is to throw your turn signal on if stranded in the box, this doesn't count as blocking the box.
It is in NYC as well and it's usually enforced by parking enforcement (doesn't carry points but it has a steep fine), if NYPD writes it also comes with points but in my experience they'd rather let the walking ticket printers do it.
Great! and if enforcement were consistent, rule-breaking behavior would probably decline:
> Quick, clear and consistent also works in controlling crime. It’s not a coincidence that the same approach works for parenting and crime control because the problems are largely the same. Moreover, in both domains quick, clear and consistent punishment need not be severe.
If you look into the fleet size serving Waymo service areas, it's remarkably small. But because they work 24/7 they serve up a lot of rides, punching way above their weight in terms of market share in ride hailing.
Their effect on traffic and how drivers behave will be similarly amplified. It could turn out to be disastrous for Waymo. But I suspect that low speed limits in New York will work to Waymo's favor.]
I've seen reports that they've been testing Driver 6 in snowy places like around Lake Tahoe and the upper Midwest last winter. I suppose this year we'll find out how well that went.
In many places outside the USA they just use cameras for box blocking, stop sign rolling, speeding...and there is a system for honking also. But many in the states think automation here is too Orwellian.
They do that in NY too. The worst offenders inevitably have fake/defaced/covered/no license plates. That should be cracked down on very hard but the police and prosecutors are strangely reluctant.
A sound solution in general, but the majority of police and firefighters and government employees with a connection to law enforcement cover their license plates with magnetic 'leaves' and so on. It's an undocumented perk for government employees.
Isn't that what speed cameras are about? Seem a lot more efficient and cheaper. I got a few tickets, nothing too serious just ran the yellow a little too close and 40 in 25. And if def changed my behavior
NYPD cops don't like enforcing traffic violations: https://i.redd.it/w6es37v1sqpc1.png (License holders and drivers on the road are up in the same period that summonses are down, too. Traffic is up since pre-covid.)
Now that I live in Toronto we face the same challenges. Politicians may introduce traffic laws to curb dangers and nuisances from drivers, but police refuse to enforce them. As they don't live in the city, cops seem to prefer to side with drivers over local pedestrians, residents or cyclists who they view antagonistically. Broken window works for them because they enjoy harassing pedestrians and residents of the communities they commute into.
So there is a bigger problem to solve than legislation.
Police quiet quitting and arbitrarily choosing what laws they feel like enforcing is a huge problem.
The most effective fix vis a vis traffic is simply automating so much of it with speed averaging cameras and intersection cameras and taking police out of the equation and retasking them to more important things that only they can do.
Don't police have quotas any more? 40 years ago everybody knew not to speed at the end of the month because a cop that would normally give you a warning for a small speed infraction would give you a ticket instead so they could make this month's quota.
Part of the problem is we have police doing far too many jobs. We need to separate out traffic enforcement, mental health responses, and other works into their own focused units. Especially the mental health responses, as far too often police refuse to or (at best) don't know how to de-escalate in those situations.
The current Democratic nominee and frontrunner for NYC mayor plans to do exactly that! He plans to create a Department of Community Safety to take over mental health responses from NYPD.
Bringing a gun and a taser to every problem guarantees that a lot of problems will be "solved" with the wrong tools. It's impossible to train enough people to carry guns and tasers and use them wisely.
I agree we need to separate these responsibilities, but when it comes to mental health response, the police themselves are often opposed to alternatives, even while they complain that they're not mental health providers and often can't do anything in those types of situations.
In my city, we've had an underfunded street response program for a few years now, but a lot of people (including a lot of people who don't live here) see it as antagonistic to police and police funding, when really it should just be part of a holistic system to address social issues.
It makes no sense to me that the people who ostensibly care the most about addressing crime and "disorder" on the streets are often the most oppositional to programs that might actually address some of the underlying issues (not all of course, but some).
My wife and I took a road trip that included time in SF last year and seeing a Waymo was pretty neat.
To save some money, we stayed in downtown Oakland and took the BART into San Francisco. After getting ice cream at the Ghirardelli Chocolate shop, we were headed to Pier 39. My wife has a bad ankle and can't walk very far before needing a break to sit, and we could have taken another bus, we decided to take a Waymo for the novelty of it. It felt like being in the future.
I own a Tesla and have had trials of FSD, but being in a car that was ACTUALLY autonomous and didn't merely pretend to be was amazing. For that short ride of 7 city blocks, it was like being in a sci-fi film.
The company selling the car is adamant that none of their cars are fully autonomous in every single legal or regularity context. Any accident caused by the car is 100% the fault of the driver. But the company markets their cars as fully autonomous. That's pretty much the definition of pretending to be autonomous.
Are you trying to draw a distinction between sleeping versus looking away from the road and not paying attention to it? I expect both situations to have similar results with similar levels of danger in a Tesla, and the latter is the bare minimum for autonomous/unattended.
They key difference is that the Teslas killed their passengers, the Waymo hit someone outside the car (and it wasn't the Waymo's fault, it was hit by another car).
Yes. [1] That incident got considerable publicity in the San Francisco media. But not because of the Waymo.[2][3]
Someone was driving a Tesla on I-280 into SF. They'd previously been involved in a hit-and-run accident on the freeway. They exited I-280 at the 6th St. off ramp, which is
a long straightaway. They entered surface streets at 98 MPH in a 25 MPH zone, ran through a red light, and reached the next intersection, where traffic was stopped at a red light. The Tesla Model Y plowed into a lane of stopped cars, killing one person and one dog, injuring seven others, and demolishing at least six vehicles. One of the vehicles waiting was a Waymo, which had no one on board at the time.
The driver of the Tesla claims their brakes failed. "Police on Monday booked Zheng on one count of felony vehicular manslaughter, reckless driving causing injury, felony vandalism and speeding."[2]
FSD is not Autopilot despite the names being equated today, and obviously Waymo operates at a far smaller scale. It's simply not enough to compare raw numbers of fatal incidents to draw this conclusion. That's not to justify taking your eyes off the road when enabling FSD on a Tesla, but the OP did not suggest that either anyway.
No, rather, if the manufacturer of the self-driving software doesn't take full legal liability for actions taken by the car, then it's not autonomous. This is the once and final criterion for a self-driving vehicle.
Right now, Tesla skirts legal liability by saying that the driver needs to watch the road and be ready to take control, and then uses measures like detecting your hand on the wheel and tracking your gaze to make sure you're watching the road. If a car driving on FSD crashes, Tesla will say it's the driver's fault for not monitoring the drive.
I had my second Waymo ride in SF 2 weeks ago and I had to press the support button: it was behind a large bus that was backing up to parallel park. The bus was waiting for the Waymo to get out of the way while the Waymo was waiting for the bus to move forward.
It took only a few seconds for a human to answer the support request and she immediately ordered the Waymo to go to a different lane. Very happy with the responsiveness of support, but there's clearly still some situations that Waymo can't deal with.
I was in a Waymo in SF last weekend riding from the Richmond district to SOMA, and the car actually surprised me by accelerating through two yellow lights. It was exactly what I would have done. So it seems the cars are able to dial up the assertiveness when appropriate.
It doesn't seem impossible technically to up the assertiveness. The issue is the tradeoffs: you up the assertiveness, and increase the number of accidents by X%. Inevitably, that will contribute to some fatal crash. Does the decision maker want to be the one trying to justify to the jury knowingly causing an expected one more fatal incident in order to improve average fleet time to destination by 25%?
Reinforcement learning is a helluva drug. I'm sure by now Waymos can time yellows in SF to within a nanosecond, whereas humans will only ever drive through so many yellows will never get that much training data.
A human can know the yellows on a few routes. A Waymo can pull over, observe a given intersection for an hour, and tell every other Waymo that exists precisely how long that light lasts.
It's not just collecting the information; it's the ability to spread it.
I'm sure "timing of yellow" is only a few parameters in its network at this point. And it's continuously training, it can probably one-shot the timing changes ( one taxi ride through maybe 3 lights ).
When red-light cameras are installed at an intersection, the number of rear-end accidents typically increases as drivers unexpectedly slow down instead of speeding up at yellow lights.
The cost of these accidents is borne by just about everyone, except the authority profitably operating the red lights. (To be fair, some statistics also show a decrease in right-angle collisions, which is kinda the point of the red-light rules to begin with.)
That seems only like a temporary problem until people get used to actually stopping at red lights, as they are supposed to. After the initial acceptance phase, it should minimise accidents over the longer term.
Unless there is a warning of how long is left on the yellow light, it’s an unsolvable problem because there is an asymmetric risk of stopping vs accelerating
The lights should be designed so that if you don't have enough space to stop with a mild deceleration you should just go through. If a mild deceleration get you rear ended then of course that's an unsolvable problem
My memory may be outdated or only local to my jurisdiction but my understanding is that yellow means “do not enter the intersection” where “intersection” begins before the box, usually with some alternate street indicator, like broken white lines turning to solid, at a braking distance that accounts for posted speed limit and yellow light duration.
Each Waymo is equipped with multiple cameras (potentially LPR), LIDAR, etc. The car knows when the vehicles around it are breaking traffic laws and can provide photographic/video evidence of it. Imagine if Waymo cars started reporting violators to the police, and if the police started accepting those reports. Someday they might.
Yes it could potentially be very dystopian for human drivers. That doesn't mean it won't happen. Police departments could make a lot of extra money from the additional traffic tickets; there is a financial incentive for them to do this.
>> could you imagine the audacity of actually not driving into an intersection when the light is yellow and you know you're going to block the crossing traffic?
I wonder how many Waymos following the rules would be needed to reduce gridlock.
Waymo in SF pretty much drives like a human, and that includes doing human things like cutting lanes, stopping wherever it feels like, driving in the bus lane etc. I think it’ll be fine in NYC
Waymo in LA also drives pretty much like a human here would, which includes: not yielding for pedestrian-only crosswalks, running red lights, driving in the oncoming traffic/suicide/bike lane, occupying two lanes, blocking entrances/driveways/intersections, and stopping/parking in no-stop/parking curbs.
They're only really phenomenal at not hitting things; they really aren't good/courteous/predictable drivers under most conventional definitions.
Still, I think rollout in NYC will be fine. NYC generally drives slower and much less aggressively than LA, and slower gives the Waymo plenty of reaction time to not hit things.
I can attest to that. I live in Orange county and occasionally see Waymos when I go to LA, and they'll do things like merging with very little gap or merging in the middle of interactions.
Traffic? The issue half the time in NYC is the drivers.
I can't compare it to SF since I haven't been there in a while but I still thought it was not as congested to compared to NYC.
NYC has a greater population and also has a greater number of registered cars compare to SF however.
I was on Market Street yesterday on my bike next to a Waymo. A bunch of cars were blocking the intersection when we had the green. The light turned red and the cars blocking the intersection were able to move. I decided to stay, but the Waymo sped through despite the light being red. I regretted not crossing.
Driving in most of the city isn't that bad. Even most of Manhattan is fairly regular driving compared to most of the country. It really isn't until you're near midtown that the insanity kicks up.
It's fascinating seeing all the comments elsewhere anytime Waymo starts testing in another city along the lines of, "ah, but how will they handle X, Y, and Z here?? Checkmate, robots!" despite having already launched service in several other cities.
Granted, NYC is the biggest city in the US, so maybe that sort of reaction is more reasonable there than when people in Dallas or Boston do it.
NYC is a new set of challenges. As you already mentioned snow and ice is new. But also missing the high density of people and cars per square area. Behavior of drivers and pedestrians are different and much less polite. I can see it working in NYC but "just fine" is a bit of an over confidence... at least not for the first few years before they learn to deal with these issues that they don't face yet in LA and SF.
People in LA wait to jaywalk on the street or even in the suicide lane all the time. The Waymos handle it fine; generally by asserting it has right of way unless collision is obviously imminent. They'll even happily swerve around you if you're too far out.
We do have narrow streets in LA with double parked cars, cars parked in the street only allowing one car through the middle at a time, and plenty of construction closures and obstacles.
Why do so many NYC people think there’s comically no cars in LA or neighborhood streets?
Also, I can assure you LA drivers are a tad bit more aggressive than NYC drivers (less honking and flicking off though, LA people are more a drive you off the road or into the shoulder sort of passive aggressive).
I was born and raised in NYC and have lived in LA for quite some time, still going home often for family. I’m really struggling with reading these “NYC is unique” comments regarding Waymo traffic.
What snow and ice? We don't get much of that anymore.
That was actually the last thing I am worried about here. I really want to see how Waymo does with NYC drivers and obstacles(double parking on block where sometimes you have to pull in your mirrors just to get by(if you even take the chance instead of just laying on your horn). In some neighborhoods it can be so annoying.
I think a well designed winter specific FSD system is probably more safe in snow and ice than a human. For instance downshifting to ensure wheels continue to spin on slippery surfaces, subtle corrective steering to keep the vehicle within its lane, etc. should be easier for a FSD car since it won't panic and over-correct like most people do in those situations.
And if the car reduces speed when appropriate and some assholes start tailgating it, it won't suffer the anxiety of holding up 10 cars that want to drive beyond the safe, reasonable speed for the snowy/icy conditions.
Pretty much all electric cars have single speed transmissions, so there's no downshifting. And modern vehicles have electronic stability control, anti-lock brakes, automatic emergency braking, and several other safety systems. It's pretty hard to overreact with those enabled. The main issue is that people exceed safe speeds for the conditions, making them unable to brake or turn in time to avoid a collision.
Right now, most self-driving software will refuse to activate in conditions of poor visibility. I've had that happen with Tesla's FSD, though in that case it was snowing so much that the road should have been closed. Also when the snow is deep enough that your front bumper becomes a plow, it will refuse to activate.
My point wasn’t to say they are the same, more that SF and LA (I would guess) have covered and defeated almost every single challenge and obstacle for an urban environment (sans..weather).
LA also has far denser areas than SF, places like DTLA and Koreatown are more dense than most boroughs in NYC (sans Manhattan).
My point was that it will be interesting to see how well it works in NYC where the only way to drive is aggressively in city streets breaking the rules. LA has its share of aggressive driving but as someone who has driven in both, NYC felt like I had to break the rules to go anywhere, LA not so much.
Hard to really compare a tiny piece of LA and say it’s more dense and compare it to borough that is in the same range but also magnitudes larger total pop.
> I genuinely think the most complicated challenge for Waymo in NYC will be…winter snow and ice.
Nah, I'm betting it'll be the locals. They'll get pissed off at it remaining stopped when it shouldn't and do shit like start ramming into it. I've had it happen on the island when I stopped at a yellow. NYC is a lot more chaotic than any other US city I've driven.
Oof I don’t know about that. Driving in NYC is much different than San Francisco. Frequent lane departures, cutting into heavy traffic despite technically lacking right of way, and other moderate rule breaking is required to get anywhere. Boston will be even more challenging due to the hundreds of convoluted intersections.
We've been getting less and less of those, though. And even then, it's just for a few days. Last year was a bit worse, but two years ago it was very, very mild, I think. Yay global warming?
The thing is waymo at least in LA specifically geolocks you from those hilly areas. Imo it also is not assertive enough and drivers seem to be learning one can bully a waymo on the road.
Traffic on the sidewalk is a daily occurrence and often necessity in NYC. I'm not sure exactly how every area in LA is but often (as in pretty much constantly every day) in Manhattan or Brooklyn drivers don't obey the lines on the road, don't care to bump objects and cars to fit into a spot, literally threaten to hit other cars to get anywhere.
There's a bit of a "do what you have to" mentality with NY traffic that I haven't seen in any other east coast or mid-western city. I think that poses some unique challenges that I've often seen video of Waymos freezing up when facing similar scenarios, which could cause huge issues in most of the city.
LA is extremely similar. Often can only make unprotected turns at lights while it’s red and you’re in the box, you have to wait at the top of a hill and have your car sideways while the oncoming car has space to drive up a hill, cars trying to give you space so you can drive through a line of traffic into the adjacent traffic pattern.
The “freezing” issues are very real though (and frustrating), and it’s what most everyone who uses Waymo in any city right now jokes/complains about. Waymo can often get into a weird game of “chicken” when there’s a four way stop with pedestrians, and any slight movement from the intersection can often make the car stop - so the pedestrian stops - the the Waymo finally moves again, but then pedestrian also started moving so the Waymo stops again and the pedestrian stops caring.
All this to say, I really don’t think there’s much that will be different. Go to Hollywood or Santa Monica
there is nowhere in LA with the complex intermingling of pedestrian, car, bicycle, and motorbike traffic of anywhere in the boroughs other than Staten.
LA it’s gridlock or go. There’s nothing complicated about it other than strategizing where is gridlock and where is Go.
NYC is also one of the most regularly built out cities, in stark contrast to Boston, for instance. OTOH roads here may be 3 or even 4 levels high at the same point (e.g. where Manhattan bridge meets Brooklyn), and GPS is sometimes way off in canyons between skyscrapers.
> OTOH roads here may be 3 or even 4 levels high at the same point
And here I thought Chicago was complex with lower lower Wacker (just 3 levels).
> GPS is sometimes way off in canyons between skyscrapers
This is probably very challenging for human drivers using navigation, but probably no nearly as much of a problem for a Waymo car with onboard 3D maps of the entire operating area.
I think the main difference for NYC is that quite a few streets and intersections routinely have 10x to 100x the pedestrian traffic of the busiest such intersections in pretty much any other American city.
That's not to say that I don't think it'll be able to handle it, just that it'll be a new challenge. I wonder if their current program of apparently trying to positively track every single moving object in range will survive that, or whether they'll need to figure out some algorithm to prioritize objects that are more likely to be of concern to it. And there probably are more than a few places where pedestrians are numerous and densely crowded enough that you can't positively track all of them, even with a bunch of LIDAR sensors.
NYC (at least the parts I've spent a bit of time in) is pretty grid like with fairly simple roads. The drivers are the hard part :)
I am excited to see them tackle Boston at some point because of how strange some of those roads are. The first time I had ever been I came to an intersection that was all one ways and there were like 7 entry/exit points. My GPS said turn left, but there were three paths I'd consider left. Thank god I was walking.
And I don't really pose much doubt because it seems like Waymo's rollout plan has been solid, but I'm just interested to see how well they tackle different cities.
Yup, the data so far seems to indicate that Waymo is substantially safe than average drivers. Obviously it's not inclusive yet since the tech is still new (and while the study I'm thinking of was done by a third party, it's still Waymo that handed over the data and paid them to analyze it).
There is a question, NYC driving gets by with everyone driving aggressive and breaking road rules. That is something that does not exist as much in other markets.
My complaint with Tesla city FSD is that it’s not quick or aggressive enough. It will come to long and complete stops and other things that will not work well in NYC.
I drive when I travel, though not in NYC these days but being a pedestrian gives you a good enough view. LA from a city driving perspective at least for me is not comparable to most of NYC.
Roads in Texas specifically just seem to do whatever they want, whenever they want. It's really apparent that Texas local roads used to be wagon trails.
The grid system in NYC seems like a good alternative for a rollout. Though the current NYC human drivers will hate these things. I also expect LOTS of vandalism.
What I don't get about the "checkmate robots" mentality is that, like, get it working in sunny California with no snow, and then get it working in the snow, seems like the way to do it, not, solve all possible problems before anyone knows you're even trying and can make fun of you.
>despite having already launched service in several other cities.
Why does having launched in other cities matter if the new city brings up things that none of the other launched cities do?
For example the first thing I can think of new for New York is snow and ice.
It's my understanding that self-driving cars don't really account their acceleration and braking for roads that could sometimes be very slippery due to snow and ice.
> "Why does having launched in other cities matter if the new city brings up things that none of the other launched cities do?"
New requirements come up all the time in technology. The existence of a new requirement isn't in and of itself justification for skepticism - is there a particular reason to believe that Waymo is not capable of solving for the new requirement?
The answer may be yes, but simply "ahah! It would need to do [new thing]!" is insufficient. "[new thing] is likely intractable because [reason]" would be more justification for skepticism.
> "It's my understanding that self-driving cars don't really account their acceleration and braking for roads that could sometimes be very slippery due to snow and ice."
Sure, but like above - is there a reason this is an intractable problem?
I'll throw this out there: your human-driven car already accounts for acceleration and braking on slippery roads, without the need for the human. Traction control systems and electronic stability control systems exist! They're in fact incredibly common on modern cars.
>I'll throw this out there: your human-driven car already accounts for acceleration and braking on slippery roads, without the need for the human. Traction control systems and electronic stability control systems exist! They're in fact incredibly common on modern cars.
These systems don't help with the problems I am talking about.
You have to drive completely differently in heavy snow, significantly slower, brake sooner, turn less sharp, accelerate much slower, leave significantly larger gaps, leave space to move out of the way and be ready to move if someone behind you is coming at you too fast and can't stop in time, etc. I've spend my entire life in the midwest.
The traction control system in my 2023 camry didn't help one bit when I applied the brakes on black ice and the car didn't react at all, it just kept sliding at the same speed across the ice.
That all sounds like something that should be easier for a robot to do than the typical human. If programmed for how to drive in heavy snow, a robot should be able to switch driving modes much easier than the typical human brain.
Waymo has been trained in Buffalo NY for winter conditions, unlike most NYC drivers.
Is it possible to train a machine to drive in snow? Yes. But consider that humans are trained to do so by means of things like: actually crashing, observing others crashing, talking to people who crashed, and all of the above is highly localized. Where I live there are many days in winter when someone not from the immediate area should not drive at all. But I might if there was a good reason because I have 25 years experience with the specific roads, conditions, how those conditions relate to wind and on and on. Training a machine to know all that seems feasible but unlikely to be commercially viable. It's just not a problem that can be solved with a simple closed loop control system like ABS or traction control.
A reasonable counterargument is that autonomous vehicles can actually do that to a degree that is much, much more effective than humans. You might have 25 years of experience, but at 8 hours a day for 365 days of those 25 years we'd only need 8 cars driving for a year to match that. After all, training data and event logs generated by cars can be shared, and models can be upgraded all around. And of course that scales to more than 8 vehicles rather easily.
The interesting snow & ice problem for me is that humans will drive in winter conditions that are unsafe -- for example white-out blizzards. Robocars won't be able to drive in a white-out blizzard, so they'll likely refuse to do so. Humans should also refuse to drive, but people drive anyways.
NYC doesn't generally get white-out blizzards, so refusing to drive in them is quite feasible.
I come from way up top on that globe of ours. I have driven in frankly apocalyptic snowstorms. They're an insidious problem to solve, but I remain optimistic. Back home, they will close specific roads due to snowstorms, but what do you do about the cars already on the road? You can't stay put for 16 hours can you? So you move as slow as possible, sometimes as low as 5 kilometres an hour. Cause that's the thing about a snowstorm; it's about visibility. You're not risking your life if a dude in skis can go faster than you.
> You're not risking your life if a dude in skis can go faster than you.
Sure you are. You can still drive off the road and into the ditch where nobody can see you. People then die because they don't clear their tail pipe and get carbon monoxide poisoning or they try and walk for help and freeze to death.
If white-out visibility is the only problem to be addressed then machines seem pretty well placed because they can use very accurate positioning and non-visible light sensors. Unfortunately they probably wouldn't know that there's a 50 yard section of the road that always drifts in when the wind comes from the south and the snow is dry.
Launching in other cities with new problems gives experience dealing with new problems, and the meta-learnings transfer to better processes for adapting to new issues. But yeah, ice and snow are definitely major new environmental factors for New York (and DC, and many other places we are starting to see more serious testing).
Autonomous vehicles can and do take into account surface conditions, there’s not really any reason not to. There are pretty good generative models of the physics of vehicles with different surface conditions, and I imagine part of the data collection they are doing is to help build statistical of vehicle performance based on sensed conditions.
A fair point about weather, but a lot of the assertions are like "how will they handle the double parking and suicidal pedestrians jaywalking across the street??" I'd say most of the concerns just don't sound very unique at all.
For weather, Waymo has clearly started out in warmer climates while slowly building out towards places with colder and colder weather, I'm guessing they're just incrementally getting better at it.
This is great long term for having cars that follow traffic laws since human drivers in NYC are awful (kill/injure pedestrians, bikers, and other street users all the time).
Not so great for getting cars out of NYC and pedestrianizing more of the city/moving towards more “low traffic neighborhoods” as I imagine Waymo and other similar companies are going to fight against these efforts.
Edit: Lots of people talking about human drivers taking advantage of self-driving cars being more cautious/timid. Good news is that once you have enough self-driving cars on the road, it probably slows down/calms other traffic (see related research on speed governors).
I'm not sure why you think waymo would fight against that. People getting rid of their own cars for daily use will increase how often a service like waymo is used for occasional usage. In the long run it would be a win for waymo. Not many people are taking taxis on a daily basis in New York for normal driving, they buy a car if they need to do that because even with the parking bill they will still come out ahead. And once they have their own car they feel like they need to get some use out of it.
> not sure why you think waymo would fight against that
If you were to pedestrianize 10% of Manhattan (or for example all of Broadway, which is being considered), then that’s less area for Waymo to operate and make money. To be clear, this is likely more of a long term issue.
Waymo would not fight against "people getting rid of their cars", many people in NY who use the incredible public transit system would like to see more car-free streets, which they absolutely would fight against.
Do you mean that in the sense of "anyone getting killed is unacceptable" or the sense of "we need complete separation between cars and pedestrians/bikers, somehow"?
I mean if we required a license to own a bike in NYC we could see a significant reduction in injuries/deaths, same for pedestrians. Cars are already heavily regulated and likely aren't the underlying issue.
There are many ways to interpret data, but one often comes to the conclusion that pedestrians and bikers are the root cause of most accidents.
It’s a step in the right direction, but they still pollute (heavy electric vehicles have a lot more tire dust) and take up a lot space (could close roads and build housing or just have more space for the millions of city residents that don’t have/use cars).
> heavy electric vehicles have a lot more tire dust
It would be interesting to know the fleet-level statistics for this. Driven by humans, EV might wear tires faster because of fast starts and the extra weight during stops. It's possible that the Waymo Driver accelerates and decelerates more smoothly, resulting in less tire wear than a human-driven ICE vehicle.
You know what kills or harms people in NYC are the motorized bikes driving the wrong way and putting people in the hospital, with no charges against the operator because they are usually an illegal alien. Not sure Waymo is going to fix that.
Waymo should add a thin layer of "assertiveness" for actual deadlock that their self-driving architecture could cause.
While in Austin, I was in a Waymo that blocked 3 lanes of incoming traffic while attempting to merge into a lane going into the opposite direction. It was a super unorthodox move, but none of the drivers (even while stopped for a red light) would let the Waymo* merge into their lane.
Thank God for the tinted windows, people were pulling their phones out to record (rightly so). It felt like I was responsible for holding up a major portion of Austin 5 pm traffic on a Friday.
Wish it just asserted itself ever-so-slightly to get itself out.
I think we're going to see more examples of this as Waymo's popularity grows. Basically human drivers taking advantage of Waymo's far more passive driving style. Maybe some rules of the road will have to change, or the Waymos will get dedicated lanes to solve this problem.
You joke, but the reality is going to be dynamic self-driving buses that don't have preset routes or stops but respond to instant demand.
You'll pay $$$ for a nonstop ride into midtown in a dedicated vehicle, or $ for a short dedicated ride to a self-driving bus you only need to wait a few minutes for, and which will drop you off on your destination block.
So yes -- self-driving buses seamlessly integrated into ride sharing are certainly going to be a major part of 21st century urban transportation. Which will save a ton of time compared to current buses.
You need to think bigger. Once we have separate lanes just for the waymos, we don't need them to be regular roadways. We can scale up the waymo even more and size the lane exactly to the vehicle, maybe even radically redesign the road surface for lower rolling resistance. What a future it will be.
By forming those waymos like aerodynamic bullets, they could reach ridiculously high speeds on those special lanes. Something like 200 mph should be possible.
Maybe the waymos could be powered by overhead wires?
I'm having a hard time even picturing such a thing, but I have no doubt that Waymo could manage to operate them in cities around the nation, with sufficient re-training.
I would like that just as much as the next guy but the problem of public transport cannot be addressed until you first address the problem of anti social behavior on public transport.
That’s just being around people. We gotta live together as people; the idea that we can atomize ourselves away from the society we live in is more disastrous to the shared social fabric than any amount of people listening to music without headphones.
I find that in LA people routinely cut off the waymo or refuse to let it in. After all why not it is a robot, not someone who might legitimately harm you like a road rager. It also tends to fail the cultural left turns. That is, sending 2-3 cars left during yellow not just one like in other places. Seeing it stuck awkwardly in an intersection for another cycle from failing to make an assertive left turn is somewhat common.
Waymo also avoids certain challenging environments by excluding it from its service coverage, namely hilly neighborhoods.
well, just couple weeks ago here on the intersection of Middlefield and Shoreline, half-mile from Google headquarters - 100 million times driven by Waymo cars, thousands by us - midday, perfect visibility, perfect intersection with all the markings, lights, etc., we and a Waymo are doing left turns from dedicated lanes on the opposite directions. We were saved from head on collision by the "lack of assertiveness" on the part of my wife as she swerved last moment as the Waymo apparently decided that its left turn point lies way into, very deep into, our trajectory, and it was assertive enough to not care that we were in its path. I almost soiled my pants upon seeing how it went for barreling into us instead of turning.
It looks like the same extra assertiveness like with Uber back then - i.e. to not have an emergency braking and similar features because it gets too much false positives.
Yeah I admit that "assertiveness" isn't the right word here. I've been in Waymo's that have also tried to dangerous moves in front off busses. Maybe "conscientiousness" would be a better definition?
This is awful. Your ride takes just a bit longer, so you want it to take more risks in decisions?
This is how you ruin trust. Take the L dude, sit back and relax. You will get to your restaurant or whatever inane activity you are doing for the day/evening.
Biggest thing I'm excited for is knowing what the cost will be ahead of time
Which Uber used to provide... Until they were infected with tipping. Hell, I will gladly pay more than I would've spent on a tip (20%) just to avoid the hassle.
I’m curious if autonomous cars will become targets for aggressive drivers. Like a driver isn’t going to be as scared cutting off a Waymo or tailgating one because the AI isn’t gonna get road rage or honk like hell. In some places I could see the Waymo’s getting severely bullied if that’s the case.
People tailgate because they're toddlers and locate their locus of control externally - if anything, they'll be very happy tailgating driverless cars because they can throw as big a fit as they want, there will be no consequences, and they'll feel they got to blame something else other than themselves.
Because there’s still someone in the car, they just have no way to defend themselves. You can tailgate and honk at them to your heart’s content. Well at least until they call the police but that’s pretty far. And there are other forms of aggression that do accomplish something. If you cut a Waymo off or beat it to merge, you get ahead of it. In some locales I could see a whole series of cars merging ahead of a Waymo if people are aggressive enough.
You are on to something here. I have started driving a small car in the UK (Ford Fiesta) and have discovered it's a magnet for the road rage people (around 50% of drivers here).
Firstly, I never back down and will come to a complete stop if slowing down doesn't work.
Secondly, I have noticed these drivers feed off any reaction and that avoiding eye contact works very effectively, even if they pull beside you to have a childish rant.
They're already learning how to handle this in SF. (I don't live there anymore, so I can't give specific examples.)
Waymo markets itself as an automated driver - same reason they're using off-the-shelf cars and not the cartoony concepts they originally showed. Like real drivers, they take the law as guidelines more than rules.
De jure (what the law says) and de facto (what a cop enforces) rules have had a gap between them for decades. It's built into the system - police judgement is supposed to be an exhaust valve. As a civil libertarian, it's maddening in both directions:
- It's not just that we have a system where it's expected that everyone goes 15mph faster than posted, because it gives police an avenue to harass anyone simply for behaving as expected, and
- It's also dystopian to see police judgement be replaced with automated enforcement. There are whole classes of things that shouldn't be penalized that are technically illegal, and we've historically relied on police to be reasonable about what they enforce. Is it anybody's business if you're speeding where there's nobody to harm? Maybe encoding "judgment" into rules will be more fair in the long run, but it is also coaching new generations to expect there to be more rules and more enforcement. Feels like a ratchet where things that weren't meant to be penalized are becoming so over time, as more rules beget more automated, pedantic enforcement.
A slight digression, and clearly one I have a lot of thoughts on.
It's really interesting to see how automation is handling the other side of this - how you build machines to follow laws that aren't enforced at face value. They can't program them to behave like actual robots - going 24 mph, stopping exactly 12" before the stop line, waiting until there are no pedestrians anywhere before moving. Humans won't know how to interact with them (cause they're missing all the nonverbal communication that happens on the road), and those who understand their limits will take advantage of them in the ways you've stated.
So Waymo is programming a driver, trying to encode the behaviors and nonverbal communication that a human learns by participating in the road system. That means they have to program robots that go a bit over the speed limit, creep into the intersection before the turn is all the way clear, defend against being cut off, etc. In other words, they're building machines that follow the de facto rules of the road, which mean they may need to be ready to break the de jure laws like everyone else does.
TBF the Zeekr minivans are a big step toward a purpose-built Waymo vehicle. I do agree that Zoox has its priorities backwards by going straight for a purpose built robotaxi vehicle. But it has advantages like friendlier ergonomics for the disabled.
Who cares? You are focusing on unimportant issues.
Movement in the USA is heavily outdated. Whether it’s "automated" won’t change anything other than encourage more cars on the road. Great your 5AM commute from the boondocks still takes 2-3 hours but at least you don’t have to put your hands on the wheel!
I haven’t lived in NYC, but I have lived in Boston. Isn’t the real concern winter? Has Waymo (or any other self driving tech company) shown that it can handle the snow well: non-visible lanes, downshifting to avoid braking, etc.?
Having a seasonal service is not a bad idea. The big problem with that is cutoff times. Too early and people will complain when they can't get a ride when no snow is on the ground. Too late and you're liable for everything that happens when the road is covered in thin ice or sleet, including leaving someone stranded. You will need very accurate weather predictions for operating over the winter months.
I exclusively use Waymo in SF, even if it costs a bit more than Uber. You'll most often get a great human Uber driver, but there's a very real possibility that the person is a bad/unsafe driver or the ride is unpleasant for a myriad of other reasons. With a Waymo, you know exactly what you're buying.
Yea, that's what I figured, but I also wonder how well anyone is driving in the slush and if the LIDAR / cameras are that disrupted by snow / ice / salt.
The permit gets them into the process for eventually deploying without safety drivers. That includes safety plans, emergency responder plans and training, and periodic reporting.
They could just drive cars around like Tesla, but that wouldn't put them on a path to a fully autonomous service.
I don't thin it's fair to say they are fully automated. There's a large remote operations team for remote assistance to help them get out of tricky situations. The cars can be nudged to perform certain actions.
New York is a long way behind California in regulating autonomous vehicles. Fully driverless vehicles are also illegal there and it will require legislation for Waymo to deploy in the state.
It's not insane for cities to permit autonomous vehicle technology. They permit almost every other type of heavy machinery. Even manually driven cars are permitted! (Driver's license test, registration fees, etc.)
I didn't say testing was stupid. I said permitting only 8 vehicles for human testing from the leader in self-driving cars, years after they've been fully autonomous in other dense cities, isn't the flex he thinks it is.
Lots of comments sharing their waymo experience, so I'll hop on the bandwagon :) I visited Austin for a work trip and went out of my way to get waymo rides for work events, reimbursed of course ;) , managed to score 3 rides.
The airport is out the coverage map so I had a real person behind the wheel both ways. Objectively, the waymo was way safer experience because one driver was a local and drove like one (e.g. rolled through stop signs, drove past a long queue to merge at the end, etc.) and the other smelt like weed in the car. Luckily, both trips we arrived unharmed. In comparison, the Waymo drove pretty well, imo and very consistent. Nothing extra ordinary but no reason to stress.
The difficult part of riding the waymo was all moral cope: it cost just as much (minus tip) as paying a real person, driving past homeless people under a bridge in an autonomous vehicle felt unsettling, and my driver from the airport in my home city was wonderful and hard working. I don't typically like to chat in the cab, and the driver didn't initiate, but I was feeling empathetic and guilty so I struck up a conversation. By the time I got home we were enjoying ourselves and the driver was sharing animal facts because I had learned he was a real enthusiast that could not make a living solely on ecology. We were laughing and joking around together. (Google, if you're reading do NOT try to replicate this experience with AI)
I'm glad I got to try it and out of my system. Still would prefer trains or more public transit over more cars :p
Once again, if you're going to test your self driving car on public streets, all data should be open and public. Car companies shouldn't be competing on how to prevent life-or-death incidents, they should be cooperating.
Downtown Manhattan is the hardest-to-navigate area of NYC. I thought they would start somewhere in Midtown, where the grid is regular, streets wide, and demand for taxis still pretty high.
We already know that Waymo can handle regular American cities quite well. I woul expect them to spend most of their expensive human-supervised training and testing budget in the most unique locations, like downtown Manhattan.
Because half the year it's freezing cold or blistering hot?
With CitiBike and so many bike lanes it basically already is a biker's paradise. Obviously there could be lots more improvements, but the people who want to bike already do.
The game-theoretic aspect of this is interesting to me. A lawful robot will never make progress in Manhattan because the people will just walk across its path continuously, forever. To be an effective driver in Manhattan you have to intimate that you're willing to hit people, without ever hitting them. If humans believe that the Waymo will categorically never hit them, then the Waymo will never get a turn.
its interesting. at beginning in SF the waymos would just stop cold anytime they saw a person or a bicyclist. now they're acting a lot more like a person. if I'm in the crosswalk they've started playing chicken just like a normal driver would, starting to go into the turn while watching to see if you're going to stop and give them the right of way. if you keep going, they will stop.
There are still plenty of humans in SF who are on to the nature of the game. A few of the shambling lunatics who inhabit the vicinity of 6th and Jessie know that they can just harass a Waymo and it will stand there forever.
I think the problem in NYC will be getting medallions, assuming that's what self driving cars will need.
There are already so many (too many?) taxis and car sharing drivers, after TLC's massive increases of the last few years. You can play a game, based on something I read about last year: stand at a corner and count all cars/trucks/for-hire. The first two combined are barely outnumbered by the last group. And the few times I checked, half of taxis and car sharing vehicles were empty. (Of course that's different at peak times or when it rains.)
Will Waymo be allowed to add as many vehicles as they want, like a new class of cars, or will they need to buy out medallions from drivers? The former might undo all the progress in traffic relief that was brought by congestion pricing.
I think taxis have a place even in cities with great transportation. I live in Toronto and 90% of my commuting is walking, 8% public transit and 2% driving. But there are some trips that would be very difficult to do a way other than driving (for example carrying lots of stuff or awkward cargo) and taxis fill that gap wonderfully. Especially if self-driving taxis could handle long trips a lot better as inter-city is a place where Toronto public transit unfortunately sucks (for example visiting my parents in cottage country).
This is true for intersection throughput, but I bet full travel throughput (walking to the bus, waiting for the bus/train, walking to your destination) is the same or better with Waymo.
There's a great YC saying which approximately says "you should get the easiest customers first". They even made a video about it, saying tech startups sometimes try to go for the hardest customers to "prove themselves" and it just hurts their business.
I sort of wonder if that's happening here. SF, Austin, LA, etc, are all great cities to build autonomous vehicle startups in. There are many more major cities which don't get snow, have minimal rain, and are well thought out in terms of driving layout. NYC seems like the most difficult city to operate in, and while I believe it's a lucrative market it seems like a mistake.
I can't wait to hear how it goes in NYC -- its going to be a total cluster - with the significantly more chaotic behavior on the streets, bike scooters, pedestrians and then the oddness of the streets/aggressive driving necessary behavior.
Give it one month if they saturate it too much there will be political blowback on waymos causing traffic chaos. Queue track record in SF as datapoints.
I don’t think people commenting and downvoting us realize how things are in NYC. Not only do you have to deal with insane chaos you also have to deal with malicious drivers. Hit-and-run in NYC is shockingly high because it’s a no-fault state. People don’t stop after accidents. It’s gotten really bad since the pandemic.
I think some people pushing for driverless cars everywhere are assuming it will necessitate much stricter driving laws and penalties for human drivers to make their driving compatible with the robots. And they're fine with that, but they know it's not a selling point, so they don't talk about it.
I live in one of the areas they are actively testing/training in. Their cars consistently behave better and more safely than most human drivers that I’m forced to share the road with.
As semi-autonomous and autonomous cars become the norm, I would adore to see obtaining a drivers license ratchet up in difficulty in order to remove dangerous human drivers from the road.
Traffic enforcement, which used to correct some bad driving, has basically evaporated in many parts of the U.S. This has been a long-term trend.
A friend who's a cop told me that only when their department got specific state grants would they set up stings of drivers driving in a pedestrian walkway while someone was crossing the street. He said he would only pull someone over if they were doing something batshit crazy and they happened to be behind the vehicle where it was easy to pull them over. Minor stuff and speeding they would almost never ticket.
The U.S. and other countries need to use automated methods of detecting and applying penalties. Some busy intersections have cameras for this, but it seems to be very limited, maybe because of cost.
Years ago New York used to calculate if you were speeding the NY State Thruway based on the time between toll booths. They cancelled this program for some reason.
The New York State Police have speed cameras set up in a few highway work zones, which is effective (double fines applicable) but it still requires a person driving a car to set up the gear.
The real issue is all the current bad drivers. A requirement to start re-testing normal people in addition to the elderly would be a large benefit to society.
I'm from the UK, took driving lessons in the UK but then passed my driving test in the USA (in California).
The USA driving test is so much easier than the UK one!
UK: Varied junctions and roundabouts, traffic lights, independent driving (≈20 minutes via sat nav or signs), one reversing manoeuvre (parallel park, bay park, or pull up on the right and reverse), normal stops and move-offs (including from behind a parked car), hill start, emergency stop.
California: Cross three intersections, three right turns, three left turns, lane change, backing up, park in a bay, obey stop signs and traffic lights.
My understanding is that the USA test is so much easier because it's hard to get by in most of the USA without a car, so if the test was harder people would likely just drive without a license instead.
Not to mention no stick shift. The driving test from hell in hilly Adriatic cities: parallel park facing downhill
To be fair even people who have been driving many years do this by grinding up the clutch.
Complete unrelated, I just wish every driver on the road re-learn that cyclists have the same rights of being on city roads like cars.
And I wish cyclists would re-learn that pedestrians have more rights of being on sidewalks. That said, the bigger plague on sidewalks are e-scooters.
Additionally, most cyclists I see never stop at stop signs no matter how busy the intersection is.
I would support re-testing on some interval like every 5 years. That said, so much could be done to make the environment safer. Lower speeds, more traffic calming, safer intersections, safer alternatives (public transit, walking, bicycle).
I can't help but think about the failures of basic human-oriented infrastructure when I can't safely ride my bike to the grocery store 2 miles from my home. I don't know what it'll take to change this in our cities, and it feels like an uphill battle when seemingly very few people care about problems like these.
Everyone agrees to this, the problem is there needs to be a way for this to be done efficiently so it's not another regressive tax on poor people's time and money.
Yeah the mindset is essentially drive to spec in the test and then skirting the law from then on.
I think a lot about this (bad drivers) and I’m not really sure how to fix it since I think it’s really a problem of underlying selfishness and perceived-exceptionalism mixed with overestimation of skill.
Nailed it.
Mostly the selfishness part. The whole idea of being courteous with other people on the road just doesn’t exist.
Sadly, this also extends to bicyclists. Entitled instead of courteous.
If you actually ride in one, you do notice some off behaviors that I didn't pick up while just driving alongside them. That said, I agree that the bad human drivers have done things far, far worse than any of the cars.
The biggest gripe with riding in one is that they're slow, both because of super cautious driving and because they won't take freeways yet.
A month ago I saw a Waymo turn left into a tiny alley in Palo Alto and continue at full 25mph speed, which was alarming. I guess the alley is marked as a regular road in the software? Highlights how even if it's safer than humans on average, they need to minimize these weird behaviors in order to get socially accepted and avoid $$$ liability when there is an accident.
I have only taken a couple Waymos but I had the opposite experience. They were much faster and more decisive than I expected. They do apparently learn from surrounding drivers and this was LA so maybe that explains the difference.
I live in the bay and occasionally ride Waymo in SF and I pretty much always have a good time.
I visited NYC a few weeks ago and was instantly reminded of how much the traffic fucking sucks :) While I was there I actually thought of Waymo and how they'd have to turn up the "aggression" slider up to 11 to get anything done there. I mean, could you imagine the audacity of actually not driving into an intersection when the light is yellow and you know you're going to block the crossing traffic?
Semi-related, but just once in my life, I want to hear a mayoral candidate say: “I endorse broken windows theory, but for drivers. You honk when there’s no emergency, block the box, roll through a stop sign — buddy that’s a ticket. Do it enough and we’ll impound your car.”
Who knows, maybe we’ll start taking our cues from our polite new robot driver friends…
This always astounds me about cities who have a reputation for people breaking certain traffic laws. In St. Louis, people run red lights for 5+ seconds after it turns red, and no one seems to care to solve it, but if they'd just station police at some worst-offender lights for a couple months to write tickets, people would catch on pretty quickly that it's not worth the risk. I have similar thoughts on people using their phones at red lights and people running stop signs.
It’s amazing how effective even a slight amount of random law enforcement can be.
Several of the hiking trails I frequent allow dogs but only on leash. Over time the number of dogs running around off leash grows until it’s nearly every dog you see.
When the city starts putting someone at the trailhead at random times to write tickets for people coming down the trail with off-leash dogs suddenly most dogs are back on leash again. Then they stop enforcing it and the number of off-leash dogs starts growing.
Random sampling over time is substantially as effective as having someone enforce the law 100% of the time. It's something like how randomized algorithms can be faster than their purely-deterministic counterparts, or how sampling a population is quite effective at finding population statistics.
It feels less fair though. When everyone is driving x mph over the limit but only you get pulled over, it sucks. So I agree for efficiency of enforcement, but I'd rather see 100% enforcement (automated if possible), with more warnings and lower penalties.
It doesn't just "feel" less fair, it often is -- bc it's not truly random, it's selective enforcement which leads to things like "driving while black".
[delayed]
In europe we use traffic cameras for this. Going through red light? A bill is in your mailbox automatically. No need for a whole police station.
In Massachusetts, USA, red light cameras were illegal until very recently, due to a 70s era law specifying that a live policeman had to issue a citation for something like that. From well before traffic cameras were common.
We had a pilot program in NJ for them, they were universally hated. People would slam brakes on and be hanging over the edge into intersection and throw their car into reverse panicking to avoid the ticket, ended up causing a ton of new accidents so the program was never continued. In newark people shot at the cameras: https://www.nj.com/news/2012/08/shoot_out_the_red_lights_2_t...
Hitting the brakes and getting rear ended is barely even a crash compared to T-boning someone or plowing over pedestrians
I didn't say that. I said they'd panic and throw their vehicle into reverse. Cars/trucks can take the hit, motorcycles/bicycles not so much.
People say Bay Area drivers are bad, but holy shit. People in New Jersey must be retarded. I wonder how they survive. This is hilarious.
Thankfully sawzalls are cheap and plentiful so people can use much safer practices to disable/remove them:
https://www.cbc.ca/news/canada/toronto/parkside-drive-speed-...
Before they were common, yes, but they existed in active use back in the 1960s in the Netherlands: https://en.m.wikipedia.org/wiki/Traffic_enforcement_camera
Put a single live policeman in front of 100 camera screens
In most the USA, or at least Arizona, you have to serve someone. Just dropping something in a mail box doesn't mean dick. The very people that invented the traffic cameras up in Scottsdale were caught dodging the process servers from triggers from their own camera.
Another words, you have to spend hundreds of dollars chasing someone down, by the time you add that on to how easy it is to jam up the ticket in court by demanding an actual human being accuse you, it's not the easy win some may think. You're basically looking at $500+ to try and prosecute someone for a $300 ticket.
NY is not Arizona. They have the plate and send the fine to whomever the vehicle is registered to. If the fine isn't paid they flag the plate and impound the car if it's driven in their state.
In FL, a speed camera can give a car's owner can a ticket without needing to know he was the driver. Your perspective is not true nation wide.
"The registered owner of the motor vehicle involved in the violation is responsible and liable for paying the uniform traffic citation issued for a violation"
http://www.leg.state.fl.us/Statutes/index.cfm?App_mode=Displ...
Arizona also did stakeouts to try and catch this guy:
https://www.nbcnews.com/id/wbna32806142
In CO we have automatic traffic cameras, and to my knowledge they just mail you the ticket, which is usually only a fine (and no license points). Its one of those “automatic plea” tickets where if you fight it, you fight (and risk conviction on) the actual offense, while if you just pay the ticket it will automatically get downgraded to a less serious offense (IE parking outside the lines).
Not in New Jersey. I visited my parents and didn’t stop for a full three seconds before making a right on red on a deserted road at night and they fined my dad.
This isn't true we've had plenty of programs where red light camera tickets were rolled out.
Voters just really don't like them.
They were rolled out but the mailed tickets are legally meaningless, someone has to actually hunt you down within a short timespan (I think 90 days) to create any binding requirement to address it.
https://rideoutlaw.com/photo-radar-tickets-in-arizona-a-comp...[delayed]
We have them in the US too, but it varies widely by jurisdiction because they're regulated at the state level and policed at the local level.
Oh and it's not a bill, it goes through the legal system so people have the right to argue it in court if they want.
Here in my country they removed the cameras in the second largest city after a trial period. It took too much effort to filter out police colleagues running a red (in police or civilian vehicles).
NYC is ramping up on this as well.
https://www.nyc.gov/site/finance/vehicles/red-light-camera-v...
Sweden: Their locations are public. There is even an official API.
They are mostly located in sane places.
Apps like Waze consume this API and warn drivers if they’re at risk of getting caught. It’s the deterrence/slowdown at known risky spots they’re after, not the fine, I guess.
I heard that apps warning drivers this way are illegal in Germany?
Aside: what's up with the traffic speed cameras in Sweden? It feels like they're not designed to catch anybody. In my recent drive there it seemed like most of the cameras were in an 80 zone just before it switch to 50 for a tiny town. They wouldn't catch a typical driver who does something like 10 over everywhere -- they would likely have already started slowing down for the 50.
In my city in Canada, that camera would be in the 50 zone.
Nobody thinks it's racist to enforce traffic laws. People think it's racist to selectively enforce traffic laws by race, which usually takes the form of police pulling over Black drivers at higher rates. (But it can also mean installing more traffic cameras in minority neighborhoods!)
Try driving anywhere in the world that's not Western Europe or The USA and you'll quickly see how advanced even our worst cities are when it comes to traffic.
Last time I was in China drivers simply go through four way intersections at top speed from all directions simultaneously. If you are a pedestrian I hope you're good at frogger because there is a 0% chance anyone will stop for you. I really wonder how self driving cars work because they must program some kind of insane software that ignores all laws or it wouldn't even be remotely workable.
When I was living in China I got used to crossing large streets one lane at a time. Pedestrians stand on the lane markers with cars whizzing by on either side while they wait for a gap big enough to cross the next lane. It's not great for safety, to put it mildly, but the drivers expect it and it's the only way to get across the road in some places. I was freaked out by it but eventually it became habit.
Then I came back to the US and forgot to switch back to US-style street crossing behavior at first. No physical harm done, but I was very embarrassed when people slammed on their brakes at the sight of me in the middle of the road.
It is kinda funny watching people complain here after visiting almost anywhere in Asia. Can't speak for Japan or Korea though.
I've never been to SK, but in Japan things are -- unsurprisingly, as one might guess -- very orderly. For the most part (in cities at least) you don't jaywalk, even when there are no cars on the road.
Wait, so all the sibling comments are actually proposing bringing NYC traffic to a gridlock?
Phone while stopped at a red light is explicitly legal here. I don't think it's been a problem?
People are risking their lives and the lives of others, and a fine is supposed to be the thing that finally gets them to comply?
Yes.
If they run a red light today there is some small chance they will injure/kill someone.
If they run a red light with a camera, there is a 100% chance they will receive a ticket.
The key factor is not the magnitude of the penalty (i.e. whether someone dies or they receive a fine) but the chance that they will encounter the penalty.
This is what the points system is for.
Any individual infraction might only be a small fine, but it adds points to your license. Collect enough points and you risk license suspension.
I’ve known a couple people who got close to having enough points for license suspension. They drove perfectly for years.
That sounds reasonable to me. Everybody makes mistakes, but nobody should be consistently making grievous mistakes capable of causing serious injury or death to other motorists on a regular basis.
I'm less concerned with a little speeding than I am with blowing through lights and stop signs.
You've got me: I believe that people respond to financial incentives. I don't think this is a radical position.
I think (or at least I hope) St Louis is primarily focused on reducing their sky-high murder rates. But who knows.
New startup idea just dropped.
> Who knows, maybe we’ll start taking our cues from our polite new robot driver friends…
I think this could be an interesting unintended consequence of the proliferation of Waymos: if everyone gets used to drivers that obey the law to letter, it could slipstream into being a norm by sheer numbers.
Ultimately I wouldn’t support this level of snitching (especially in our current political env) but I’ve had the idea of:
A bounty program to submit dash cam video of egregious driving crimes. It gets reviewed, maybe even by AI initially and then gets escalated to formal ticket if legit. Once ticket is paid, the snitch gets a percentage.
Again, I am fundamentally against something like this though, especially now.
Blocking the box is a ticket in London. It works.
Edit: let me clarify: there is a camera on every intersection which automatically gives a ticket to everyone who blocks for >5sec. That works.
It is in NYC also, except it's entirely unenforced. We need a lot more red light cameras.
The nominal regulations on automotive behavior is pretty sufficient throughout the US, the main problem is that in most parts of the country traffic law may as well be a dead letter.
> It is in NYC also, except it's entirely unenforced.
It's enforced in the worst congested zones, the intersections around tunnel entrances and midtown, but as I said in my other comment usually by parking enforcement not NYPD.
A workaround in the law is to throw your turn signal on if stranded in the box, this doesn't count as blocking the box.
It is in NYC as well and it's usually enforced by parking enforcement (doesn't carry points but it has a steep fine), if NYPD writes it also comes with points but in my experience they'd rather let the walking ticket printers do it.
I paid a ticket for this in NYC.
Great! and if enforcement were consistent, rule-breaking behavior would probably decline:
> Quick, clear and consistent also works in controlling crime. It’s not a coincidence that the same approach works for parenting and crime control because the problems are largely the same. Moreover, in both domains quick, clear and consistent punishment need not be severe.
https://marginalrevolution.com/marginalrevolution/2015/09/wh...
If you look into the fleet size serving Waymo service areas, it's remarkably small. But because they work 24/7 they serve up a lot of rides, punching way above their weight in terms of market share in ride hailing.
Their effect on traffic and how drivers behave will be similarly amplified. It could turn out to be disastrous for Waymo. But I suspect that low speed limits in New York will work to Waymo's favor.]
Real question for waymo will be snow and ice, or do they just get parked in that situation when demand is highest?
I've seen reports that they've been testing Driver 6 in snowy places like around Lake Tahoe and the upper Midwest last winter. I suppose this year we'll find out how well that went.
In many places outside the USA they just use cameras for box blocking, stop sign rolling, speeding...and there is a system for honking also. But many in the states think automation here is too Orwellian.
They do that in NY too. The worst offenders inevitably have fake/defaced/covered/no license plates. That should be cracked down on very hard but the police and prosecutors are strangely reluctant.
We have all of those things in the states too. Just not ubiquitous.
We don't have much of it, not compared to Europe or Australia. This is a solved problem, but we don't want to solve it.
A sound solution in general, but the majority of police and firefighters and government employees with a connection to law enforcement cover their license plates with magnetic 'leaves' and so on. It's an undocumented perk for government employees.
If only Mitch Hedberg was still alive: https://youtu.be/zonQXdmIlqQ?si=EBrpJiCk2XlhGJIs&t=97
Isn't that what speed cameras are about? Seem a lot more efficient and cheaper. I got a few tickets, nothing too serious just ran the yellow a little too close and 40 in 25. And if def changed my behavior
We don't go after moving violations anymore (in NYC) because the driver might have a bad reaction. True story.
NYPD cops don't like enforcing traffic violations: https://i.redd.it/w6es37v1sqpc1.png (License holders and drivers on the road are up in the same period that summonses are down, too. Traffic is up since pre-covid.)
Now that I live in Toronto we face the same challenges. Politicians may introduce traffic laws to curb dangers and nuisances from drivers, but police refuse to enforce them. As they don't live in the city, cops seem to prefer to side with drivers over local pedestrians, residents or cyclists who they view antagonistically. Broken window works for them because they enjoy harassing pedestrians and residents of the communities they commute into.
So there is a bigger problem to solve than legislation.
Police quiet quitting and arbitrarily choosing what laws they feel like enforcing is a huge problem.
The most effective fix vis a vis traffic is simply automating so much of it with speed averaging cameras and intersection cameras and taking police out of the equation and retasking them to more important things that only they can do.
Don't police have quotas any more? 40 years ago everybody knew not to speed at the end of the month because a cop that would normally give you a warning for a small speed infraction would give you a ticket instead so they could make this month's quota.
Part of the problem is we have police doing far too many jobs. We need to separate out traffic enforcement, mental health responses, and other works into their own focused units. Especially the mental health responses, as far too often police refuse to or (at best) don't know how to de-escalate in those situations.
The current Democratic nominee and frontrunner for NYC mayor plans to do exactly that! He plans to create a Department of Community Safety to take over mental health responses from NYPD.
Bringing a gun and a taser to every problem guarantees that a lot of problems will be "solved" with the wrong tools. It's impossible to train enough people to carry guns and tasers and use them wisely.
It’s also expensive training and on-going cost when you add it all up.
Canada budgeted the cost of arming its border officers at ~$1 billion.
In the first 10 years, they fired them 18 times. 11 were accidents and the rest were against animal, usually to euthanize it rather than defend.
Works out to ~$55 million per bullet.
https://www.cbc.ca/news/politics/cbsa-border-guards-guns-1.4...
I agree we need to separate these responsibilities, but when it comes to mental health response, the police themselves are often opposed to alternatives, even while they complain that they're not mental health providers and often can't do anything in those types of situations.
In my city, we've had an underfunded street response program for a few years now, but a lot of people (including a lot of people who don't live here) see it as antagonistic to police and police funding, when really it should just be part of a holistic system to address social issues.
It makes no sense to me that the people who ostensibly care the most about addressing crime and "disorder" on the streets are often the most oppositional to programs that might actually address some of the underlying issues (not all of course, but some).
My wife and I took a road trip that included time in SF last year and seeing a Waymo was pretty neat.
To save some money, we stayed in downtown Oakland and took the BART into San Francisco. After getting ice cream at the Ghirardelli Chocolate shop, we were headed to Pier 39. My wife has a bad ankle and can't walk very far before needing a break to sit, and we could have taken another bus, we decided to take a Waymo for the novelty of it. It felt like being in the future.
I own a Tesla and have had trials of FSD, but being in a car that was ACTUALLY autonomous and didn't merely pretend to be was amazing. For that short ride of 7 city blocks, it was like being in a sci-fi film.
Why does tesla pretend to be autonomous? My friends with tesla fsd use it fully autonomously. It even finds a spot and parks for them.
The company selling the car is adamant that none of their cars are fully autonomous in every single legal or regularity context. Any accident caused by the car is 100% the fault of the driver. But the company markets their cars as fully autonomous. That's pretty much the definition of pretending to be autonomous.
It's a level 2 system, it can't be operated unattended. Your friends are risking thier lives as several people (now dead) have found out.
Sources? Havent heard of deaths except total idiots sleepping at 80mph.
If the car needs any occupant to be awake, it is not an autonomous vehicle.
Some of the best marketing ever behind convincing people that the word "autonomous" does not mean what we all know it means.
Are you trying to draw a distinction between sleeping versus looking away from the road and not paying attention to it? I expect both situations to have similar results with similar levels of danger in a Tesla, and the latter is the bare minimum for autonomous/unattended.
You don't need to cite accidents when you're stating the true fact that the system is not approved for unattended use.
Wikipedia lists two fatal crashes involving Tesla FSD and one involving Waymo.
They key difference is that the Teslas killed their passengers, the Waymo hit someone outside the car (and it wasn't the Waymo's fault, it was hit by another car).
Yes. [1] That incident got considerable publicity in the San Francisco media. But not because of the Waymo.[2][3]
Someone was driving a Tesla on I-280 into SF. They'd previously been involved in a hit-and-run accident on the freeway. They exited I-280 at the 6th St. off ramp, which is a long straightaway. They entered surface streets at 98 MPH in a 25 MPH zone, ran through a red light, and reached the next intersection, where traffic was stopped at a red light. The Tesla Model Y plowed into a lane of stopped cars, killing one person and one dog, injuring seven others, and demolishing at least six vehicles. One of the vehicles waiting was a Waymo, which had no one on board at the time.
The driver of the Tesla claims their brakes failed. "Police on Monday booked Zheng on one count of felony vehicular manslaughter, reckless driving causing injury, felony vandalism and speeding."[2]
[1] https://www.nbcbayarea.com/investigations/waymo-multi-car-wr...
[2] https://www.sfchronicle.com/sf/article/crash-tesla-waymo-inj...
[3] https://www.youtube.com/watch?v=ULalTHBQ3rI&
Only one of the two, and it's not nearly enough data to draw a conclusion one way or another in any case.
Wikipedia lists at least 28 fatal crashes involving Tesla FSD:
https://en.wikipedia.org/wiki/List_of_Tesla_Autopilot_crashe...
FSD is not Autopilot despite the names being equated today, and obviously Waymo operates at a far smaller scale. It's simply not enough to compare raw numbers of fatal incidents to draw this conclusion. That's not to justify taking your eyes off the road when enabling FSD on a Tesla, but the OP did not suggest that either anyway.
If I can't use the center console to pick a song on Spotify without the car yelling at me to watch the road, it's not autonomous.
No, rather, if the manufacturer of the self-driving software doesn't take full legal liability for actions taken by the car, then it's not autonomous. This is the once and final criterion for a self-driving vehicle.
Sounds like we're in agreement then.
Right now, Tesla skirts legal liability by saying that the driver needs to watch the road and be ready to take control, and then uses measures like detecting your hand on the wheel and tracking your gaze to make sure you're watching the road. If a car driving on FSD crashes, Tesla will say it's the driver's fault for not monitoring the drive.
It's just pretending to do that, seemingly?
I had my second Waymo ride in SF 2 weeks ago and I had to press the support button: it was behind a large bus that was backing up to parallel park. The bus was waiting for the Waymo to get out of the way while the Waymo was waiting for the bus to move forward.
It took only a few seconds for a human to answer the support request and she immediately ordered the Waymo to go to a different lane. Very happy with the responsiveness of support, but there's clearly still some situations that Waymo can't deal with.
I was in a Waymo in SF last weekend riding from the Richmond district to SOMA, and the car actually surprised me by accelerating through two yellow lights. It was exactly what I would have done. So it seems the cars are able to dial up the assertiveness when appropriate.
It doesn't seem impossible technically to up the assertiveness. The issue is the tradeoffs: you up the assertiveness, and increase the number of accidents by X%. Inevitably, that will contribute to some fatal crash. Does the decision maker want to be the one trying to justify to the jury knowingly causing an expected one more fatal incident in order to improve average fleet time to destination by 25%?
Nah, it's not that simple. Excessive passiveness causes ambiguity which causes its own risks.
You want the cars to follow norms, modifying them down slightly for safety in cases where it's a clear benefit.
Reinforcement learning is a helluva drug. I'm sure by now Waymos can time yellows in SF to within a nanosecond, whereas humans will only ever drive through so many yellows will never get that much training data.
A human can know the yellows on a few routes. A Waymo can pull over, observe a given intersection for an hour, and tell every other Waymo that exists precisely how long that light lasts.
It's not just collecting the information; it's the ability to spread it.
An autonomous vehicle's hivemind knows the exact duration of all yellow lights, even ones that vary based on traffic flow.
Not if they change the timing.
I'm sure "timing of yellow" is only a few parameters in its network at this point. And it's continuously training, it can probably one-shot the timing changes ( one taxi ride through maybe 3 lights ).
When red-light cameras are installed at an intersection, the number of rear-end accidents typically increases as drivers unexpectedly slow down instead of speeding up at yellow lights.
The cost of these accidents is borne by just about everyone, except the authority profitably operating the red lights. (To be fair, some statistics also show a decrease in right-angle collisions, which is kinda the point of the red-light rules to begin with.)
That seems only like a temporary problem until people get used to actually stopping at red lights, as they are supposed to. After the initial acceptance phase, it should minimise accidents over the longer term.
Unless there is a warning of how long is left on the yellow light, it’s an unsolvable problem because there is an asymmetric risk of stopping vs accelerating
The lights should be designed so that if you don't have enough space to stop with a mild deceleration you should just go through. If a mild deceleration get you rear ended then of course that's an unsolvable problem
Then they shorten the yellow so that it isn't "with a mild deceleration" but a full-on stomp-on-the-brakes stop.
>speeding up at yellow lights
I remember reading somewhere accelerating at orange light is actually a ticket-able offense?
My memory may be outdated or only local to my jurisdiction but my understanding is that yellow means “do not enter the intersection” where “intersection” begins before the box, usually with some alternate street indicator, like broken white lines turning to solid, at a braking distance that accounts for posted speed limit and yellow light duration.
Each Waymo is equipped with multiple cameras (potentially LPR), LIDAR, etc. The car knows when the vehicles around it are breaking traffic laws and can provide photographic/video evidence of it. Imagine if Waymo cars started reporting violators to the police, and if the police started accepting those reports. Someday they might.
Isn't it too dystopian to have cars follow you around and report you to authorities? I can easily imagine some bad scenarios.
Yes it could potentially be very dystopian for human drivers. That doesn't mean it won't happen. Police departments could make a lot of extra money from the additional traffic tickets; there is a financial incentive for them to do this.
>> could you imagine the audacity of actually not driving into an intersection when the light is yellow and you know you're going to block the crossing traffic?
I wonder how many Waymos following the rules would be needed to reduce gridlock.
Waymo in SF pretty much drives like a human, and that includes doing human things like cutting lanes, stopping wherever it feels like, driving in the bus lane etc. I think it’ll be fine in NYC
Waymo in LA also drives pretty much like a human here would, which includes: not yielding for pedestrian-only crosswalks, running red lights, driving in the oncoming traffic/suicide/bike lane, occupying two lanes, blocking entrances/driveways/intersections, and stopping/parking in no-stop/parking curbs.
They're only really phenomenal at not hitting things; they really aren't good/courteous/predictable drivers under most conventional definitions.
Still, I think rollout in NYC will be fine. NYC generally drives slower and much less aggressively than LA, and slower gives the Waymo plenty of reaction time to not hit things.
I can attest to that. I live in Orange county and occasionally see Waymos when I go to LA, and they'll do things like merging with very little gap or merging in the middle of interactions.
SF traffic is but a single speck of nyc
Traffic? The issue half the time in NYC is the drivers. I can't compare it to SF since I haven't been there in a while but I still thought it was not as congested to compared to NYC.
NYC has a greater population and also has a greater number of registered cars compare to SF however.
As a comparison, I feel safe riding a motorcycle in SF. I don’t think I would ever ride a motorcycle in NYC.
Riding safely requires predicting what the cars around you are about to do. I find it an order of magnitude harder to predict driver behavior in NYC.
I was on Market Street yesterday on my bike next to a Waymo. A bunch of cars were blocking the intersection when we had the green. The light turned red and the cars blocking the intersection were able to move. I decided to stay, but the Waymo sped through despite the light being red. I regretted not crossing.
Driving in most of the city isn't that bad. Even most of Manhattan is fairly regular driving compared to most of the country. It really isn't until you're near midtown that the insanity kicks up.
occasionally ride Waymo in SF and I pretty much always have a good time
Surreal. You have to step back and absorb what you just said. We have self driving cars, insane.
It's fascinating seeing all the comments elsewhere anytime Waymo starts testing in another city along the lines of, "ah, but how will they handle X, Y, and Z here?? Checkmate, robots!" despite having already launched service in several other cities.
Granted, NYC is the biggest city in the US, so maybe that sort of reaction is more reasonable there than when people in Dallas or Boston do it.
Since Waymo is very reliable in LA and SF, you will be just fine in NYC.
Your grid system is far less of a challenge than the amount of hills, twists, narrow streets and low visibility back streets in California.
I genuinely think the most complicated challenge for Waymo in NYC will be…winter snow and ice.
NYC is a new set of challenges. As you already mentioned snow and ice is new. But also missing the high density of people and cars per square area. Behavior of drivers and pedestrians are different and much less polite. I can see it working in NYC but "just fine" is a bit of an over confidence... at least not for the first few years before they learn to deal with these issues that they don't face yet in LA and SF.
Don't forget the unique NYC challenge of people waiting to cross the street not on the sidewalk but just into the street itself.
People in LA wait to jaywalk on the street or even in the suicide lane all the time. The Waymos handle it fine; generally by asserting it has right of way unless collision is obviously imminent. They'll even happily swerve around you if you're too far out.
We do have narrow streets in LA with double parked cars, cars parked in the street only allowing one car through the middle at a time, and plenty of construction closures and obstacles.
Why do so many NYC people think there’s comically no cars in LA or neighborhood streets?
Also, I can assure you LA drivers are a tad bit more aggressive than NYC drivers (less honking and flicking off though, LA people are more a drive you off the road or into the shoulder sort of passive aggressive).
I was born and raised in NYC and have lived in LA for quite some time, still going home often for family. I’m really struggling with reading these “NYC is unique” comments regarding Waymo traffic.
There's a weird thing where people like to brag(?) that their city has the craziest, worst, drivers/roads.
What snow and ice? We don't get much of that anymore. That was actually the last thing I am worried about here. I really want to see how Waymo does with NYC drivers and obstacles(double parking on block where sometimes you have to pull in your mirrors just to get by(if you even take the chance instead of just laying on your horn). In some neighborhoods it can be so annoying.
I think what OP means is Waymo's most challenging rollouts will be to places that do get lots of snow and ice.
I think a well designed winter specific FSD system is probably more safe in snow and ice than a human. For instance downshifting to ensure wheels continue to spin on slippery surfaces, subtle corrective steering to keep the vehicle within its lane, etc. should be easier for a FSD car since it won't panic and over-correct like most people do in those situations.
And if the car reduces speed when appropriate and some assholes start tailgating it, it won't suffer the anxiety of holding up 10 cars that want to drive beyond the safe, reasonable speed for the snowy/icy conditions.
Pretty much all electric cars have single speed transmissions, so there's no downshifting. And modern vehicles have electronic stability control, anti-lock brakes, automatic emergency braking, and several other safety systems. It's pretty hard to overreact with those enabled. The main issue is that people exceed safe speeds for the conditions, making them unable to brake or turn in time to avoid a collision.
Right now, most self-driving software will refuse to activate in conditions of poor visibility. I've had that happen with Tesla's FSD, though in that case it was snowing so much that the road should have been closed. Also when the snow is deep enough that your front bumper becomes a plow, it will refuse to activate.
I would argue those two areas are very different though. The Bay Area is not as dense or as many aggressive drivers as NYC.
My point wasn’t to say they are the same, more that SF and LA (I would guess) have covered and defeated almost every single challenge and obstacle for an urban environment (sans..weather).
LA also has far denser areas than SF, places like DTLA and Koreatown are more dense than most boroughs in NYC (sans Manhattan).
My point was that it will be interesting to see how well it works in NYC where the only way to drive is aggressively in city streets breaking the rules. LA has its share of aggressive driving but as someone who has driven in both, NYC felt like I had to break the rules to go anywhere, LA not so much.
Hard to really compare a tiny piece of LA and say it’s more dense and compare it to borough that is in the same range but also magnitudes larger total pop.
> I genuinely think the most complicated challenge for Waymo in NYC will be…winter snow and ice.
Nah, I'm betting it'll be the locals. They'll get pissed off at it remaining stopped when it shouldn't and do shit like start ramming into it. I've had it happen on the island when I stopped at a yellow. NYC is a lot more chaotic than any other US city I've driven.
Oof I don’t know about that. Driving in NYC is much different than San Francisco. Frequent lane departures, cutting into heavy traffic despite technically lacking right of way, and other moderate rule breaking is required to get anywhere. Boston will be even more challenging due to the hundreds of convoluted intersections.
We've been getting less and less of those, though. And even then, it's just for a few days. Last year was a bit worse, but two years ago it was very, very mild, I think. Yay global warming?
The thing is waymo at least in LA specifically geolocks you from those hilly areas. Imo it also is not assertive enough and drivers seem to be learning one can bully a waymo on the road.
LA and SF are not close to close to the complexity of nyc traffic and pedestrian culture
TIL.
LA doesn’t have complex traffic? What sort of traffic do we have in LA then?
LA is walkable, it’s lazy (and mostly incorrect) to say LA isn’t walkable.
LA County is massive, and depending on where you want to pick a comparison from, you may prove yourself either right or wrong.
Traffic on the sidewalk is a daily occurrence and often necessity in NYC. I'm not sure exactly how every area in LA is but often (as in pretty much constantly every day) in Manhattan or Brooklyn drivers don't obey the lines on the road, don't care to bump objects and cars to fit into a spot, literally threaten to hit other cars to get anywhere.
There's a bit of a "do what you have to" mentality with NY traffic that I haven't seen in any other east coast or mid-western city. I think that poses some unique challenges that I've often seen video of Waymos freezing up when facing similar scenarios, which could cause huge issues in most of the city.
You articulated this very well, thanks.
LA is extremely similar. Often can only make unprotected turns at lights while it’s red and you’re in the box, you have to wait at the top of a hill and have your car sideways while the oncoming car has space to drive up a hill, cars trying to give you space so you can drive through a line of traffic into the adjacent traffic pattern.
The “freezing” issues are very real though (and frustrating), and it’s what most everyone who uses Waymo in any city right now jokes/complains about. Waymo can often get into a weird game of “chicken” when there’s a four way stop with pedestrians, and any slight movement from the intersection can often make the car stop - so the pedestrian stops - the the Waymo finally moves again, but then pedestrian also started moving so the Waymo stops again and the pedestrian stops caring.
All this to say, I really don’t think there’s much that will be different. Go to Hollywood or Santa Monica
there is nowhere in LA with the complex intermingling of pedestrian, car, bicycle, and motorbike traffic of anywhere in the boroughs other than Staten.
LA it’s gridlock or go. There’s nothing complicated about it other than strategizing where is gridlock and where is Go.
is it just me or its common that in nyc people bike in most cases like there is no traffic lights at all? this to me is prob the most challenging
NYC is also one of the most regularly built out cities, in stark contrast to Boston, for instance. OTOH roads here may be 3 or even 4 levels high at the same point (e.g. where Manhattan bridge meets Brooklyn), and GPS is sometimes way off in canyons between skyscrapers.
> OTOH roads here may be 3 or even 4 levels high at the same point
And here I thought Chicago was complex with lower lower Wacker (just 3 levels).
> GPS is sometimes way off in canyons between skyscrapers
This is probably very challenging for human drivers using navigation, but probably no nearly as much of a problem for a Waymo car with onboard 3D maps of the entire operating area.
HD mapping gives much better than GPS I would imagine.
I think the main difference for NYC is that quite a few streets and intersections routinely have 10x to 100x the pedestrian traffic of the busiest such intersections in pretty much any other American city.
That's not to say that I don't think it'll be able to handle it, just that it'll be a new challenge. I wonder if their current program of apparently trying to positively track every single moving object in range will survive that, or whether they'll need to figure out some algorithm to prioritize objects that are more likely to be of concern to it. And there probably are more than a few places where pedestrians are numerous and densely crowded enough that you can't positively track all of them, even with a bunch of LIDAR sensors.
NYC (at least the parts I've spent a bit of time in) is pretty grid like with fairly simple roads. The drivers are the hard part :)
I am excited to see them tackle Boston at some point because of how strange some of those roads are. The first time I had ever been I came to an intersection that was all one ways and there were like 7 entry/exit points. My GPS said turn left, but there were three paths I'd consider left. Thank god I was walking.
And I don't really pose much doubt because it seems like Waymo's rollout plan has been solid, but I'm just interested to see how well they tackle different cities.
Yeah I’m skeptical that robots will ever be perfect drivers but the bar isn’t perfect it’s better than human and that’s certainly possible.
Yup, the data so far seems to indicate that Waymo is substantially safe than average drivers. Obviously it's not inclusive yet since the tech is still new (and while the study I'm thinking of was done by a third party, it's still Waymo that handed over the data and paid them to analyze it).
There is a question, NYC driving gets by with everyone driving aggressive and breaking road rules. That is something that does not exist as much in other markets.
My complaint with Tesla city FSD is that it’s not quick or aggressive enough. It will come to long and complete stops and other things that will not work well in NYC.
I think you'd be surprised how aggressive driving has become normalized throughout the world?
I drive when I travel, though not in NYC these days but being a pedestrian gives you a good enough view. LA from a city driving perspective at least for me is not comparable to most of NYC.
Roads in Texas specifically just seem to do whatever they want, whenever they want. It's really apparent that Texas local roads used to be wagon trails.
The grid system in NYC seems like a good alternative for a rollout. Though the current NYC human drivers will hate these things. I also expect LOTS of vandalism.
This type of edge case covering is pretty essential to the jobs of most on this board. It doesn't surprise me to see it.
What I don't get about the "checkmate robots" mentality is that, like, get it working in sunny California with no snow, and then get it working in the snow, seems like the way to do it, not, solve all possible problems before anyone knows you're even trying and can make fun of you.
>despite having already launched service in several other cities.
Why does having launched in other cities matter if the new city brings up things that none of the other launched cities do?
For example the first thing I can think of new for New York is snow and ice.
It's my understanding that self-driving cars don't really account their acceleration and braking for roads that could sometimes be very slippery due to snow and ice.
> "Why does having launched in other cities matter if the new city brings up things that none of the other launched cities do?"
New requirements come up all the time in technology. The existence of a new requirement isn't in and of itself justification for skepticism - is there a particular reason to believe that Waymo is not capable of solving for the new requirement?
The answer may be yes, but simply "ahah! It would need to do [new thing]!" is insufficient. "[new thing] is likely intractable because [reason]" would be more justification for skepticism.
> "It's my understanding that self-driving cars don't really account their acceleration and braking for roads that could sometimes be very slippery due to snow and ice."
Sure, but like above - is there a reason this is an intractable problem?
I'll throw this out there: your human-driven car already accounts for acceleration and braking on slippery roads, without the need for the human. Traction control systems and electronic stability control systems exist! They're in fact incredibly common on modern cars.
>I'll throw this out there: your human-driven car already accounts for acceleration and braking on slippery roads, without the need for the human. Traction control systems and electronic stability control systems exist! They're in fact incredibly common on modern cars.
These systems don't help with the problems I am talking about.
You have to drive completely differently in heavy snow, significantly slower, brake sooner, turn less sharp, accelerate much slower, leave significantly larger gaps, leave space to move out of the way and be ready to move if someone behind you is coming at you too fast and can't stop in time, etc. I've spend my entire life in the midwest.
The traction control system in my 2023 camry didn't help one bit when I applied the brakes on black ice and the car didn't react at all, it just kept sliding at the same speed across the ice.
That all sounds like something that should be easier for a robot to do than the typical human. If programmed for how to drive in heavy snow, a robot should be able to switch driving modes much easier than the typical human brain.
Waymo has been trained in Buffalo NY for winter conditions, unlike most NYC drivers.
Is it possible to train a machine to drive in snow? Yes. But consider that humans are trained to do so by means of things like: actually crashing, observing others crashing, talking to people who crashed, and all of the above is highly localized. Where I live there are many days in winter when someone not from the immediate area should not drive at all. But I might if there was a good reason because I have 25 years experience with the specific roads, conditions, how those conditions relate to wind and on and on. Training a machine to know all that seems feasible but unlikely to be commercially viable. It's just not a problem that can be solved with a simple closed loop control system like ABS or traction control.
> talking to people who crashed
A reasonable counterargument is that autonomous vehicles can actually do that to a degree that is much, much more effective than humans. You might have 25 years of experience, but at 8 hours a day for 365 days of those 25 years we'd only need 8 cars driving for a year to match that. After all, training data and event logs generated by cars can be shared, and models can be upgraded all around. And of course that scales to more than 8 vehicles rather easily.
The interesting snow & ice problem for me is that humans will drive in winter conditions that are unsafe -- for example white-out blizzards. Robocars won't be able to drive in a white-out blizzard, so they'll likely refuse to do so. Humans should also refuse to drive, but people drive anyways.
NYC doesn't generally get white-out blizzards, so refusing to drive in them is quite feasible.
I come from way up top on that globe of ours. I have driven in frankly apocalyptic snowstorms. They're an insidious problem to solve, but I remain optimistic. Back home, they will close specific roads due to snowstorms, but what do you do about the cars already on the road? You can't stay put for 16 hours can you? So you move as slow as possible, sometimes as low as 5 kilometres an hour. Cause that's the thing about a snowstorm; it's about visibility. You're not risking your life if a dude in skis can go faster than you.
> You're not risking your life if a dude in skis can go faster than you.
Sure you are. You can still drive off the road and into the ditch where nobody can see you. People then die because they don't clear their tail pipe and get carbon monoxide poisoning or they try and walk for help and freeze to death.
Also snow and ice in NYC is a rare event now, not a given like it used to be.
If white-out visibility is the only problem to be addressed then machines seem pretty well placed because they can use very accurate positioning and non-visible light sensors. Unfortunately they probably wouldn't know that there's a 50 yard section of the road that always drifts in when the wind comes from the south and the snow is dry.
Launching in other cities with new problems gives experience dealing with new problems, and the meta-learnings transfer to better processes for adapting to new issues. But yeah, ice and snow are definitely major new environmental factors for New York (and DC, and many other places we are starting to see more serious testing).
Autonomous vehicles can and do take into account surface conditions, there’s not really any reason not to. There are pretty good generative models of the physics of vehicles with different surface conditions, and I imagine part of the data collection they are doing is to help build statistical of vehicle performance based on sensed conditions.
A fair point about weather, but a lot of the assertions are like "how will they handle the double parking and suicidal pedestrians jaywalking across the street??" I'd say most of the concerns just don't sound very unique at all.
For weather, Waymo has clearly started out in warmer climates while slowly building out towards places with colder and colder weather, I'm guessing they're just incrementally getting better at it.
This is great long term for having cars that follow traffic laws since human drivers in NYC are awful (kill/injure pedestrians, bikers, and other street users all the time).
Not so great for getting cars out of NYC and pedestrianizing more of the city/moving towards more “low traffic neighborhoods” as I imagine Waymo and other similar companies are going to fight against these efforts.
Edit: Lots of people talking about human drivers taking advantage of self-driving cars being more cautious/timid. Good news is that once you have enough self-driving cars on the road, it probably slows down/calms other traffic (see related research on speed governors).
I'm not sure why you think waymo would fight against that. People getting rid of their own cars for daily use will increase how often a service like waymo is used for occasional usage. In the long run it would be a win for waymo. Not many people are taking taxis on a daily basis in New York for normal driving, they buy a car if they need to do that because even with the parking bill they will still come out ahead. And once they have their own car they feel like they need to get some use out of it.
> not sure why you think waymo would fight against that
If you were to pedestrianize 10% of Manhattan (or for example all of Broadway, which is being considered), then that’s less area for Waymo to operate and make money. To be clear, this is likely more of a long term issue.
They will probably gain way more by the removal of parking lots that comes with it than by losing rides to pedestrian traffic or bikes.
NYC doesn’t have a lot of surface parking lots.
Waymo would not fight against "people getting rid of their cars", many people in NY who use the incredible public transit system would like to see more car-free streets, which they absolutely would fight against.
Believe it or not, NYC is actually the safest city in the country for pedestrians and bicyclists.[1]
[1]https://www.wagnerreese.com/most-dangerous-cities-cyclists-p...
I’d believe it, but the fact that any pedestrians/bikers are killed/injured by cars in NYC is unacceptable.
Do you mean that in the sense of "anyone getting killed is unacceptable" or the sense of "we need complete separation between cars and pedestrians/bikers, somehow"?
I mean if we required a license to own a bike in NYC we could see a significant reduction in injuries/deaths, same for pedestrians. Cars are already heavily regulated and likely aren't the underlying issue.
There are many ways to interpret data, but one often comes to the conclusion that pedestrians and bikers are the root cause of most accidents.
> pedestrianizing more of the city
Replacing dangerous, dirty, noisy cars with safe, clean, and quiet ones seems like a huge pedestrianizing step.
What's a "low traffic neighborhood"? Does that allow busses or deliveries?
It’s a step in the right direction, but they still pollute (heavy electric vehicles have a lot more tire dust) and take up a lot space (could close roads and build housing or just have more space for the millions of city residents that don’t have/use cars).
LTN still allow buses, emergency vehicles, deliveries, etc. https://en.m.wikipedia.org/wiki/Low_Traffic_Neighbourhood
EVs usually produce much less brake dust, not more, than combustion vehicles.
More tire dust on EVs tho cuz of the added weight and ability to accelerate a lot faster. EVs can really chew through tires.
Brake dust composition is improving tho: https://www.epa.gov/npdes/copper-free-brake-initiative
Thanks for the correction. Updated.
> heavy electric vehicles have a lot more tire dust
It would be interesting to know the fleet-level statistics for this. Driven by humans, EV might wear tires faster because of fast starts and the extra weight during stops. It's possible that the Waymo Driver accelerates and decelerates more smoothly, resulting in less tire wear than a human-driven ICE vehicle.
That would be great! Even if EV pollution is zero, cars still take up a lot of space in a city where space is very limited.
It just means that feral bikers will take over the roads ;)
You know what kills or harms people in NYC are the motorized bikes driving the wrong way and putting people in the hospital, with no charges against the operator because they are usually an illegal alien. Not sure Waymo is going to fix that.
> Not so great for getting cars out of NYC
This will never happen. Not in our lifetimes. And as I get older and less able to walk, I don't want it to happen.
Waymo should add a thin layer of "assertiveness" for actual deadlock that their self-driving architecture could cause.
While in Austin, I was in a Waymo that blocked 3 lanes of incoming traffic while attempting to merge into a lane going into the opposite direction. It was a super unorthodox move, but none of the drivers (even while stopped for a red light) would let the Waymo* merge into their lane.
Thank God for the tinted windows, people were pulling their phones out to record (rightly so). It felt like I was responsible for holding up a major portion of Austin 5 pm traffic on a Friday.
Wish it just asserted itself ever-so-slightly to get itself out.
According to this article, they are doing some of that already. Presumably it will improve:
> Waymos are getting more assertive. Why the driverless taxis are learning to drive like humans
https://www.sfchronicle.com/sf/article/waymo-robotaxis-drivi...
Oh very cool. My only complaint is that it didn't lay on the horn for the last example of the reckless driver nearly causing an accident.
I wonder if this is related to the Foundation Model: https://waymo.com/blog/2024/10/ai-and-ml-at-waymo
I think we're going to see more examples of this as Waymo's popularity grows. Basically human drivers taking advantage of Waymo's far more passive driving style. Maybe some rules of the road will have to change, or the Waymos will get dedicated lanes to solve this problem.
Imagine if we had dedicated lanes for giant Waymos, that could hold dozens of people. The future of transport.
You joke, but the reality is going to be dynamic self-driving buses that don't have preset routes or stops but respond to instant demand.
You'll pay $$$ for a nonstop ride into midtown in a dedicated vehicle, or $ for a short dedicated ride to a self-driving bus you only need to wait a few minutes for, and which will drop you off on your destination block.
So yes -- self-driving buses seamlessly integrated into ride sharing are certainly going to be a major part of 21st century urban transportation. Which will save a ton of time compared to current buses.
You need to think bigger. Once we have separate lanes just for the waymos, we don't need them to be regular roadways. We can scale up the waymo even more and size the lane exactly to the vehicle, maybe even radically redesign the road surface for lower rolling resistance. What a future it will be.
This is surely impossible. Such a thing has never been tried, it could never work.
By forming those waymos like aerodynamic bullets, they could reach ridiculously high speeds on those special lanes. Something like 200 mph should be possible.
Maybe the waymos could be powered by overhead wires?
I'm having a hard time even picturing such a thing, but I have no doubt that Waymo could manage to operate them in cities around the nation, with sufficient re-training.
I would like that just as much as the next guy but the problem of public transport cannot be addressed until you first address the problem of anti social behavior on public transport.
That’s just being around people. We gotta live together as people; the idea that we can atomize ourselves away from the society we live in is more disastrous to the shared social fabric than any amount of people listening to music without headphones.
That isn’t what they meant, and you know it.
This already exists outside of America and is abundant and cheap. It's called public transit.
Public transit doesn't always have dedicated lanes. That's BRT.
that was the joke, yes
Imagine if we went further and put them on rails and interconnected them. Maybe even built dedicated tunnels for them.
I find that in LA people routinely cut off the waymo or refuse to let it in. After all why not it is a robot, not someone who might legitimately harm you like a road rager. It also tends to fail the cultural left turns. That is, sending 2-3 cars left during yellow not just one like in other places. Seeing it stuck awkwardly in an intersection for another cycle from failing to make an assertive left turn is somewhat common.
Waymo also avoids certain challenging environments by excluding it from its service coverage, namely hilly neighborhoods.
>Waymo should add a thin layer of "assertiveness"
well, just couple weeks ago here on the intersection of Middlefield and Shoreline, half-mile from Google headquarters - 100 million times driven by Waymo cars, thousands by us - midday, perfect visibility, perfect intersection with all the markings, lights, etc., we and a Waymo are doing left turns from dedicated lanes on the opposite directions. We were saved from head on collision by the "lack of assertiveness" on the part of my wife as she swerved last moment as the Waymo apparently decided that its left turn point lies way into, very deep into, our trajectory, and it was assertive enough to not care that we were in its path. I almost soiled my pants upon seeing how it went for barreling into us instead of turning.
It looks like the same extra assertiveness like with Uber back then - i.e. to not have an emergency braking and similar features because it gets too much false positives.
Yeah I admit that "assertiveness" isn't the right word here. I've been in Waymo's that have also tried to dangerous moves in front off busses. Maybe "conscientiousness" would be a better definition?
This is awful. Your ride takes just a bit longer, so you want it to take more risks in decisions?
This is how you ruin trust. Take the L dude, sit back and relax. You will get to your restaurant or whatever inane activity you are doing for the day/evening.
Man I love Waymo everytime I'm in SF. Truly feel like I'm living in the future when I sit in one
Biggest thing I'm excited for is knowing what the cost will be ahead of time
Which Uber used to provide... Until they were infected with tipping. Hell, I will gladly pay more than I would've spent on a tip (20%) just to avoid the hassle.
I’m curious if autonomous cars will become targets for aggressive drivers. Like a driver isn’t going to be as scared cutting off a Waymo or tailgating one because the AI isn’t gonna get road rage or honk like hell. In some places I could see the Waymo’s getting severely bullied if that’s the case.
But why would you tailgate a driverless car? I think usually people tailgate to intimidate the person ahead of them to drive faster.
People tailgate because they're toddlers and locate their locus of control externally - if anything, they'll be very happy tailgating driverless cars because they can throw as big a fit as they want, there will be no consequences, and they'll feel they got to blame something else other than themselves.
Because there’s still someone in the car, they just have no way to defend themselves. You can tailgate and honk at them to your heart’s content. Well at least until they call the police but that’s pretty far. And there are other forms of aggression that do accomplish something. If you cut a Waymo off or beat it to merge, you get ahead of it. In some locales I could see a whole series of cars merging ahead of a Waymo if people are aggressive enough.
Because that works every time?
Or maybe the agressive drivers get a kick out of inciting a reaction, which they won't get from a robot
If they don't get any feedback, they might not get anything from it anymore.
You are on to something here. I have started driving a small car in the UK (Ford Fiesta) and have discovered it's a magnet for the road rage people (around 50% of drivers here).
Firstly, I never back down and will come to a complete stop if slowing down doesn't work. Secondly, I have noticed these drivers feed off any reaction and that avoiding eye contact works very effectively, even if they pull beside you to have a childish rant.
"Don't engage" are words to live by on the roads
They're already learning how to handle this in SF. (I don't live there anymore, so I can't give specific examples.)
Waymo markets itself as an automated driver - same reason they're using off-the-shelf cars and not the cartoony concepts they originally showed. Like real drivers, they take the law as guidelines more than rules.
De jure (what the law says) and de facto (what a cop enforces) rules have had a gap between them for decades. It's built into the system - police judgement is supposed to be an exhaust valve. As a civil libertarian, it's maddening in both directions:
- It's not just that we have a system where it's expected that everyone goes 15mph faster than posted, because it gives police an avenue to harass anyone simply for behaving as expected, and
- It's also dystopian to see police judgement be replaced with automated enforcement. There are whole classes of things that shouldn't be penalized that are technically illegal, and we've historically relied on police to be reasonable about what they enforce. Is it anybody's business if you're speeding where there's nobody to harm? Maybe encoding "judgment" into rules will be more fair in the long run, but it is also coaching new generations to expect there to be more rules and more enforcement. Feels like a ratchet where things that weren't meant to be penalized are becoming so over time, as more rules beget more automated, pedantic enforcement.
A slight digression, and clearly one I have a lot of thoughts on.
It's really interesting to see how automation is handling the other side of this - how you build machines to follow laws that aren't enforced at face value. They can't program them to behave like actual robots - going 24 mph, stopping exactly 12" before the stop line, waiting until there are no pedestrians anywhere before moving. Humans won't know how to interact with them (cause they're missing all the nonverbal communication that happens on the road), and those who understand their limits will take advantage of them in the ways you've stated.
So Waymo is programming a driver, trying to encode the behaviors and nonverbal communication that a human learns by participating in the road system. That means they have to program robots that go a bit over the speed limit, creep into the intersection before the turn is all the way clear, defend against being cut off, etc. In other words, they're building machines that follow the de facto rules of the road, which mean they may need to be ready to break the de jure laws like everyone else does.
TBF the Zeekr minivans are a big step toward a purpose-built Waymo vehicle. I do agree that Zoox has its priorities backwards by going straight for a purpose built robotaxi vehicle. But it has advantages like friendlier ergonomics for the disabled.
Who cares? You are focusing on unimportant issues.
Movement in the USA is heavily outdated. Whether it’s "automated" won’t change anything other than encourage more cars on the road. Great your 5AM commute from the boondocks still takes 2-3 hours but at least you don’t have to put your hands on the wheel!
I haven’t lived in NYC, but I have lived in Boston. Isn’t the real concern winter? Has Waymo (or any other self driving tech company) shown that it can handle the snow well: non-visible lanes, downshifting to avoid braking, etc.?
Definitely interested in how this turns out.
downshifting? these are all electric vehicles IIUC
They tested in Buffalo last year and have in tested Michigan.
https://www.wkbw.com/news/local-news/inside-the-self-driving...
Didn’t know that. Good for them!
Even if they never actually solve winter driving, they could just… not drive during the winter?
If there’s a high probability of below freezing temperatures, cars can just make their way out of the city to some parking lot to hunker down.
Or move them elsewhere in the country during the winter months.
Having a seasonal service is not a bad idea. The big problem with that is cutoff times. Too early and people will complain when they can't get a ride when no snow is on the ground. Too late and you're liable for everything that happens when the road is covered in thin ice or sleet, including leaving someone stranded. You will need very accurate weather predictions for operating over the winter months.
GM and Ford do quite a lot of self-driving testing in Michigan.
I exclusively use Waymo in SF, even if it costs a bit more than Uber. You'll most often get a great human Uber driver, but there's a very real possibility that the person is a bad/unsafe driver or the ride is unpleasant for a myriad of other reasons. With a Waymo, you know exactly what you're buying.
Is this the first time Waymo is doing winter / snow testing at scale?
I think some of the Pittsburgh-based self-driving firms may have tried this, but unaware how far they got.
We'll see what happens when there is snow in the forecast. They might just call them all back for the storm.
Yea, that's what I figured, but I also wonder how well anyone is driving in the slush and if the LIDAR / cameras are that disrupted by snow / ice / salt.
Waymo has done some winter testing in Buffalo NY.
It's insane that they need permits for 8 cars that have humans driving them in 2025, when they're already fully automated in SF.
> We’re a tech-friendly administration
Clearly not.
I think caution is good here. We all saw what reckless admin + Uber did before they shut it down for good.
The permit gets them into the process for eventually deploying without safety drivers. That includes safety plans, emergency responder plans and training, and periodic reporting.
They could just drive cars around like Tesla, but that wouldn't put them on a path to a fully autonomous service.
> already fully automated in SF
I don't thin it's fair to say they are fully automated. There's a large remote operations team for remote assistance to help them get out of tricky situations. The cars can be nudged to perform certain actions.
New York is a long way behind California in regulating autonomous vehicles. Fully driverless vehicles are also illegal there and it will require legislation for Waymo to deploy in the state.
It's not insane for cities to permit autonomous vehicle technology. They permit almost every other type of heavy machinery. Even manually driven cars are permitted! (Driver's license test, registration fees, etc.)
You sound like a junior admin. "Why do we need to keep testing? It works in the SF office?"
Because they are completely different environments.
I didn't say testing was stupid. I said permitting only 8 vehicles for human testing from the leader in self-driving cars, years after they've been fully autonomous in other dense cities, isn't the flex he thinks it is.
There's gonna be people driving them? What's the point then?
Sounds like it's a person actively monitoring but not driving. The point is minimizing risk until safety can be proven.
Lots of comments sharing their waymo experience, so I'll hop on the bandwagon :) I visited Austin for a work trip and went out of my way to get waymo rides for work events, reimbursed of course ;) , managed to score 3 rides.
The airport is out the coverage map so I had a real person behind the wheel both ways. Objectively, the waymo was way safer experience because one driver was a local and drove like one (e.g. rolled through stop signs, drove past a long queue to merge at the end, etc.) and the other smelt like weed in the car. Luckily, both trips we arrived unharmed. In comparison, the Waymo drove pretty well, imo and very consistent. Nothing extra ordinary but no reason to stress.
The difficult part of riding the waymo was all moral cope: it cost just as much (minus tip) as paying a real person, driving past homeless people under a bridge in an autonomous vehicle felt unsettling, and my driver from the airport in my home city was wonderful and hard working. I don't typically like to chat in the cab, and the driver didn't initiate, but I was feeling empathetic and guilty so I struck up a conversation. By the time I got home we were enjoying ourselves and the driver was sharing animal facts because I had learned he was a real enthusiast that could not make a living solely on ecology. We were laughing and joking around together. (Google, if you're reading do NOT try to replicate this experience with AI)
I'm glad I got to try it and out of my system. Still would prefer trains or more public transit over more cars :p
As a regular bike rider, I'd rather take the lane in front of Waymo vehicle than a human driver.
Saw my first Waymo car yesterday in Manhattan (SoHo) and wasn't sure if it was finally happening! Super excited!
FYI if you live near a Waymo charging and cleaning station, it will be constant BEEP BEEP BEEP BEEP BEEP and shopvacs running all the time.
Gemini has an API for gesture recognition. It might come in handy.
if these take over, people are just going to walk in the street in front of traffic (even more) since they know they'll stop lol
Once again, if you're going to test your self driving car on public streets, all data should be open and public. Car companies shouldn't be competing on how to prevent life-or-death incidents, they should be cooperating.
if it can do WDC or NYC it can drive anywhere
I saw one of these on Chambers Street just yesterday afternoon, but it must have been in manual mode, of course.
Downtown Manhattan is the hardest-to-navigate area of NYC. I thought they would start somewhere in Midtown, where the grid is regular, streets wide, and demand for taxis still pretty high.
We already know that Waymo can handle regular American cities quite well. I woul expect them to spend most of their expensive human-supervised training and testing budget in the most unique locations, like downtown Manhattan.
Midtown on a busy Saturday or Sunday afternoon with a driverless vehicle would be... amusing. No one-including busses-give af about traffic rules.
I saw them in downtown Jersey city right across the Holland tunnel. I guess that's where they are parked when the human operator is off duty.
I would expect an "automatic, but human ready to intervene" mode for development and testing.
The Waymo’s I see in Austin just circle the same path to pump the number of miles their fleet is training on. Unless they are on an actual ride.
Is Chambers St busy during the afternoons?
NYC is so dense it could be a bikers' paradise in the US. Why are we supporting even more car infrastructure :/
Because half the year it's freezing cold or blistering hot?
With CitiBike and so many bike lanes it basically already is a biker's paradise. Obviously there could be lots more improvements, but the people who want to bike already do.
Bike infra is forever stuck in the limbo of building half-assed solutions and then complaining not enough ridership is taking it up.
I hope Waymo can get their cars to stop blocking crosswalks and running red lights.
The game-theoretic aspect of this is interesting to me. A lawful robot will never make progress in Manhattan because the people will just walk across its path continuously, forever. To be an effective driver in Manhattan you have to intimate that you're willing to hit people, without ever hitting them. If humans believe that the Waymo will categorically never hit them, then the Waymo will never get a turn.
its interesting. at beginning in SF the waymos would just stop cold anytime they saw a person or a bicyclist. now they're acting a lot more like a person. if I'm in the crosswalk they've started playing chicken just like a normal driver would, starting to go into the turn while watching to see if you're going to stop and give them the right of way. if you keep going, they will stop.
There are still plenty of humans in SF who are on to the nature of the game. A few of the shambling lunatics who inhabit the vicinity of 6th and Jessie know that they can just harass a Waymo and it will stand there forever.
Very cool. I wonder what scale it has to hit for this to become a profitable line item for Google and what their revenue targets are for it.
I think the problem in NYC will be getting medallions, assuming that's what self driving cars will need.
There are already so many (too many?) taxis and car sharing drivers, after TLC's massive increases of the last few years. You can play a game, based on something I read about last year: stand at a corner and count all cars/trucks/for-hire. The first two combined are barely outnumbered by the last group. And the few times I checked, half of taxis and car sharing vehicles were empty. (Of course that's different at peak times or when it rains.)
Will Waymo be allowed to add as many vehicles as they want, like a new class of cars, or will they need to buy out medallions from drivers? The former might undo all the progress in traffic relief that was brought by congestion pricing.
I'm cautiously optimistic about this self-driving thing. Waymo at least seems to have figured out a lot of it.
Would it be way better to make walkable neighborhoods, mixed-use developments, and reliable and frequent public transit?
Yes. Yes it would.
But, in lieu of that, self-driving has a lot of advantages in the long run, even if the technology isn't 100% perfect right now.
I think taxis have a place even in cities with great transportation. I live in Toronto and 90% of my commuting is walking, 8% public transit and 2% driving. But there are some trips that would be very difficult to do a way other than driving (for example carrying lots of stuff or awkward cargo) and taxis fill that gap wonderfully. Especially if self-driving taxis could handle long trips a lot better as inter-city is a place where Toronto public transit unfortunately sucks (for example visiting my parents in cottage country).
In most respects, Waymo is the modern version of reliable and frequent public transit, with a lot of additional benefits.
The modern version of reliable and frequent public transit is reliable and frequent public transit.
Not in terms of throughput though. Buses and trains still have em beat.
This is true for intersection throughput, but I bet full travel throughput (walking to the bus, waiting for the bus/train, walking to your destination) is the same or better with Waymo.
> Would it be way better to make walkable neighborhoods, mixed-use developments, and reliable and frequent public transit?
This x10000
I would wager Waymo fails to integrate in NYC. Austin, and SF are child’s play.
How much? What time horizon?
No chance Waymo can operate in many parts of NYC. Good luck getting through double parked cars in Astoria and elsewhere.
There's a great YC saying which approximately says "you should get the easiest customers first". They even made a video about it, saying tech startups sometimes try to go for the hardest customers to "prove themselves" and it just hurts their business.
I sort of wonder if that's happening here. SF, Austin, LA, etc, are all great cities to build autonomous vehicle startups in. There are many more major cities which don't get snow, have minimal rain, and are well thought out in terms of driving layout. NYC seems like the most difficult city to operate in, and while I believe it's a lucrative market it seems like a mistake.
Maybe not immediately but gathering the data on those areas will eventually lead to their ability to drive there.
That type of tolerance in moving vehicles will take at least a decade.
I can't wait to hear how it goes in NYC -- its going to be a total cluster - with the significantly more chaotic behavior on the streets, bike scooters, pedestrians and then the oddness of the streets/aggressive driving necessary behavior.
Give it one month if they saturate it too much there will be political blowback on waymos causing traffic chaos. Queue track record in SF as datapoints.
I don’t think people commenting and downvoting us realize how things are in NYC. Not only do you have to deal with insane chaos you also have to deal with malicious drivers. Hit-and-run in NYC is shockingly high because it’s a no-fault state. People don’t stop after accidents. It’s gotten really bad since the pandemic.
I think some people pushing for driverless cars everywhere are assuming it will necessitate much stricter driving laws and penalties for human drivers to make their driving compatible with the robots. And they're fine with that, but they know it's not a selling point, so they don't talk about it.