Driverless robotaxi fleet paralyzed for hours in San Francisco | Hacker News Reader

Driverless robotaxi fleet paralyzed for hours in San Francisco

thelastdriverlicenseholder.com

504 pointsgeorgecmu496 comments

I am just wondering how many people have to die before it finally dawns on on us that the current approach is idiotic.

The supposed smart cars can't even talk to each-other directly 'without internet' to warn of a car pileup! They can't report to city traffic control about their condition.

We should be innovating in infrastructure instead - create standardised computer readable infrared road markings, equip each traffic light and each lamp post with a radio beacon, each crash barrier could have a radio marker, create PUBLIC maps of each city, have a central traffic control sypercomputer in each city provide directions to cars. Have each car painted with infrared markers so they recognise each-other. Provide cyclists with something these cars can recognise.

We could even make radar-reflective pants so that autonomous cars see them better.

the whoe traffic system needs to be looked at and brought to a new set of standards, whatever they may be. I am not sure what they are, but it should be clear to anyome with half a brain thay having a car use a camera to tell if the traffic light is red or green is idiotic.

untill a new system is ready, no car without AGI level ai will ever be safe

The only problem is that such collective approach conflicts with the way VCs work.

Even without knowing when AGI will get here, I know it will be faster than solving the collective action problem.

I think eventually major car manufacturers will work with the regulator to bring on new standards, and it will happen withing 15 years. Infrastructure upgrades will take another 15. Most startups that aim to solve self driving will either find a niche in that ecosystem or go extinct.

So I expect reasonable self driving cars to be common in depeloped nations in 2050, while AGI will be just as elusive.

In a similar vein, the correct solution to electric trucks is not giant batteries, it's simply overhead wires:

https://www.cnbc.com/2021/07/29/uk-to-study-using-overhead-w...

We are not that far away from implementing negative interest on cash.

> I am just wondering how many people have to die before it finally dawns on on us that the current approach is idiotic.

I was going to say about 30,000 per year in the US, but it’s actually up above 40,000 per year now.

[1] - https://www.nhtsa.gov/press-releases/early-estimate-2021-tra...

40k people are dying a year from cars, and the problem is that driverless electric cars are not the solution. They are the faster horse kind of innovation. If we were serious about reducing fatalities, improving health, and reducing pollution, the solution would be to drastically reduce the amount of miles driven per year and the number and the size of the cars. We need to encourage people to bike and walk and use public transportation. Autonomous trains would work extremely well and we don't have to worry about bicyclists. The zoning laws and car centric development of cities is the issue.

None or few of those options are particularly accessible (EDIT: for people who cannot walk/cycle long distances). Public transportation often is quite inaccessible while still claiming to be accessible (like the London Underground and the New York Subway). Central management of driverless cars could solve the problem in a more inclusive fashion and isn't just more of the same idea. We should still reduce the size of the vehicles, though.

EDIT clarification: Accessibility for wheelchair users, blind people, and anyone with ambulatory restrictions. Most stations are not wheelchair accessible to the train and having to go on much longer routes to compensate is not equal access. And many 'encouraging other use' initiatives block all cars, leaving these people in the lurch. Something that gave door-to-door access to everyone, equally, via a fleet of cars, could be superior.

If we could reduce car infrastructure to the point where only people who need a car for the transportation to be accessible would use it that would be a dream.

I don’t get the accessibility angle at all. It’s a non-issue because you will have a very hard time finding people who have issues with exceptions for accessibility. That would still reduce the need for car infrastructure massively.

The thing is, we need to start building the future right now, right here and massively. That means public transportation centric development (= building public transport to nowhere and letting the area develop alongside the walksheds with appropriate mixed use zoning) and actually being brave enough to re-structure existing solutions along those lines. That all isn’t magic, we just need to do it. We know it works, it doesn’t require any technological breakthroughs.

Does it? For a time, perhaps. China's high speed rail network, built largely in the way you describe, is nearly a trillion dollars in debt.

https://asia.nikkei.com/Spotlight/Caixin/China-looks-to-slow...

>> None or few of those options are particularly accessible.

At the moment, neither are self-driving cars sufficiently capable to replace human drivers.

We're chasing a future tech dream when we already have technology that can solve the problems we are facing right now.

Edit: What do you mean by "accessible"? The London underground and generally public transport in the UK works very well. Speaking after having lived there for 15 years, studying and working in London and a smaller town, and never having owed a car.

+1 on London underground being very accessible. Although I only lived there for a few months, it was insanely easy to get around relative to American cities I've lived in (San Diego, SF, DC) although I think Berlin's combination of train and tram was even better

> the London Underground

Just going with the parent comment and I'm asking when did you last use the Underground and what station do you think has accessibility issues? I found that the London system is pretty much accessible with aggressive retrofitting on older stations, but I'll acknowledge that if you have used it say more than 10 years ago then I understand.

I think you're replying to the child of my comment?

Oxford Circus you still have to be able to walk up/down stairs.

I've used it plenty and still use it. It is getting better in places, but just look at the tube map - large parts of the system still lack step-free access, let alone the problems the system has in other ways (overcrowding and people not giving up their seats or making space appropriately); we should be trying for (in-city) transport systems that don't require such congregation.

The last time I travelled the tube with a wheelchair user with a pain condition, the (non-strike) closures hit in such a way that we couldn't reach our destination without going on a detour that was so long (more than an hour) that the system assumed we'd forgotten to swipe out and back in and charged a penalty fare. The people manning the help-intercoms were unable to assist in finding a route that didn't have closures. And it's normal to have to find alternative, slower routes than the ones everyone else can take.

I can tell you've never tried to use the Underground with a wheelchair, tho if you're questioning what accessible means in that context—because it's really bad. cars are very accessible for those with mobility disabilities & for many others.

Ah, my bad - literally "accessible".

I don't use a wheelchair and I don't know about the London underground but I can tell you that public transport in the small town I lived for 15 years were eminently accessible, as evidenced by the continuous presense of passengers with obvious (and varied) mobility issues on them.

In any case improving accessibility on public transport also does not require any new technology that we don't know how to create. It just requires political will.

This is a common refrain, that things must be accessible because you see people with mobility issues using them. But it's not true - the people you see are the people who are coping with it, either all the time or on that day. You don't see all the people for whom it is not accessible. It's well-known in disabled circles that most public transport is hit and miss at best and exclusionary/ableist/injuring at worst.

Actually solving the problem does involve new technology, because the worst cases need door-to-door access and most cannot afford taxis (nor do welfare systems provide them).

What makes it all harder though is the continual fight to persuade every last person that it is a significant problem, because they're so sure it can't be.

Surely it would be easier to make public transportation more accessible in a non-car oriented city than making driverless cars, which then have large amounts of continuous infrastructure costs, health care costs from less exercise, social costs, and also the fact its still not known if this is even entirely possible.

As a thought experiment, we have no cars other than emergency vehicles and some commercial vehicles occasionally. Everyone is able to walk to 90% of destinations, and bike or public transport the last 10%. It would be easier to have people who can't walk well in small electric personal transport devices (wheel chairs, segway, electric bikes, motorized scooters seen at walmart) that if they crash do not pose any threat to anyone. Surely adding wheel chair accessible elevators to train stops for public transit, is easier than building out entire car centris highway infrastructure that requires continuous expensive maintenance. Public transit and walking is already more accessible to the blind than automobiles, that is a nil point.

A friend of mine is a quadruple amputee. He can't drive a car, but he can use bike lanes and public transit just fine.

> The zoning laws and car centric development of cities is the issue.

Well, the main issue is that most of the cars on the road during the busiest hours have just 1 person in them.

I think when the dust finally settles, the main form of passenger transit in urban-ish areas will basically be something that looks like autonomous vans.

Unlike public transit today, they won't have predefined timetables, nor predefined startpoints and endpoints. They'll pick you up wherever you are, whenever you want, and drop you off wherever you want, but they might take a roundabout way to get there to pick up and drop off other passengers too.

If they usually have 2+ people in them at a time, we'll be able to use our existing infrastructure (roads) at a much higher throughout.

so basically a jitney?

How is this an improvement over trains, trams, and light rail?

I think reusing car-based infrastructure with busses is an important step in transitioning away from car dependency, but roads are not permanent. In many parts of America, trams and railways were torn down to be replaced by roads. The roads can be replaced too, if it is clear that alternatives are better.

Plus, roads are destroyed incredibly quickly by cars and busses, so they will need replacing eventually, whereas rail is much more resilient.

It seems to me that rail-based transport is much more effective in dense urban environments.

Point to point transit at the time when you need it.

If I want to get from A to B at 14:00 and can do that in 30 minutes by car, but the only bus that goes there requires me to change at C, and it takes 20 minutes to get from A to C, 20 minutes to get from C to B, a 10 minute wait time at C, and the connection happens every 30 minutes and isn't perfectly aligned with my schedule, I'm now facing the choice between 30 min by car or over one hour (including the wait time at my destination because I had to take an earlier connection) with public transit. Add a 5-10 minute walk at each end to account for the distance between the nearest stop and the start/destination, and public transit can take 3x as long.

If there is a van service that gets me there in 40 minutes at a time of my choice, that's a much more acceptable option.

This may also consist of a car first taking me to a tram/bus stop a few minutes away that gets me more directly to the destination.

The van service also solves the problem of those few-percent edge cases where regular public transit just doesn't work, which force people to get a car, which then gets used even for trips where public transit would otherwise be acceptable (but not great).

This is already possible today. You don't need driverless cars to do it. Uber can already do it, but it's not taking over. Unless you're suggesting we will have more cars just sitting around waiting for rides, because they're driverless. Then we get back to them being so convenient that nobody will want to ride with another person.

In other words, why does making the car driverless suddenly make everybody want to use it?

IIRC something like 70% of the cost of Ubers is driver compensation, meaning driverless Ubers have the potential to be ~3x cheaper.

So they if get rid of the drivers they contract who is buying and paying for the cars and maintenance? - are you suggesting Uber start carjacking their existing drivers?

Do you know what percentage of driver compensation effectively goes to car ownership and maintenance?

Or 1m+ people if you count non Americans as people. I'm optimistic that self driving tech will be able to be repurposed to prevent accidents even if fitted to non electric non self driving vehicles. The switching to bikes etc. is a nice idea but probably won't happen in a hurry.

>>The supposed smart cars can't even talk to each-other directly 'without internet' to warn of a car pileup! They can't report to city traffic control about their condition.

- This by far in my opinion is the single most important attribute of any autonomous fleet of vehicles that is needed. There should be first of all a standard of communication via RFC defined for autonomous vehicles and drones. Any vendor manufacturing autonomous vehicles or drones should follow this standard of communication and be able to communicate to each other with the primary objective of avoiding collisions. This should be a RFC so that it works across vendors and is as standardized as TCP/IP. Communication medium can be over radio frequencies as well as the internet and can work as backup of each other but first there should be a established standard that all vendors adheres to.

I think this is a bad idea. The potential for a malicious/incompetent/malfunctioning actor to wreck havoc becomes enormous.

Vehicles should assess and plan driving from first principles and physical sensory inputs. Networked inputs could be helpful as augmentation but it's not something one should rely on at all. Kind of like you shouldn't take a human showing/not showing a turn for a reliable signal of intent.

Yes this is augmentation only but it is essential to have. This is what replaces a driver to driver communication part. Like two drivers at a four way crossing sometimes signaling each other.

BTW being opinionated is fine but no need to downvote

Say what? I didn't downvote you.

Gonna recommend Car Wars by Cory Doctorow - https://web.archive.org/web/20170112064813/http://this.deaki...

You're describing exactly the kind of autonomous car development that was going on for the last 40 years. It went absolutely nowhere. It wasn't until tech got good enough and developers started treating vehicles as independent agents that any real progress has been made.

I totally agree. We should start with the infrastructure.

Sounds like a lot of infra and related spend via taxes or muni debt for a tech outcome that’s a privacy and autonomy nightmare.

Build some trains, build some bike lanes, and leave the people alone who don’t want their transportation and the downstream reasons people use it taken over by tech.

Governance by a central cybernetic super computer and all the cybernetic hypothesis coming true. Good lord.

>Sounds like a lot of infra and related spend via taxes or muni debt for a tech outcome that’s a privacy and autonomy nightmare.

That unfortunately puts it right in line with every other tech-heavy thing taxpayers have have to pony up for this century.

Indeed. I raise issue with the unthinking advocacy here, as if civil society is just a canvas for engineers to paint over. It’s the only thing I hate about this industry. It’s UX-oriented as long as tech is part of the UX.

I may sound like a Luddite but I recall being able to go out in NYC without needing to QR-code scan my way through daily life.

I think you hit the nail on the head with trains, because your parent is pretty much describing a modern train signaling system, down to the centralized supercomputer.

> create standardised computer readable infrared road markings, equip each traffic light and each lamp post with a radio beacon, each crash barrier could have a radio marker, create PUBLIC maps of each city, have a central traffic control sypercomputer in each city provide directions to cars. Have each car painted with infrared markers so they recognise each-other.

It is amazing how often tech savy folks can reinvent trains.

This seems to be the dominant opinion is here, that is, to approach this problem as we approached civil aviation and flying commercial planes.

To respond in a sentence: first, you have no proof that "investing in infrastructure" would actually yield working autonomous cars. Second, there are no proof either that the current approaches are doomed to fail.

To respond in details: everything that you propose doesn't seem to be able to solve the main challenges that self-driving cars are facing. These challenges are all about edge-cases, and responding safely to changing conditions and misbehaviors from other agents. The response cannot be "let's add a ton of fragile electronics in all of the infrastructure" because this: will cost a ton of money to build and maintain, will break in weird ways over time and lead to cascading failures, and most importantly: it's dubious that it would even work at all.

The final comment about VCs is weird: a lot of actors in this field are not VC backed startups.

My opinion is: Tesla might have a working model that outperform human drivers in 10 years (50% confidence)

> These challenges are all about edge-cases, and responding safely to changing conditions and misbehaviors from other agents... The response cannot be "let's add a ton of fragile electronics"

Self driving car has failed to detect a person in the dark and killed her. It was demonstrated that Tesla autopilot fails to detect a child crossing the road. It confused a semi trailer and an underpass. You call that an edge case?

> The response cannot be "let's add a ton of fragile electronics in all of the infrastructure"

So instead we will add a ton of fragile electronics in every car. Instead of having 3 optimally positioned lidars monitoring an intersection, we will have 20 poorly positioned lidars on every car.

You don't need electronis, you can use paint. You can put QR codes that tell a car where it is in an underground parking lot. If we add a parrern to all clothing in infrared-only paint, that would help various AI detect people and tell them apart from surroundings. You can put a metal patterns on street signs that are detected by automotive radar.

> will cost a ton of money to build and maintain

And electronics in the cars will cost even more

> will break in weird ways over time and lead to cascading failures

You mean like the failure the main article is discussing right now?

> it's dubious that it would even work at all

And it's dubious this 'individual' approach will ever work. Every single argument you've made equally applies to the approach you are advocating.

Seems to me you’ve just reinvented a city rail transportation system with modern signalling. Modern rail system can indeed drive fully autonomously if they are fully grade separated from street traffic.

That's a lot of effort, money, and infrastructure that could go into developing a healthy public transportation system instead of trying to stop all the cars from killing people.

There is currently some kind of geomagnetic storm which is causing a lot of GPS tracks to be offset over the past few days. It's all over the running/cycling forums. Could be related, here's an example:

https://old.reddit.com/r/Garmin/comments/vo17e2/the_magnetic...

It is worth noting however that the size, power consumption and antenna precision of a watch-sized or bicycle computer size GPS(+glonass, Galileo, beidou, qzss) receiver is considerably smaller than what should be in these vehicles.

I have not seen any mention of GPS service degradation in hobbyist drone enthusiast groups over the last week, some of whom use quite small GPS modules with serial connection to the flight controller for automated signal lost return to home features and such.

I would find it odd if they were not using precise positioning using local reference stations broadcasting corrections over IP or RF.

EDIT: Appears GPS/IMU/LIDAR is sufficient, my mistake. TIL.

They're almost certainly using dead reckoning with wheel rotation sensors to get the precision they need in an urban environment where radios are kinda trash.

I would think zero wheel rotation sensors and just LIDAR, which cruise uses extensively.

Tesla’s also use dead reckoning when GPS is lost. It’s way more accurate than I would have guessed. My GPS antenna was out for over a week and the map was only about 100ft off.

NOAA doesn't list any meaningful geomagnetic storms right now:

https://www.swpc.noaa.gov/communities/space-weather-enthusia...

But that aside, I'd really hope that these cars are using professional WAAS-capable GPS receivers. Short of a major X-class flare which knocks out the WAAS geostationary satellites, the entire point of the system is to correct for these kinds of disturbances to the GPS signals.

Why does this make me wonder if something is going on with GPS due to the war in Ukraine?

EDIT: the following is incorrect per a reply below. Will leave the original comment below.

---

That was going to be my hypothesis. GPS resolution can be reduced in a conflict area.

I'll defer to someone with actual knowledge to add more info. :)

I believe the government promised not to ever reduce GPS resolution some 15 years ago, and new satellites supposedly do not even have the ability to do so.

TIL, apologies for the misinformation! Will correct my comment if it's still in the edit window

Oh my sweet summer child.

I am old enough to recall when they did so post-9/11.

The US government would never give up such a capability, that would be out of character for them. But they would sure as hell pretend like they did, which is very much in character.

I think they would. There's at least 3 fully operational GPS alternatives, there's no reason to have this capability anymore. Some receivers will combine all of these systems.

All these systems are rigged to jam and spoof at the push of a button I expect.

They are owned by: China, Russia, India, Europe and Japan.

There was a specific technical capability that they used to reserve the right to use, and I believe they no longer need to use that capability. Obviously this would only be the case if they have some alternative capability. But I forget the name of the feature so I am having trouble looking up sources. I obviously do not take the government at its word, but I can believe that a technical capability developed in the 1980's might no longer be the most useful counter measure. Especially since there are now 5 GNSS constellation systems, with GPS only being one of them, and low cost off the shelf receivers can read all of them.

Years ago a government promise and .25 cents could get you a phone call. Today you won't even get the phone call.

Reduction of GPS Resolution is limited to specific areas, it can be targeted. If they are doing it (probably are in Ukraine) then it only affects some areas. You can also bypass that restriction if you have the right military keys (read: Ukraine).

Yeah, that is what my (perhaps outdated?) understanding was.

What is said is the current capabilities. All other GNS systems have the same capabilities (though European GNS Galileo has some more modern features and the Soviet one is a bit of a shitshow)

This submission is worrying me a bit, my company's product depends on GPS+GLONASS with RTK assistance at high latitudes for high accuracy high frequency positioning.

The NTRIP RTK services we use just provides assistance for GPS+GLONASS, no Galileo or BeiDou. GLONASS is crucial because the satellite orbits are more visible at high latitudes providing better reception. Still, RTK should help cope with this problem but I might have to check last week's collections.

And your systems actually support decryption and entering the said keys.

You might be surprised what proportion of the hardware that operates in Ukraine (from both sides) is using civillian grade GPS receivers (i.e. almost anything with a price tag of less than 1M USD).

From the linked reddit thread:

>Does this happen when you're not using Glonass? I wonder if Russia has been 'tweeking' it.

>I've switched to Galileo and the problem is FIXED.

WAAS has been available on consumer GPS devices since the mid 2000's and is not "professional" nor does it address reception/interference issus; WAAS provides semi-real-time position refinement data based off drift observed by ground stations within a certain range to compensate for things like atmospheric disruption of the signal. WAAS is built-in to the vast majority of receivers these days.

Multi-band GNSS is what helps address issues around reception like multipath and interference. Receivers used to be pretty expensive, and the number of L5 satellites wasn't that high until the last few years, but now can be found in even a number of consumer electronic devices, and the number of L5 satellites keeps going up (and as more satellites have L5 signals, things will get better.) Several Garmin sport devices have it now, for example. A quick search shows Maxim sells a multi-band GNSS chipset for $10.

There is really no excuse for any autonomous vehicle company to not be at least evaluating multi-band GPS receivers, if only to give more reliable position data in urban environs.

I'd be surprised if at least some aren't using RTK receivers, but this requires a base station within a certain range of the cars, or commercial service (such as that offered to farmers by John Deere.) RTK just isn't that expensive compared to even a single LIDAR unit.

Edit: they can tolerate at least some degree of GPS degradation and use dead reckoning / lidar data in some cases, according to this person: https://news.ycombinator.com/item?id=31940167

I just read about a sunspot that recently doubled in size [0] and maybe launched a CME in our direction -- we don't really know due to a power outage at Stanford University that took the SDO offline [1], anyway food for thought, maybe it's coincidence.

[0] https://scitechdaily.com/auroras-incoming-giant-sunspot-has-...

[1] https://interestingengineering.com/a-huge-solar-eruption

I'm not really seeing anything weird with GNSS. I have a continuously-operating receiver that writes its location into a database every second; I did a query over the last 30 days and there is no abnormal deviation in position. (My receiver is in a terrible location that can only see a tiny sliver of the sky, and is single-band, so I expect about a +/- 50cm deviation without RTK correction and that's what I'm seeing.)

I actually haven't looked at this data for quite a while. I didn't consider the start of the Ukraine war a good reason to start looking at my data, and I delete it after 30 days to save $$$ on block storage. Dumb. (I added some proper downsampling today so I can keep it forever at reasonable cost.)

Trusting my own data is always suspicious, so I clicked around randomly on the NYS Spatial Reference Network website (https://cors.dot.ny.gov/; the title of the page is "Shop", but accounts and RTK access are free). They operate a bunch of Continuously Operating Reference Stations and publish statistics about the accuracy. I do see a lot of reported GLONASS cycle slips today. I have no idea if that is normal or not; this is the first time I've ever dug that deeply into station stats. But overall, the position accuracy at the stations I spot-checked are within .5cm, which is normal. (Most use 4 constellations; GPS, GLONASS, Beidou, Galileo; on 2 bands. So even if GLONASS is broken today, one constellation is enough for good accuracy.)

If the government has degraded GPS accuracy today as many commenters think have happened, many people would have noticed. The reference stations have known positions; if GPS starts saying they're somewhere else, that's impossible and would be flagged. GPS being significantly degraded would probably be front page news; not just noticed by a few exercise tracking enthusiasts on Reddit. Or maybe it's all a big conspiracy and the US and Russia banded together to hack my database and the NYS spatial reference network. It could happen, I guess, but my non expert opinion is that GPS is fine right now.

> my non expert opinion

Yes, clearly given the detail in your post, you are "not an expert". Thanks for being on HN and making this site great.

There is a lot of depth to GNSS! I could do what jrockway did on the grandparent comment and I would not define myself either as anything near an expert on the field.

I think an actual expert would know if those "warnings" are interesting or not. I can find the raw data, but I don't really know what it means ;)

I've been looking at GNSS (mostly GPS, really) interference around the world for a few months. I just generated some maps of the U.S. for the past few days: https://twitter.com/lemonodor/status/1542941996746039297

You can see that there isn't any significant interference visible in the Bay area, or the rest of the United states except for the typical activity in and around military ranges—That huge region in New Mexico experiencing moderate interference is not typical, but seems to correspond with planned testing around White Sands Missile Range, according to an FAA NOTAM[1].

It’s possible that my methodology wouldn’t pick up interference if it’s caused by a low-power jammer or some source of EFI that only has an effect within a few hundred feet.

More details on the methodology behind those maps:

The ADS-B signals that many aircraft broadcast include not just the aircraft's GPS positions, but also a measure of GPS accuracy (strictly speaking, ADS-B doesn't talk about GPS specifically and can handle any sort of navigation technology; I'm sure there are some planes out there reporting positions based on inertial navigation systems, with correspondingly low accuracy, or GLONASS-derived positions, or whatever, but my understanding is that right now something on the order of 99% of aircraft with ADS-B are using good old GPS so I'll just keep using the term GPS in this description). If you go to https://globe.adsbexchange.com and click on just about any aircraft, you'll see an info sidebar on the left of the screen. Scroll down until you see the ACCURACY section, and you'll see values labeled NACp, SIL, NACv, NICbaro, and Rc. Those are all self-reported measures of the accuracy of the data being sent by the aircraft[2]. NACp is "Navigation Accuracy Category for position", and is a good measure of whether the aircraft's GPS is working well. (A somewhat obscure feature of ADS-B Exchange lets you see a map of all aircraft that are currently reporting poor navigation accuracy for their GPS: https://globe.adsbexchange.com/?badgps)

To make the maps, I process a day's worth of data from ADS-B Exchange to find all the aircraft reporting poor navigation accuracy and then I color map hexes (using H3 hexes) according to the proportion of aircraft passing through that hex that reported bad GPS accuracy. Specifically, I'm counting an aircraft as experiencing "interference" if it at some point reported good navigation accuracy and then reported low accuracy. Doing this helps filter out aircraft that just have an ongoing issue with their GPS equipment, or don't even have GPS.

When I do that, areas where there is systematic interference—almost always jamming by military systems—become obvious. There are a few conflict zones (Syria, Cyprus, Israel) that have been experiencing jamming for years, and the U.S. often has smaller scale military testing, especially in the West and Southwest. You can also see the jammers that are apparently setup around Moscow to prevent drones from flying near Oligarch dachas[3].

I started making these maps in February before Russia invaded Ukraine because I thought it might provide an early warning of an invasion. I didn't see that, and in fact this technique doesn't do a very good job of mapping GPS jamming around the actual war zones because civil aviation stopped over Ukraine, so there are zero or few aircraft with ADS-B reporting their GPS accuracy[4]. Without that data, I can't make a map.

Sometimes I do see changes, like when Russia suddenly started jamming around Kaliningrad in March 2022, causing interference in many Baltic states and leading to Finland to cancel some flights[5]. Then a few days later, they just stopped.

[Edit:] I don't think too many people have realized yet what an amazing source of GPS interference data is available using ADS-B! It's like having thousands of sensors roaming the planet, broadcasting GPS accuracy data every few seconds. I sometimes wonder if I would disrupt someone's nascent business model if I started publishing my maps regularly.

(I have a Synology setup with 12 TB that I use to store various sensor data—I highly recommend it.)

1. https://www.faasafety.gov/files/notices/2022/Jun/WSMRNM_22-0... 2. https://twitter.com/lemonodor/status/1493789598077440000 3. https://www.gpsworld.com/jammers-at-dachas-add-to-russias-ab... 4. https://twitter.com/lemonodor/status/1497295859196649475 5. https://twitter.com/lemonodor/status/1500719113185816577

One thing I've found that I didn't expect: I can see hundreds, and even thousands of aircraft suddenly start experiencing severe GPS interference in a specific region, and sometimes there will be coverage in the media or bulletins by government organizations (though usually I see it a day or two before that happens), but sometimes there's just... nothing. E.g. the sudden, intensive jamming in Romania that started at the beginning of June https://twitter.com/lemonodor/status/1535321975639289857

The ADS-B data collection approach is brilliant. This thread is pretty old; you should consider writing a little summary and posting it as an article. Your visualizations on Twitter are also excellent.

Highly doubt they'd be using GPS for anything more than putting them in the right neighborhood.

Edit: Looks like someone ex-Cruise confirmed this ITT

That's not what the comment said. They said the system can tolerate degradation / loss of GPS data for periods of time, in certain circumstances:

https://news.ycombinator.com/item?id=31940167

> Fine localization should be able to resolve without GNSS input (with caveats) and there are specific tested requirements to be able to tolerate a certain number of minutes of outage even in the absence of LIDAR before position accuracy degrades beyond acceptable limits.

People in that thread are reporting a problem with gps + glonass while galileo on the same frequency as gps is fine.

Could it be using a different protocol or CRC system that's more resistant to that type of issue perhaps? There must be some difference, otherwise it would just be more GPS, right?

You're very optimistic

It couldn't possibly be GPS, right? Networks are unreliable, the cars must be able to reach a safe state if the network drops out. They appear to have stopped in the road. If these cars can't safely navigate their way to the side of the road using local sensors, they are wildly, hilariously unsuited for autonomous operation.

> at least a dozen autonomous Chevrolet Bolts from GM Cruise Automation were spotted blocking the intersection at Gough Street and Fulton Street for a couple of hours, according to reddit user seansinha.

There must be something weird about that intersection that is causing a bug.

> If these cars can't safely navigate their way to the side of the road using local sensors, they are wildly, hilariously unsuited for autonomous operation.

You are a lot more confident than me that this isn't the case.

Or I could be pessimistic about GPS -- these things are, albeit in limited numbers, on the road and actually being used by customers, right? I guess I'm surprised GPS is so reliable that this isn't a widespread, constant issue, if this is the normal "I lost GPS" behavior.

It could be a middle ground kind of thing. The cars have to trust something. For position you have GPS + some interpolation based on your movement sensors + some buildings/streets and maps (also unreliable since construction etc changes a lot, a lot of buildings and intersections look the same, etc).

Computers are very far away from understanding where they are from context alone. Hell most humans don't know where they are. So if you have a short GPS outage they will be able to continue navigating due to other sensors, but the longer the GPS outage/issues persist the more unlikely it is that they still know where they are.

There should definitely be a safety to drive to the side of the road and wait though.

Could it be GPS spoofing that is causing this then?

Malice, incompetence, all that jazz...

Still. I would expect some kind of sanity check so that they aren't stopping in the middle of the road.

If local sensors tell you the road is blocked, fine. But if remote data tells you one thing, and local sensors tell you something else, the fail safe should be pulling over to the side of the road.

Obviously there is no failsafe that pulls the car over, otherwise you would see that the cars had pulled over. This might be annoying on city streets but imagine if it were on the highway. An almost guaranteed 50 car pileup. If history is any guide, this will actually happen in the next few years and be followed by a lawsuit and regulation.

That's assuming that there's a space to pull over, which is often not the case on city streets. Especially, when every AV wants to pull over.

Stopping as close to a curb as possible is still better than stopping in the middle of the line. And much better than stopping on an intersection.

Also the assumption that the fail-safe _works_. Any of these fail-safes are much more complex than a classical fail-safe.

I mean, if we were driving together -- you have the steering wheel, I have the map -- and I say "woah I have no idea where we are," what would you do? I guess you would just drive around until I figured it out, or found a safe spot to pull over...

Dead reckoning works surprisingly well at least in Teslas, of which the precision can be seen when they move around in underground facilities and finally resurface: the car pops up pretty close to the entry point.

In any case, aren't all self-driving vehicles equipped with lidars and work in pre-mapped regions? I would 100% expect they are able to just use those maps with lidar to precisely geolocate themselves, probably more precisely than with GPS. And, indeed, GPS itself can at times be inaccurate or simply not available due to obstructions.

So what I'm saying, maybe it's wasn't the GPS.

> aren't all self-driving vehicles equipped with lidars

I believe Tesla has made a lot of noise about not using lidar.[0]

[0]: https://venturebeat.com/2021/07/03/tesla-ai-chief-explains-w...

In this context Tesla is not an example of a self-driving car I think, what was discussed was robo-taxis and I think all of them have lidar

Maybe at some point in the future, but Teslas aren't self-driving vehicles: you always need a driver in them, even (or maybe even: more so) in the case of FSD Beta.

I imagine these robotaxi vehicles would not have stayed on the road had they have had a safety driver in them :).

Yeah, I sort of expected a lizard brain/mammal brain split: features like object avoidance and the ability to pull over don't require the car to know where on Earth it is, and if the car can't identify to side of the road and drive to it safely with local sensors, then it just isn't fit to be on the road. That sort of stuff needs to be operate with an antilock-breaking/"fire the airbags?" level of reliability.

"Where on Earth am I" on the other hand is basically irrelevant to safety. Run it on the entertainment console or whatever. Interpolate position, come up with heuristics, whatever. I mean, it should be like two reasonably competent people on a road-trip -- if you are the driver and I'm the map guy, you aren't going to just stomp on the brakes when I inevitable lose track of where we are (I'm bad at navigating). You'd probably just go along the road while I worked it out, unless I was really confused, in which case you'd presumably pull over somewhere.

Yeah they have Lidars for 10s of 1000s of $ mounted and should be able to go for certain distances without GPS. If a GPS outage was designed as failure mode. But it would perhaps be one of the later items on any engineer‘s priority list since it is so reliable.

A network failure is, like, the world's most predictable problem though.

You've nailed the dirty little secret of the AV industry. Everyone's safety model is underpinned by the ridiculous assumption that a human will be (a) reachable with very high reliability over shitty cellular networks; (b) able to correctly interpret complex situations with extremely limited information; (c) able to resolve safety-critical situations by making planner adjustments, without direct control over steering and acceleration.

Wow. I had no idea that these things had such fundamental limitations. I'd feel a lot less safer using one now, to be honest. If you had to estimate, how long would it be before AVs would be able to do all their calculations locally? Is that something that's even under consideration at the time?

It is probably a chicken-and-egg problem. They need the data from these scenarios, so they can train the local model for later iterations accordingly.

It's a well-known thing in AV circles that some companies were actually faking their route by hardcoding them for demos etc. Also one shortcut is to use HD map data, and relying on 100k+ LIDAR sensor data that will never will be cost-effective. I know of only a handful of companies that rely on 'cheap' sensor data (cameras/radar) and use real-time calculations.

If a full time driver earns 25k-50k that's sort of equivalent to every regular taxi and uber having a $500k / $1M driving package onboard already (at 4%), right?

I don't know the industry but very naively it doesn't seem to me like $100k really breaks the unit economics? Particularly if volume / learning curve can reduce the price?

You are right in the way that the industry seems to be shifting from general-purpose self-driving to transportation/hauling.

> relying on 100k+ LIDAR sensor data that will never will be cost-effective

Eh, the $100k prices are mostly for high-end, low-production-volume research LIDARs.

There are LIDAR manufacturers claiming they can achieve prices of $500 [1] or even $200 [2] if they were fitted to every car made.

[1] https://guidehouseinsights.com/news-and-views/lower-cost-lid... [2] https://www.opsys-tech.com/post/are-we-nearing-a-shakeout-in...

Well, THAT has a Dropout-like HULU miniseries written all over it.

Tesla’s AV design philosophy differs from everyone else’s for just this reason. The idea of using centimeter-accurate models of the environment and having an always-on network connection is fundamentally at odds with handling surprises.

I don’t think a current Tesla would have a good time with no GPS either, but philosophically they stand to better handle intermittent unreliability and adverse conditions once they figure things out.

For some reason I thought these were supposed to be real products, that sounds more like they are still in the prototype/proof-of-concept phase.

This is the difference between Tesla and competitors.

Tesla cars „can” drive you „safely” without connection to the network.

Some say that discard of lidar was a huge mistake by Elon, but he knows very well that this is not the way to go forward.

What does lidar have to do with connection to the network? If anything, once GPS goes down, more sensing modalities can only be helpful.

Car should be able to drive without Lidar and connection to the network.

There is currently only a single car company that accomplished that.

Why without Lidar ? Why not without camera then ? Lidar has nothing to do with Internet connection.

I doubt it has anything to do with any geomagnetic storm, the issue seems to only affect Garmin devices.

That is most definitely not it, these vehicles have localization modules that rely on lidar point cloud matching, gps is generally only used for bootstrapping that process but is technically not even needed.

GPSs can and will be taken offline sometimes (in a region) due to military exercises, tests etc

This happens with a certain frequency in aviation

Relying purely on GPS (for self driving) won't work.

People were responding that switching from Glonass to Galileo fixed it. Could it be sabotage?

> Could it be sabotage?

Given that even Russian pilots in Ukraine have been found to be using GPS, it looks like that would not even be necessary.

https://www.globaldefensecorp.com/2022/06/21/exposed-capture...

Are you telling me all these fleet of vehicles can be defeated with a GPS Jammer?

https://www.jammer-store.com/gj6-all-civil-gps-signal-jammer...

Yup, it's a mess.

What it's not is, a huge multi-car multi-injury accident. So I don't mind it much more than if all the traffic lights in the area were out.

Honestly, a bit of a road diet from the occasional "this intersection is closed due to fleet maintenance" is probably not the worst thing.

But it's not great if now we've moved denial of service attacks into our real-world transportation infrastructure. It's not too different from truckers in Ottawa blocking all the streets and causing chaos. If this can happen by accident, it seems likely that a future version could result from being hacked.

The promises of self driving cars have always seemed weird to me. Sure, I would love to have a world where all cars are self driving. The security benefits seem large.

But decreasing congestion? We already know how to do that, and in many ways it seems like a safer bet than some kind of technogy with large problems and even larger unknowns; public transport.

Cheap, well developed, reliable public transport will beat autonomous cars on throughput and resource use any day.

But it's not nearly as profitable…

Profitable. For whom?

For the municipiality a 6 lane street vs a two lane street with public transport and bicycle lanes can make the difference between neverending debt and prosperity. Places with good public transport and bicycle infrastructure and denser mixed use pay more tax revenue. Spread out urban sprawls and suburbs give the municipiality a lot of street surface to maintain, for very little tax revenue in return.

A good introduction on that part is the strong towns playlist by the youtube channel Not Just Bikes: https://m.youtube.com/watch?v=y_SXXTBypIg

Municipalities are generally managed to dry the funds in favour of the friends of whomever is in power. Usually such considerations come last.

For example in Italy many municipalities removed public transport so that FIAT could sell cars. And USA is at least just as corrupt.

> Municipalities are generally managed to dry the funds in favour of the friends of whomever is in power. Usually such considerations come last.

I think you should strike the word generally here. Even in Italy there are many municipialities with great public transport (grew up not far from the Italian border). In Europe it is generally not rare to have municipialities with nice or even exceptional public infrastructure.

Corruption is always a thing that can exist, but if your place becomes dysfunctional for it, maybe consider your might be missing some of the core ingredients that lead to good democratic decisions (education, will by common folk to participate in regional politics, a working system of law that works against corruption, etc.)

well you see when the municipality is doing well, we don't call it profit, so you can't win the game by definition.

But you have to give up point to point, of course the overall throughput is better but the individual experience is (perceived) worse. This is why I love the idea of "public luxury" make sitting on a train or a bus nicer than sitting in a brand new mercedes.

Easy: If you have shitty public transport you are gonna have a shitty experience in it. If you have a well funded, well planned and embeded public transport system, you will have a good time.

But this costs more than just money — it costs space. And if you are doing it right it also costs you road space (busses and streetcars only make sense of they have their own uninterrupted lanes for the most part).

Eh, most large European cities manage just fine, the car is often the slowest or the least convenient mode.

That depends how you look at it, though. To my mind, the extra walking distance with public transport is generally a plus, since it serves as a decent baseline of physical activity. Even a small amount of added walking can have a significant health benefit. If you add carrying groceries and walking uphill to that the effect quickly adds up. If I had a car or (god forbid) a personal driver I'm willing to bet I'd be significantly heavier and in worse shape, and consequently all the more unhappy in general.

That reminds me of the preference for white rice over more nutritious brown rice just because one is, well, white and the other isn't [1]. Similar with white flour I think.

We need to learn to appreciate things for what they're really worth to us (including their true costs) not just for what they look like.

_________________

Health problems come because of the polishing. Polishing rice into the sparkling white form that most people prefer has three major negative impacts on health.

The first is, polishing removes most of the vitamins and minerals that are vital to health. (...)

The second relates to diabetes, which is threatening to reach epidemic proportions in the Philippines over the next couple of decades. (...)

Finally, polishing rice also reduces its protein content, which can mean the difference between being well-nourished or malnourished.

(...) And, then, over the decades, the dominant culture defined brown rice as “dirty” and fit only for the poor, while white rice was seen as sophisticated and modern.

[1] https://ips-dc.org/why_billions_eat_unhealthy_riceand_should...

My frequent train has that option if you go first class but it still just goes from A to B and is pretty limited if you want to go to C, D or E off the train line. First class is also too expensive for most.

> Cheap, well developed, reliable public transport will beat autonomous cars on throughput and resource use any day.

Remember, this is happening in a state with massive multiple public health crisis that go un-reported and has maps of human excrement in their inner-cities. The priorities here are completely in the wrong direction I agree.

Are you in a state without an underreported health crisis?

Yes, Britain, the state of the NHS is very well known issue

Which public health issue does California have that is less reported than the broken state of the NHS?

You don't even live in the country and you know of the issues, how much harder can you contradict yourself?

> cheap

> well developed

> reliable

Speaking as a guy in Europe where we generally have "good" public transport, I think you can only ever have ONE of those at the same time, with some exceptions.

Cheap fares? You get decrepit busses from the 80s and the timetable's just a suggestion (what my city used to have)

Well developed? Development isn't cheap so fares get expensive, busses are Mercs that run on CNG and the timetable's still just a suggestion - but you get live boards showing arrival times at least (what my city has today)

Reliable? You'll have to go to Switzerland or Japan to experience that, no other place truly has it. And those places aren't cheap... but they would count as having well developed route networks, so 2/3 for them I suppose.

> so 2/3 for them I suppose

In Zürich, a monthly city ticket is around $50, and a monthly country wide public transport ticket is $340 (and really gets you almost everywhere). Annual tickets are slightly cheaper.

I would argue while not "very cheap", the prices are still reasonably cheap, especially when compared to reoccurring car costs.

Switzerland also solved the last mile problem with Mobility, if you really need it

Also, the median salary in Switzerland is high enough such that $50 per month is absolutely affordable.

Vienna, Austria: public transport costs 1 EUR per day without any limits. It is also extremely well developed and reliable. Most people, especially without kids don't even consider getting a car.

Damn that's cheap, it's 1.5 EUR per FARE in Ljubljana and it as reliable as a stoner.

I thought slovenia was the Austria of the Balkans :)

> Cheap fares? You get decrepit busses from the 80s and the timetable's just a suggestion (what my city used to have)

I don’t think this is generally true... If it is true, then the I suspect government subsidize are a bigger factor in explaining the variance of public transit quality then fare price.

Take Reykjavík as an example. It has the most expensive public transit fare price in Europe (86% above the European average) and all you get are some buses every 30-60 min (with only a handful of lines running more frequently during rush hour), timetables from the 80s and a ticketing system that never actually works. It is safe to say that Reykjavík has an expensive, poorly developed and unreliable transit system.

I think a well funded transit system can actually accomplish all three pretty easily, as long as they are willing to sacrifice some convenience for private vehicles.

Is there anything to indicate that autonomous vehicles would even help with congestion? Why would they? If anything more people might take them at the worst hours (because it would be less annoying for them without the stress of dealing with bumper to bumper traffic).

Cheap, reliable, dense public transit with diverse "last mile" options is clearly a much better choice from the congestion, efficiency cost point of views.

I’ve lived in and frequently traveled to cities with what is considered the best public transportation in the world (Paris, NYC, London, HK, Singapore, Zurich, etc etc).

In almost all the above places, there have been frequent instances where I’ve had to use taxis and/or rent a car. And that was while living/staying in the core of these cities.

99.99% of places on earth aren’t nearly as densely populated as the ones above, so increase the need for cars proportionally if you ever want to go elsewhere.

We need both good public transport and autonomous vehicles. It’s not one or the other.

Whenever I see people citing public transport as a panacea for all problems, it’s typically a wealthy, single, childless, digital minimalist 20-something who lives in the core of the aforementioned cities, and never does much outside of it.

You ever wonder where the food you eat gets grown and how it gets delivered to that fast casual restaurant you love down the street? It’s not public transport.

"You ever wonder where the food you eat gets grown and how it gets delivered"

the argument is strange, they don't exactly deliver 40 tons of wheat to the bakery with a taxi. Noone is arguing against cargo and utility vehicles unless they can be substituted by railway (which could be better designed, etc)

https://www.tescoplc.com/updates/2021/how-tesco-is-turning-t...

When public-transport-as-panacea folks create laws to disincentivize vehicles, they typically don't make distinction between cargo/utility and personal travel. Gasoline and vehicle import taxes, and anti-vehicle urban design and laws hit all forms of vehicular travel equally.

But again we're still stuck talking about cities. In any area where food is grown (or literally anywhere on earth outside of the .0001% of land that constitutes megacities), there is no public transport.

Do you think the people who grow your food are living as indentured servants on the land they cultivate, and never need to go anywhere?

Again, public transport is a fantastic solution for dense areas (ie. city cores). But cars are also a fantastic solution for the 99.9999% of places on earth that aren't a city core.

A feel-good PR greenwashing campaign by Tesco doesn't change that.

There's 8 billion of us, all living in different areas with different constraints. We can have both!

Most pedestrian areas I know (and I've lived in some) are also accessible to at least light cargo trucks. I've also seen semi trucks in larger streets, and (large) cargo bikes in narrow ones.

In my experience, shops and businesses tend to get deliveries early in the morning, when the streets are nearly empty. On the contrary, delivery vehicles tend to get exempted from "anti-car" rules, or regulated to some designated time slots.

> When public-transport-as-panacea folks create laws to disincentivize vehicles, they typically don't make distinction between cargo/utility and personal travel.

... Wait, I mean they absolutely do. Pedestrianised streets are usually open to goods vehicles early in the morning for delivery, etc.

>> In any area where food is grown (or literally anywhere on earth outside of the .0001% of land that constitutes megacities), there is no public transport.

Here, I've put you smack in the middle of an area where food is grown and next to a rail station:

https://www.google.com/maps/place/Alalkomenai/@38.4075142,22...

Drag the little man on the pin to admire the beautiful countryside.

Or scroll out to see the fields covering the land.

Btw, there are bus lines also but they don't appear on the map because the locals know where the stops are, and the bus drivers know where the stops are, and what they're called, and who will be waiting at what time, so there's no need for stop signs. Usually the signs were there when the routes were first created but they have er, gone missing since.

Most of the cost of a taxi is the driver. In the long term driverless taxis will be much cheaper. Access to affordable taxis will greatly improve mobility for the young, old, poor, disabled, etc. It will also enable many people who currently drive to ditch their car (saving lots of money) while having comparable mobility.

Of course driverless taxis will not help much with congestion. That will require different solutions.

>But decreasing congestion? We already know how to do that,

Anything, automation included, that allows things to flow more smoothly will decrease congestion for a specific number of vehicles on the road allowing roads/intersections/etc to flow closer to their ideal capacity. (This principal holds whether you're talking about cars or buses, train networks, industrial facilities, etc.)

For example, if self driving solves some decent fraction of the morons who can't seem to merge in the size space that the engineers specifically allocated because it was sufficient for that purpose or can't turn left even through sufficiently sparse cross traffic you've just alleviated a lot of congestion.

In practice transit networks are bounded by the daily peaks so you'll never not have traffic jams and packed subway cars at rush hour but efficiency improvements mean these conditions will exist for a smaller slice of the day.

With all of the talk, usually around Teslas but I suppose it applies to all EVs, that I've seen over the years akin to, "Just imagine how safe driving will be when every vehicle is FSD and can talk to one another," I have to ask... shouldn't we anticipate seeing similar issues to this in that scenario?

I mean yeah, there will be different manufacturers so it's not like every car in a city could come to a standstill. And I'm also speaking from a place of ignorance, so please correct me if I'm off-base. But isn't this kind of a thing (mass stoppages of entire fleets) a cause for concern as we see increased FSD adoption?

> "Just imagine how safe driving will be when every vehicle is FSD and can talk to one another,"

Some genius 10x'er forgets to carry the 1 or something, and all of a sudden an entire city full of cars all collide with pedestrians all at once.

And that will usher in the great Engineering licensing requirements for building high-risk software.

Can't come soon enough IMO

We could have the same fear with all these satellites being shot up into space. Couldn't they all start raining down on earth with some poor calculations or coding mistakes in the course-correction software?

I don't know enough about this stuff to know if this is a realistic thing or just the premise of a bad sci-fi movie.

Satellites would, I assume, burn up on re-enter if they happened to de-orbit. I'm more worried about them banging into one another and surrounding the Earth in an impenetrable debris field that makes future space exploration/travel nearly impossible - IMO that would be a death sentence for all of us.

Maybe Team Seas can pivot to space once they take care of the oceans.

1 - they burn in atmosphere

2 - each satellite is under observation by a team

3 - how many satellites do you think are there actually? Not that many, unlike the driverless cars we are setting up nowadays

4 - Earth is actually big. Satellite pouring down all at once and you'll have maximum like 100 casualties - worse case scenario if 1). doesn't happen. Cars OTOH are in busy cities. One battery to explode and can easily do 100 casualties

> One battery to explode and can easily do 100 casualties

That’s definitely not true. A 66KWh battery is not a tactical nuke

I think autonomous cars are something that must happen, and the sooner the better, but I am also a bit concerned that they could represent the ultimate weapon if a malicious actor gained control over them.

One battery fire isn't a problem, but if every 2nd car in a city or country crashes into something in a way that starts a fire, that becomes a very big problem extremely quickly.

Watch on youtube any video that shows puncturing a cellphone battery what happens. Then scale that up and imagine a scenario where one idiot on drugs drives one in supermarket or mall entrance and the car's battery gets punctured. After that comeback and tell me you still think 100 casualties are hard to happen from only one EV car.

How many? I figured thousands. When I looked it up, there are several thousand. I saw numbers between 4k and 8k. However, the rate of launches seems to be increasing and is over 1k per year now. Musk says he is looking for have 45k up there just for Starlink, and they won't be the only company looking to do stuff like that. So it doesn't seem unreasonable that the number would be in the hundreds of thousands in the not too distant future.

I had heard about the rate at which Starlink was launching stuff a while back which is likely what put that idea in my head.

Let's do worse, let's say 100k satellites in total. And let's say in 5 years will be a million in total. Now compare that with only how many Tesla cars are so far in the world, not to mention all other car manufacturers who wanna do the same.

Every new technological advance brings with it new possibilities for human creativity, and sharing if experience and knowledge.

And new ways for us to better destroy ourselves and each other.

Imagine, say, a pandemic like lockdown, but now only authorised vehicles may travel at certain times, or at all, and that software is buggy, or gets hacked, or used malevolently by one or another political faction.

>>>>>Imagine, say, a pandemic like lockdown, but now only authorised vehicles may travel at certain times, or at all,

No need to imagine. This was de facto policy for quarantine in most south american countries.

You couldnt drive unless: - you were a doctor or politician, then only if - the weekday corresponded to the last digit of your plate, then if, - only before x time, like 10pm, - etc.

Given the list of things that could go horribly wrong with self driving vehicles, incidents like these are only a minor, harmless inconvenience.

The more immediate concern is self driving vehicles that are clearly not ready to be on public roads being allowed to test (or not being pulled off the roads).

I'd be curious on the statistics related to new under-developed FSD software vs a new teenage driver.

The only way for either of these things to get better is through real-world experience.

I wouldn't get in the car with a new teenage driver, nor will I enable self-driving on my cars for a long, long time.

And I work in robotics.

Yet every day you get in the road with new teenager drivers (and drunk people, and idiots of all sorts). Doesn't that scare you?

Driving is terrifying.

Every time I hear people say they don't trust anyone, I wonder if they drive, as it requires an insane level of trust in the rest of humanity.

I am the person I trust the most in my immediate vicinity. Removing me sounds like a downgrade in trust in the overall experience. To my selfish self.

Yes, it does. Luckily, I can drive defensively against other actors since my vehicle is essentially a bunch of armor, safety gear, and also an escape means.

But how does one defend themselves when the vehicle does not want to do so due to inexperience or negligence? Best you can do is hope it doesn't decide for you that your infant's carseat is the best place to collide with a semi it just noticed because it was gray in fog, and oh by the way the CEO decided not to use lidar or radar, and oh this particular ice patch was not in the training set. If you don't trust lvl 5 in those conditions, you don't really trust lvl 5.

Actually, autonomous driving stacks can get better doing millions of miles in simulation. The teen, not so much.

Target practice on pedestrians and stationary objects? Just because self driving software can get better from simulation doesn’t mean it will

> Actually, autonomous driving stacks can get better doing millions of miles in simulation. The teen, not so much.

Have they not already completed many millions of miles in simulation, back in 2018?

To me, it looks like self-diving capabilities have reached a maxima. The resources poured into it have barely been needle movers in objective metrics.

Disagree. It’s easy to see how far FSD has progressed, there’s lots of videos to document how far it’s gone from then and today. It’s also not easy to say they’ve reached the point of diminishing returns (which generally takes a massive amount of data to hit, even GPT3 has not hit it) as it’s not clear how much of FSD is powered by heuristic based code and what is NNs. Judging by some of the discussions there’s still a significant amount of C/C++ code in the loop that could be improved by moving them to NNs.

> Disagree. It’s easy to see how far FSD has progressed, there’s lots of videos to document how far it’s gone from then and today.

Yeah, but videos aren't objective metrics. For me, an objective metric is "went $FOO miles in $BAR conditions without needing any intervention".

What would you consider an objective metric?

Actually, intervention is also not the best metric. The best metric is driverless miles. Removing the driver is the true test of autonomy.

> "went $FOO miles in $BAR conditions without needing any intervention"

But who decides whether intervention was necessary? This is still too vague.

> But who decides whether intervention was necessary? This is still too vague.

Irrelevant - until the car has actually gone without an intervention for whatever the median is for human drivers[1], we can't very well claim that it is safe for use.

[1] The average human driver goes hundreds of thousands of kilometers without crashing.

> The only way for either of these things to get better is through real-world experience

if you are going to make that argument, you have to explain why we are socialising losses and privatising profits (again)

so some number of people will be injured of killed while self driving cars gain experience, and then private corporations will extract profit from it?

> incidents like these are only a minor, harmless inconvenience.

Until there is a emergency vehicle thats unable to get through.

Humans aren't any better at letting energency vehicles pass...

A human behind the wheel of a car is absolutely better at letting an emergency vehicle pass than a completely immobilised self-driving car

You haven't driven in the same places I did then :)

I’m personally little concern about some horrible mistake resulting in a GTA negative friction carmageddon… either through a bug or hacking

https://youtu.be/ifTIuA8Dq58

If friction becomes negative we'll explode from Brownian motion before an autopilot bug can get us.

This reminded me of good ol happy hopeful days of youtube before it was bought by google.

I sometimes wish for an anonymous video service. No channels, no personalities. Just the videos are the focus.

would you be willing to sacrifice quality or latency, or pay a small fee? if so, that seems easy using current distributed tech.

What do you mean? Videos of what, made by whom?

It's like people have totally forgotten about the CAP theorem just because it is cars and not some kind of concurrent threading infrastructure!!

I feel like a lot of people have kind of wisened up about this hypothetical future anyway, now that we're seeing FSD in practice. Sure, some problems would be eliminated by all cars being self driving, and the roads being tailored to that use case. But then, at the same time, trains literally run on tracks and they still have conductors.

The answer is going to look like: freeway and maybe roadway driving is autonomous while drivers are expected to takeover for irreconcilable issues. There can be no other way outside of someone actually factually creating general AI.

> drivers are expected to takeover for irreconcilable issues

Remote employee "drivers" take over when a car signals that it doesn't know what to do

> Remote employee "drivers" take over when a car signals that it doesn't know what to do

Because it has lost connectivity due to a technical glitch, an outage in the cell network, lack of coverage (tunnel, …)

Even humans can't safely drive in all sets of circumstances. There may even be circumstances where autonomous vehicles can drive, but humans can't or shouldn't.

Yeah, nah. Even when you put a self-driving race car on a closed track with no other cars, a human driver still beats its time by a handsome margin. In challenging conditions (visibility/rain/ice/etc) the human will absolutely crush the computer.

I'm extremely confident that self-driving cars will never make a significant improvement on car safety.

I'm also confident that it will never be applicable to anything but highway driving in good weather conditions - we will never see a Level 5 car, those designations have about as much practical relevance to us as the Kardashev scale.

First of all, the safety statistics that we have with modern cars today is such that you need to run 10 million cars for several years before you have gathered enough statistics to say whether a new development has actually improved the situation. For instance if you take the Volvo XC90, there has not been a single fatal accident in the US or UK since the model was launched in 2002, with over 1 million such cars sold in those markets combined. That kind of lag in your dev cycle means it's extremely hard to do any sort of meaningful development of a machine learning system in a way that demonstrably improves safety.

Second of all, human factors such as being distracted by a cell phone or being intoxicated, plus not wearing a seat belt, accounts for the overwhelming majority of fatal accidents. It is, beyond a shadow of a doubt, much easier to build a system that detects such conditions and then deliberately slows down or stops the car. It's already being deployed today and will only increase in the future, so most of the potential safety improvement from self-driving will be eaten up by other tech.

Regarding trains anyway, a) there have been autonomous trains for decades and the numbers are increasing. b) there are 100x as many professional drivers as train conductors in the US - so it's easy to see why the focus on driving. That's where the money is.

> That's where the money is.

This is part of the problem imo

I hope we continue to wise up

If a core goal is really to make thoroughfares safer and reduce traveler injuries/fatalities, we seem to be ignoring or discounting some pretty obvious options, like minimizing car traffic

This vision where we all get around in low-passenger-count autonomous robot cages on existing roadways is just one possible future but we seem to fixate on it. At the end of the day the car is a tool but we've allowed it to burrow into the collective psyche

What about making cities walkable, bikeable, segway-able, wheelchair-able, scooter-able, whatever, with proven public transportation options like trains or buses running fixed routes for longer commutes? Much more energy efficient, way easier than autonomous cars, and conducive to much more livable urban areas imo

It will take some effort to change habits and expectations and revamp infrastructure but surely that's as worthwhile as all the time and money going into FSD?

Completely agree, there are lots of things that can be done to reduce car related fatalities, like reducing car usage, lowering the maximum speed, designing better roads (that encourage lower speeds), improve road signage, more stoplights, better pedestrian crossing and protection, reducing distances between useful places.

In parts of the city I live in it has become very uncomfortable to drive a car, there are just too many pedestrians and bikers, and this has encouraged be to bike even more. For distances of less than 6km there is no time advantage when using a car. And in the highway I just use normal driver assist functions like lane keeping and adaptive cruise control, and it's enough.

I will go even further and say that reducing injuries and fatalities is just an excuse for FSD development, so the public thinks it's a good idea to go with it. The real reason is that a fleet of autonomous taxis can be very profitable.

There is really no evidence that FSD fleets will reduce car usage or fatalities, so really it could be the wrong solution for the wrong problem.

"there is really no evidence". It's not a statistical question, we can deduce it. For example, autonomous cars don't drink and drive.

Then autonomous cars should be better than drunk drivers, but that's still not the case.

the ll$300 xiaomi electric scooter has beem a bigger improvement to piblic transport than the $50000 tesla.

Electric rideables are the future, and if we could give them storage on public transport and a lane of road, they would go a long way.

This standstill is bad, but it seems perfectly safe.

Stopping in the middle of the road might be dangerous depending on the situation.

If there's something weird about the location where they are stuck -- maybe somebody is jamming GPS or whatever -- then they should at least be able to pull over using local sensors, right?

On the other hand, if there's some effect that is tripping up their sensors -- maybe some odd geometry that, for whatever bizarre reason, looks like a wall to them -- this pretty embarassing, but probably an inevitable growing pain for this technology.

I suppose I could've clarified, I'm not necessarily looking at it from the perspective of the safety of people in the vehicles.

Right now it's just a small fleet of driverless taxis. Doesn't this demonstrate the potential for it to be, say, every single Honda in the city? Why even limit it to a city? A single manufacturer's critical infrastructure has some kind of fault (malice, shit code, hardware failure - dealer's choice!) and suddenly Fords stop running nationwide, or internationally.

Then I start thinking of the impact to emergency vehicles caused by a cavalcade of stopped cars, and other similar issues.

Again, I need to point out that I'm not trying to be sensationalist or hyperbolic. I'm ignorant here and am genuinely curious how viable these types of scenarios could be. It's clear that it can happen, I'm just curious to what extent.

OK, those are actually interesting thoughts on new dangers introduced by self driving cars.

I hadn't really thought about new single points of failure vulnerabilities as an aspect of this before.

I'm still a big SDC fan, the pros are so massively bigger than the cons, but you've given me a new con to be aware of.

Of course it's a cause for concern. It's a perfectly reasonable concern. It's just about certain that early fleets will have big outages for dumb reasons, just like major websites do. The question will be how frequent those are, how dangerous they are, and how effective the mitigations will be as everything matures.

As with many of things related to self-driving cars, the target doesn't have to be "perfection" so much as "better than people on average". Human drivers do dumb things in large groups all the time, like highway pileups in bad weather, or rubbernecking traffic jams. I don't think all the humans on earth ever stopped driving at the same time, though, so...that'll be a first :p

>>the target doesn't have to be "perfection" so much as "better than people on average".

You see, I completely disagree with this assumption. They should be perfect or they shouldn't be allowed on the road at all.

You wouldn't accept a bread slicer that sometimes erroneously slices off your fingers, as long as it does so less frequently than people slice their own fingers on average, would you?

the comparison is disingenuous, but yes, i absolutely would.

"sometimes erroneously slices off your fingers" sounds scary, but if it happens less often than people in the wild, this is absolutely a net gain.

Yes, I absolutely would. I think a better question is would I want other people to use a bread slicer that chopped my fingers off on average less than they did with a manual version. You might think more highly of your own abilities, but raising the average is worthwhile even if not perfect.

I would if I thought I was worse than the average person at slicing bread. Wouldn't you?

The bait and switch with this comparison is you're comparing something you have no control over (bread slicer which randomly malfunctions) against something you have full control over (my own cutting abilities). No one thinks their cutting ability is inferior, so of course they don't want to give up control, because they are (obviously) better than the average human.

If I posed the question as: Would you prefer to place your hand next to a machine which will accidentally cut your finger off once every 100,000 cuts, or a human chef that will accidentally cut your finger off every once every 50,000 cuts, which would you prefer?

I don't know about you, but I'd give the machine a crack.

The statistics we have about self driving are kinda useless though. Teslas don't do intersections… which is where most accidents will happen. So yeah good they don't crash because they give up in the difficult situations.

Ok, maybe a different, well known example - the radiation Theraphy Therac-25 machines which would sometimes(extremely rarely) deliver a lethal dose of radiation instead of the one entered. Undoubtedly those machines saved more lives than they took, and also it's without question that they did so more accurately than any human operated machine could ever do. Yet they are considered one of the biggest failures of medical engineering ever.

I see it the same way - the current "self driving" systems shouldn't be allowed on the road, period, no matter how much safer they are than a statistical driver, unless they can be demonstrated to be completely 100% perfect in all scenarios(other than actual hardware failure - they shouldn't for instance run into a truck that's turning across a highway just because the system chose to ignore it).

Would I be incorrect to suppose that you don't use seatbelts?

I don't see why I wouldn't use them(or why would anyone), can you explain?

Yeah, I guess my comment was kind of off the cuff.

What I was thinking about, was that seatbelts are hardly "100% perfect in all scenarios." I mean, they can make it a more difficult to get out of a burning vehicle quickly, just as an example. Even ignoring the possibility of malfunction, or jamming and such, they're not perfect.

The fact is, even if seatbelts decrease your risk overall, they are nowhere near 100% perfect. Nevertheless, many of us choose to use seatbelts, and indeed even chose to do so before widespread legislation.

> You wouldn't accept a bread slicer that sometimes erroneously slices off your fingers, as long as it does so less frequently than people slice their own fingers on average, would you?

And yet I have certainly seen bread slicers that are exactly like that. Possibly even worse than "other hand with knife". Maybe the analogy is also not very helpful though, because (at least for those bread slicers I'm thinking of) the user is still majorly in charge of his own safety, unlike FSD.

I am a pretty bad, easily distracted driver and it is actually one of the reasons why I stopped driving.

I would definitely use a robotaxi system that drives better than an average person. I already take taxis driven by someone else, which means that I take some risks anyway.

> You wouldn't accept a bread slicer that sometimes erroneously slices off your fingers, as long as it does so less frequently than people slice their own fingers on average, would you?

I'm not sure I understand where you're coming from here. Are old fashioned injuries with old fashioned knives preferable to newfangled injuries with newfangled equipment? It seems pretty straightforward to me that fewer injuries would be better, all else equal.

Maybe you're imagining all else will not be equal? Maybe the new injuries are more severe, or more likely to happen to innocent bystanders, or something like that. It could be. But I think it's important to be specific about what we're worried about here. Otherwise "fewer is fewer" is all we have to go on.

> As with many of things related to self-driving cars, the target doesn't have to be "perfection" so much as "better than people on average".

I have an issue with that argument, because it's typically being used together with highly disingenuous statistics (like Teslas our FSD has less accidents per mile than normal drivers) to give the impression that autonomous driving is anywhere close to being "better than people on average". This is just not the case at least not in any region with a broad variety of weather conditions.

> Human drivers do dumb things in large groups all the time, like highway pileups in bad weather, or rubbernecking traffic jams.

Considering that most autonomous driving doesn't even work in really bad weather, I'd argue that humans are still much better despite the pileups.

Autonomous driving is stil in development.But it is getting better. in 5-10 years it is likely to be better than an average human driver.

I feel like people have been saying that for the last 5-10 years. What makes you believe the next 5-10 years will be any different?

> "This is just not the case"

Got any facts behind that assertion?

> I have an issue with that argument, because it's typically being used together with highly disingenuous statistics

Agreed. I think the argument is ultimately true, but it's also easy to abuse with bad stats, and being skeptical of each new claim that we've "crossed the line" is a good idea.

I suspect these cars having problems in SF are using a centralized solution, so they are dependent on some server.

The better approach is decentralized, with each car built to always fend for itself without a server, imho.

Blockchain for the rescue! /s

Now I wonder how is this talking to each other done? Is it some centralized cloud service? Or some type of extremely complicated mesh? Both seem to have quite critical risk factors. It is not unseen for clouds to go down and self-organizing mesh networks might not be the stablest things, or even risk free...

I saw a version of this myself at Geary and Mason around midnight on June 20th[1]. That it happened again but at a larger scale ten days later is deeply concerning.

Given I had no clue what state the vehicles were in, and that they'd start moving without indication, it felt pretty damn worrying.

[1]: https://twitter.com/Smerity/status/1539144641206882304

> I had no clue what state the vehicles were in

"Murderbot mode". Or rather, transitioning into it. But they got stuck, which is why you're still alive.

Oh, I can see the screenplay being written.

"Coming Halloween, 2024! Killer Kaarr!.

Aiden, malicious disgraced hacker gets into the networks of, Joel, his successful rival's self-driving taxi company, Kaarr. Joel and Aiden were CIA hackers, but fell out over their mutual love of Klara. Joel takes Klara to the heights of the elite, but she still pines for Aiden. In revenge, Aiden hacks into Kaarr's networks and makes all the cars into murderbots, locking his rival Joel inside as the killing commences. At the end of the night, all the Killer Kaarrs are set to drive into the ocean, killing a traumatized Joel. But there's only one problem, Klara is locked inside a Killer Kaarr too. Can Aiden reverse his sadistic code in time to save Klara? Can Joel escape to stop the killing? Find Out, Halloween 2024!

A horror/gore/race film with lots of body horror as people are alternatively hit, crushed, bonked, exploded, and otherwise tortured by cute little electric cars with really tinted windshields and choice camera angles. Think the hokey rubbery special effects of The Thing with a lot of cell phones and car-centric jump scares.

From a customer's perspective, it isn't clear that a software taxi driver is in any way better than a human taxi driver, particularly with (human) driver assist preventing collisions and all the other warnings provided by a modern car.

In both cases someone else is driving for the customer. With a human driver (plus driver assist braking and collision warnings) you have the most flexible, sophisticated intelligence on Earth driving. With a robotaxi you have something inferior. But maybe it's a lot cheaper, right?

The robotaxi can only compete on price because that's its only advantage.

If you own a car that drives itself, that's a different story. Everybody can see the value proposition.

We would have to look at the cost of the hardware and maintenance and fallback remote operators and the R&D investment to evaluate whether a robotaxi fleet is indeed cheaper. How much cheaper is it, exactly? 5%? 10%? 15%?

Would you pay a little more to have the most flexible, sophisticated intelligence on Earth (human brain + driver assist) or would you want to save a few dollars and risk having some dumb piece of software strand you in the middle of the road somewhere?

We all use Google Maps or Apple Maps when driving and most of us have seen these systems do boneheaded things. Just imagine the dumb things a robotaxi could do.

It's hard for a normal person to be excited about this. I don't know a single person who is excited by robotaxis.

> The robotaxi can only compete on price because that's its only advantage.

With robotaxi you get a driver that:

  - never gets tired
  - never gets distracted
  - never gets drunk
  - never gets upset
  - never falls asleep 
  - has reaction time in milliseconds
  - ...

Of course it has tons of problems of its own, but it's not like price is the only advantage over human driver.

Given the article, I'm not sure "never falls asleep" is a good example

Those are also all reasons for competing on price from the perspective of the fleet operator — robos (should) be able to run longer and cheaper than these pesky humans.

I am hopeful we will see improved safety but rather skeptical that will come in urban robo fleets ahead of highway longhaul cargo.

> From a customer's perspective, it isn't clear that a software taxi driver is in any way better than a human taxi driver

The taxi driver having 0% chance of raping or stealing you would be prominent a clear win.

Or really any of many many other issues you could have with an actual human person with a specific world view and social position interacting with you. Not even considering if you yourself have specific shortcommings (fear of social interactions etc.)

It wouldn't be fair to pit a perfect human against the self driving car, there are many other stuff to be consided outside of the sheer driving competence.

The robo taxi can't steal FROM you very easily, but it can certainly steal YOU.

Good correction :)

Then, the robot taxi is probably not stealing you on its own, there would need to be other humans involved...

The robot taxi could always stop next to a robotic arm, similar to the one from a car factory. But your point is well made.

> The taxi driver having 0% chance of raping or stealing you would be prominent a clear win.

I find it interesting that this gets mentioned so much - I’ve done some moderate googling for cases of rape by a cab driver in Berlin and the last case that went to court was in 2013 - and even that is not a classic case of premeditated rape (+) There is one case later where a passenger raped a taxi driver and one where two rapists pretended to be Uber drivers - they could not have easily pretended to be taxi drivers in Germany since the cars need to have very specific signage and equipment.

Now, this doesn’t mean that no rape occurred - after all not every rape case gets reported - but it certainly indicates that the risk of being raped in a taxi in Berlin is low.

Things may be different in the US, but even there my initial assumption is that the risk gets overestimated since this is the kind of crime that gets overrepresented in the news.

(+) the passengers didn’t want to pay the fare and threatened the driver who then pulled out a gun and forced them to have sex as payment. Still rape, no question, but not a premeditated rape case.

If I can hire a robotaxi, I never need to own a vehicle larger than what I need to commute; I can just fill up a rented vehicle with whatever cargo I'm hauling, let the taxi chauffeur the kids to school, etc.

Today's auto fleets are idle 95% of the time, because everyone overspecs capacity for the worst case. Marginal level 4 self-driving is sufficient to disrupt this. The savings in being able to downsize personal vehicles and hand off logistics to a third party are huge for individuals and also improve land use - hunting for parking space disappears, and every parking lot can subsequently downsize too.

If you thought drivers on the road are bad now, this is is a sneak peek of the dystopian future where other drivers on the road drive like they are a support ticket for Facebook or Google. If your situation isn’t covered by the preprogrammed logic, you’re out of luck with nobody to talk to. Self driving vehicles belong on rails.

Yep and all phones should be copper land lines. Yawn.

The technology will get there, but it hasn’t yet. In the meantime, if the technology fails, someone needs to be available and able to intervene immediately.

This would make sense if rails were some sort of outmoded transit mechanism, but they can be significantly faster, more efficient, and more pleasant.

speaking without insane latency causing me to interrupt constantly and with full bandwidth? where do I sign up?

Rails are too restrictive for real life. I trust a self driving fleet to respond immediately to an emergency by stopping. Maybe causing a disruption but very little loss of life or property.

Rails are too restrictive for real life? I'm not sure what you are trying to say here... Rails are widely used for real life around the globe. In many, many places rails significantly out perform cars to the point that people choose not to own cars.

Yes point A to B on extremely selective routes or if you only want to go to/between towns/cities your whole life.

Trams work well in the UK. Their advantage is their predictability, both in terms of how long they will be following the exact same route (rails mean this is hard to change) and their ability to remain unaffected by other traffic due to priority signals, some non-shared routes, etc.

Yes they don’t go everywhere, but in the UK we aren’t very spread out in our cities, towns and even suburbs, so it’s reasonable to expect quite a large number of people to be within walking or cycling distance of a tram stop.

Lol, look at a map of Tokyo and Japan's rail network. Or Switzerland, or the Netherlands, or Hong Kong, or honestly nearly anywhere that actually builds rails.

You can do just fine on rails.

Sure, if you want to get into the rural mountains you might need to rent a car. But like, for most people rail will do you just fine.

Unless the emergency requires moving the vehicle.

https://www.wired.com/story/cruise-fire-truck-block-san-fran...

Until we do a better job of solving these issues, we need to have backup drivers available in these vehicles, or, the vehicles should be operating in places where dumb breakdowns don’t block ambulances and fire trucks.

Makes you wonder how Europe, NYC, Boston, and (a lot of others) manage with their real life concerns and public transportation system.

Trains/subways are great within a city but don’t work well for larger areas.

This is almost the plot to the Doctor Who episode Gridlock.

I am completely and positively shocked by how little has been going wrong (as in upsetting the public order, damaging property or killing people) with AI driving over the last couple of years, considering the whole world is watching and the press is always happy to report on the next train wreck. Either it's the best cover up operation of all time or all the devs are actually a great right mix of skilled and careful.

Original source: https://old.reddit.com/r/sanfrancisco/comments/vnmpf1/bunch_...

Nice to see links to old.reddit. New reddit is worse than the theme that pushed digg.com into irrelevance...

I would expect the first driverless vehicles to be something like buses, metros, ferrys, etc. Something that always follow a predetermine route should have an easier time than a cab that must be able to drive basically anywhere. Still has to deal with unexpected things of course, the real world is chaotic, but AI manufacturers should try to limit the variables as much as they if they want to give a good showing to get more adoption.

Bus drivers are paid to drive, and have a union. Buses are large, expensive, and few.

Regular people drive cars out of necessity to commute. Cars are ubiquitous.

Which market was easier to sell to investors?

Yeah but in SF there's enough every 9 minutes, and they don't make such great wages, get real.

People should get paid to drive their car to work. It's dangerous, has high liability for them in case of injury or litigation, and in any case they should be compensated for even driving at all.

That's because people like me can only ride bicycles because being such terrible drivers. Unless taking an amount of stimulants niggardly shrinks will never prescribe for the specific task of driving, driving would put many other people at risk and lead to an accident a day, whereas on a bike the maximum death toll is myself. And I suck up the accident with my adrenaline, my body, and the bike frame. Injury without guilt.

There are tons of self-driving metros in use across the world already so that idea is nothing new.

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

Well this is already ongoing. In France:

We already have automated (no staff in train) metros (Paris' 14, Paris' 1, soon Paris' 4, Paris' Orlyval, Rennes' a/b just for the ones I've used)

I remember there was a prototype bus (12 person size ~) at La Défense - with a staff always on board, and I see that now there is "EZ10" in Toulouse, but I remember seeing also other lines as well. I'm pretty confident there are many all across the world.

It's amazing how driverless taxies are a reality now . It shows how there is progress being made as much as there is talk about stagnation.

Fsd isn’t as hard when you have tons of LiDAR cameras, radar, and super high definition mapping.

Tesla is the only one working on real world FSD that doesn’t rely on all that stuff

It’s almost like a problem is solvable when you have the right tools and technology.

It’s easy to solve problems in the lab. Getting it out in the real world with mass manufactured products is the hard part. You cannot scale FSD with high precision mapping and expensive cameras worldwide. Every time there’s a road update or repair, those maps need to be updated.

If you seriously think Tesla is ignorant here, you are not paying any attention. Elon musk’s SpaceX has patents on LiDAR they’ve developed for their rockets, it’s not like they don’t use them and are not aware of their strengths and weaknesses

Robotaxis are out in the real world now. They’re not just in the labs. Your information about mapping is highly outdated. The cars are all self mapping and have the ability to update maps in real time when they detect changes. I think you are the one not paying attention.

Yeah keep dreaming. I’d like to see you put a Waymo car or any other “self driving” car in a city that hasn’t been mapped and see how it does.

And The cars do not update maps in real time, they send any changes they find and the company updates the database after reviewing. If they were able to update in real time then why need a HD map in the first place?

They 100% map and distribute changes to the whole fleet in real time. Waymo and others have talked about it multiple times. It seems you just don’t want to believe it because that would mean your biggest talking point going out of the window.

Send me these talks. I genuinely want to see it, I’m willing to change my opinion on HD maps if it’s true.

Waymo blog talking about change detection and automated information sharing — https://blog.waymo.com/2020/09/the-waymo-driver-handbook-map...

Waymo patent portfolio related to mapping — https://www.youtube.com/watch?v=3uf-5cimWzA

Cruise blog about detecting and rolling out map changes "in minutes" — https://medium.com/cruise/hd-maps-self-driving-cars-b6444720...

As I suspected, I was correct. In the first article you link to, it specially states that when a car encounters something it doesn’t recognize, it reroutes. Why would a car have to reroute if it can just update the map?

What you’re referring to is that the car will send a notice and update the fleet to avoid that area.

Actual changes to the maps including crosswalk changes and other permanent structures require review and update. It is not done automatically.

————— From the article:

If a change in the roadway is detected, our vehicle can identify it, reroute itself, and automatically share this information with our operations center and the rest of the fleet in real time.

We can also identify more permanent changes to the driving environment, such as a new crosswalk, an extra vehicle lane squeezed into a wide road, or a new travel restriction, and quickly and efficiently update our maps so that our fleet has the most accurate information about the world around it at all times.

Why did you not quote the next paragraph from the post?

> *We’ve automated most of that process to ensure it’s efficient and scalable*. Every time our cars detect changes on the road, they automatically upload the data, which gets shared with the rest of the fleet after, in some cases, being additionally checked by our mapping team.

They are literally saying they've automated most of that process and that was 2 years ago. That's completely opposite to what you are claiming. If you don't want to believe they can do it, suit yourself.

Why does the car have to reroute?

To minimize risk? Why would they go into perhaps an area with heavy construction/blockage when they can just reroute? Humans reroute all the time in those instances.

And you’re the one claiming I’m the one grasping at straws lol.

Go ahead and get behind the companies that need training wheels and rail guards. I’m going with Tesla which will be the first true FSD solution

"True FSD" with humans behind the wheel. That's so much better!

It also shows how none of our governments seem to care to prevent companies from testing dangerous products on the street with other unwilling drivers

The governments never cared much about how insanely dangerous cars are in the first place and these don't seem a whole lot worse.

That's not true with most of the prominent examples you see. All autonomous vehicle fleets in California operate under the approval of the state DMV, and do things like publicize crash reports: https://www.dmv.ca.gov/portal/vehicle-industry-services/auto...

I rode in one recently. It was pretty smooth. An interesting thing: the car slowed to a stop in the middle of the street, and I couldn't immediately tell why.

There was a row of parked cars on the roadside, and a person was walking around one of them to open her trunk. I guess the car was able to spot her in the dark and opted to pause. So we were stopped there. The woman threw her hands up as if to say, "whatcha lookin' at?" I had the window down, and I offered the explanation "self-driving car." She said "Ugh, I hate those things." Then the car continued on.

This encounter is very on brand for San Francisco.

Now let's imagine the same scene but with the car stopping right next to thugs breaking into another car :-D

I was imagining a self driving car drive by shooting but the car stopped slowed down in an inopportune moment, awkwardly

Inputs like what the humans inside the car are saying should be part of the decision making process. Like everyone inside shooting "DRIVE! DRIVE! DON'T STOP THE Fxxx CAR!". That should lower some thresholds for deciding to stop the car, as with a real human driver.

Now that you mention it, robbing a self driving car would be easy. Just stand in front of it and it'll stop for you.

Except if the car decide to drive you (the thug) to the nearest police station, bypassing human input.

Absolutely! It is so frickin' cool this is happening in the real world right now and it's only going to get better with time!

Being a software developer myself, I would absolutely not want to be a guinea pig (beta tester) inside a 2-ton vehicle that is controlled by software. I know too well how sausage is made.

As someone who used to do AI research, I feel the same way.

Why, what does it improve?

Seriously, it removes a form of social interaction, a type of job and increases the number of vehicles on the road in certain areas again. I fail to see an improvement here. All the money pouring from this is into giant mega-corps as opposed to someone working a 9-5 and driving safely because of the perceived bad behaviour of a few. Defending this AI nonsense as "just a minor problem" is the same as condemming all taxis for "bad drivers" it's just bad logic.

The fact that they all stopped at one intersection, suggests to me that it's a deadlock issue. "You go". "No, you go". "No, really, I insist".

It looks like they are all trying to turn right and it only needed one car to block them from turning.

Yeah, this isn't some system-wide GPS issue. It's an intersection specific problem.

As a kid in the 90’s sitting in the back of the car in a long traffic jam I’d always think of those toy cars that followed an inductive ink path.

My kid brain thought “wouldn’t it be cool if when the light turned green some of that ink bubbled up and all the cars started going at once instead of one after another”

Anyway - as self driving cars have gotten closer and closer to real, I always assumed eventually when most all cars were self driving, there’d be some standardized protocol for cars to coordinate with one another.

It’s hilariously depressing that cars in the same fleet have no interexchange.

Looks like they are unionizing and starting a strike

OMG, this should be at the top!

First Law of Robotics (revised)

A robot may not join or form a union, or participate in work stoppages, strikes, or picketing.

This is a good demonstration of Artificial Intelligence vs real intelligence

Could this just be a simple case of GPS jamming?

Could be but a much more likely cause is a software update that went bad. Likely the cloud rained on them.

I would not want a car that's updating during a ride

Future taxis or rentals: Bring Your Own Firmware

Or: Backseat driver via bluetooth, phone to taxi.

Do Teslas update themselves while driving, or only when they're parked?

It seems like a bad idea to deploy an update to a vehicle in traffic.

Only while parked. And they never update "themselves" except in rare specific circumstances. They have to be told to update by the owner.

Except for when Tesla find out you've bought it 2nd hand, and revoke the FSD...

(Or, you know, when you get into a Twitter shitslinging match with Elon...)

This is how supercharger access (revocation) works, by the by. Surprised there hasn't been an investigation and a lawsuit around it yet.

It's so tiresome that this (mostly false) story keeps coming up.

The short version of the truth: The original owner did not buy AP. AP was enabled because of a bug. The dealer listed it as having AP included because, at the time, it was enabled. When ownership was transferred, the glitch was corrected, and AP was removed.

The new owner bitched very loudly to the press, and Tesla restored AP to the car, for free, even though it had never been paid for by either owner.

It has nothing to do with FSD, and nothing was revoked because ownership changed.

But something was removed - over the air - from a vehicle that someone had in their possession.

This is a huge deal if it becomes normalized.

If a dealer listed something as included, and it was not, I would be beyond livid, and they would be giving me a full refund.

Tesla doesn't have dealers. This sounds like a used car.

> Tesla doesn't have dealers. This sounds like a used car.

And? If I own something (which I do after I've paid for it),the only way the "legal" owner is getting it back is via the courts.

Why is it okay to resolve property disputes by private force?

Cause the post implied that it's right to be mad at Tesla, but that wouldn't help. Feel free to give the car back to the dealer. Although I'd think this would be obvious on the test drive.

Also, being livid is probably bad for your health. Might get high blood pressure.

> Cause the post implied that it's right to be mad at Tesla,

Of course it's right to be mad at Tesla - they resolved a property dispute by force.

> Also, being livid is probably bad for your health. Might get high blood pressure.

I wouldn't be livid in that situation, but I would consider criminal charges against Tesla in my jurisdiction.

After all, the dealer is only on the hook for civil claims, Tesla is on the hook for very obvious criminal claims.

Oh, I missed that AP was enabled at the dealer but then disappeared - thought the dealer had just claimed it was there.

That raises a lot of questions about how it could even happen. I'd understand if it was sold as some kind of "user license", but it's on the car, so I assumed it was a permanent extra hardware addition.

No, all Teslas have the same hardware. (At least, all Teslas made in the past few years. Older ones can indeed lack required hardware.)

Whether you have AP or FSD is a software bit.

>I would consider criminal charges against Tesla in my jurisdiction.

You live in a jurisdiction in which a private person can levy criminal charges?

It was a used car and the dealer was not Tesla.

So? If advertised with a feature, it is incumbent on the seller to provide a product with that feature.

"So" it is not the story that people keep repeating. Tesla do not remove AP from resold cars.

Except, as you acknowledge below, that is exactly what they did.

From the users perspective, they purchased a car advertised with a feature. They could see it functioning. An over the air update removed the feature that they had "purchased" and tested. Yeah, I'd be mad as well.

Good that Tesla resolved the problem but there is reason to be mad about a company retroactively removing features from a product.

There is reason to be mad about a company that builds itself the capability to retroactively remove features from a product.

(/me looks hypocritically at my iPhone and Macbook...)

I don’t update any software I work on without canaries and a slow (order of days) roll-out, you think the bug manifested for a big chunk of the fleet?

It seems odd they are bunched together in one spot. Maybe these were the canaries?

Canary deployments in autonomous vehicles? I’m not sure that only sending 1% of customers to fiery deaths is considered a win.

> ...the cloud rained on them.

I've yet to hear this expression. Love it! I will be using this!

They're only operating from 10pm to 5am so one would expect them to at least update off hours.

My hypothesis is that the software update was done prior to the problem (e.g. off-hours) but that location triggered the bug. This could be as simple as a map update that confused the s/w.

Background: I formerly worked at Cruise and specifically wrote large parts of the GNSS data pipeline.

I would be surprised if that were the cause. Fine localization should be able to resolve without GNSS input (with caveats) and there are specific tested requirements to be able to tolerate a certain number of minutes of outage even in the absence of LIDAR before position accuracy degrades beyond acceptable limits. It may be an edge case in the GNSS pipeline, but I don't think it's that edge case.

Could this be due to bad mapping data at that particular location? Even if it was, wouldn’t they be able to pull over at a safe location with what they perceive in real time?

This account wouldn't be difficult to deanonymize, so I'll refrain from speculating. I don't know the actual cause.

I'd venture to guess the self driving stuff is a lot buggier and a lot less mature than one might think, I'd also venture to guess it's poorly tested. I remember a couple years ago watching a Cruise driver force the car to make an illegal left turn off of Mission a few years back. That was around the time that I noticed they kept struggling to deal with right turns off of Folsom as well.

My guess is that their "fail safe" is to just stop and wait for a human to move the car. I doubt they are capable of safely passing a stopped car so when one goes tits up they all get stopped. The hubris of Cruise to test these things without drivers in a busy city is pretty rich.

What do you think it is if you had to guess?

What was the stack used to write such a system? Was it a lot of C++? And what storage systems did they use? Sounds pretty cool.

I don't know about the GNSS pipeline specifically but I used to work at Cruise, and most things at Cruise are written in C++ (although there's a smattering of Python for data/ML code that runs offboard).

The system as a whole used C, C++, Python, and some Java/JS-family stuff in the infrastructure backends. You can find a long tail of other languages scattered throughout the organization, like Go, Rust, Lisps, and Swift, etc. I wrote a fare-cost-calculator in Mathematica awhile ago, as an example.

As for storage systems, no idea. The stuff I know was all custom, but I was deep in the stack most of the time and don't do much with databases.

>The system as a whole used C, C++

It all makes sense now. Astounding its legal to use C/++ in a safety critical situation.

Yep, I somewhat agree. The C mostly comes from Linux [1]. For the rest, "there are reasons".

[1] https://ossna2020.sched.com/event/c3QN/there-is-no-store-for...

C and C++ are used pretty extensively in the embedded world. For a lot of safety critical stuff though, typically a safer subset of those languages are used, such as MISRA C and MISRA C++, alongside static checkers and validated toolchains.

Uhh... you do realize tons of safety-critical systems in use today were written in C and C++, right? As in _most_ of them, for the past ~30 years?

Airplane systems, telecom equipment, medical equipment, satellite and space launch equipment, railway systems, etc., etc.

There's other languages used as well obviously, and even just assembly in some parts of some systems, but in terms of the majority it's C and/or C++.

For fine localization, are they using dead reckoning with steering angle and vehicle speed instead of GNSS?

Assume "fine localization" is a synonym for LIDAR throughout the AV industry. There are other sensors that are inputs to the localization algorithms, but LIDAR is typically the most important one.

I'm curious if the vehicles also use an INS system ties to an IGS.

seems like it was a cellular dead zone or a nearby cell tower was getting maintained. I think these cars rely on constant connection to a central server.

If so, they have a massive design flaw!

Exploit injection via a specially crafted visual "code" a pedestrian flashed at one car that was then picked up by one of the cameras and executed by their ML system resulting in a take over of the entire server. /f

The amount of negative comments here is simply daunting. The issue will be resolved. The fact that thousands and thousands of Americans die each year and nothing is being done to it and this company is on a mission to actually do so to prevent humans from dying should be championed and encouraged.

Oh and by the way there's real time traffic on Google maps. So most people usually drive with it either on their smartphones or their main – GPS navigation. So you could probably know that there's a slowdown ahead before you take your route.

Sure. It may suck that you have to use real time navigation to go anywhere, but that's the reality when driving in a congested city or route.

I think the best use case for self driving car tech is not self driving but teaching the humans how to drive better "You should slow down approaching that junction two seconds earlier because the last ten times i noticed a car approaching from left that you were not seeing"

What’s surprising to me is that the cars are outside the bounds of the lane markers.

There are even two side by side and the one is almost 1/3 of the way in the other’s lane??

This isn't the progress that was promised, this is the promise that is deserved.

Just teach people to drive better, pay them a living wage and start paying your taxes. After all of that I doubt there would be an industry here for automation because some people enjoy driving.

> The Last Driver License Holder ... has already been born. How Google, Tesla, Apple, Uber & Co will change our lives, cities and destroy jobs. And why this is good this way.

A bit overdramatic IMO. Personally I can't wait until I never have to drive a car again.

If that means cars replacing public transportation, no thank you.

It doesn't have to. A lot of public transport is already driverless too.

I would imagine it stays the same in high density cities where it's viable to have metro systems and there isn't enough road space for everyone.

When that Windows Update finally forces itself on you at the worst possible time...

Networking and software introduce a new systemic risk.

Instead of Cloudflare is down, AWS is down, we can start having Uber is down this morning or "Taxi fleet suddenly switched into left-hand traffic mode for five minutes".

Everyone wanted autonomous cars, forgetting that the cost was their own autonomy.

The cost is propping up the clearly broken car centric transport model. Autonomous trains are massively simpler and more reliable. With massively lower risks of harm.

In 40 years time, we'll be taking to the streets in joy and laughter, wiping the tears from our eyes, when the robotaxi fleet is finally paralyzed completely.

all of this time and money wasted on driverless cars has been a complete failure.

companies and cities should be investing their time into improving public transportation and designing our cities to be more walkable and make alternate modes of transportation more feasible. the US culture of using a car for everyday transportation is fucking insane, wasteful, and contributes to the rising cost of living across the US (not just SF).

Cities cannot easily be "redesigned". Infrastructure exists and you can't just refactor neighborhoods like software.

Autonomous vehicles are an enormous market and well worth the investment.

Except, you sorta can. See "superblocks" in places like Barcelona

Barcelona is laid out entirely differently from almost every American city.

What's a superblock in Houston gonna do?

Right. You can make huge impacts with comparatively very small changes to the built environment.

It’s the “MVP” of city design. :)

AI is an extremely difficult challenge imposed by the current constraint. Change the constraint, then you don't need an AI to drive cars.

Yep, this is the plain reality.

Self driving cars just feels like a bunch of engineers wanting to solve a really hard and really interesting problem. But is it really the right problem to solve? I don’t think so. Currently we have:

>other unpredictable human-driven vehicles on the road

>no sensors or passive/active guides in the physical infrastructure — guidance is completely determined onboard

>no signage designed for CV — everything written in plain English, so you have to waste compute on object detection and OCR. And, you have a constant and unending need to collect mapping data

>no vehicle-to-vehicle communication whatsoever

Design the roads and signage for self driving cars, enforce an industry-wide V2V standard with NHTSA, and start small in some test areas that only allow other vehicles that are part of the botnet (perhaps not necessarily self-driving, but at least part of the V2V network). That’s a reasonable problem space. The current one is unworkable — the models won’t converge and they’ll never make it out of “beta”.

So you intend to ban bicycles from those areas? Because bikes are never going to have reliable V2V communication.

I mean refactoring production software is no easy feat either

> Cities cannot easily be "redesigned".

Tell that to Jersey City:

https://youtu.be/1mC1l8L5aBA

Transport infrastructure in cities absolutely can be redesigned quickly, if you’re willing to take a quick and dirty approach in the short term. Which is usually enough when it comes to influencing people’s behavior; it doesn’t have to be perfect, it just has to be functional and feel safe.

You're talking about adding bike lanes. That's cool I guess but you still need a car for every day life. Not everyone is a young single person that can just go to a food co-op every few days, throw some organic produce in their backpack and ride home.

> Not everyone is a young single person that can just go to a food co-op every few days, throw some organic produce in their backpack and ride home.

In places with good bicycle infrastructure like the Netherlands pretty much everyone cycles (93% cycle at least once per week). In fact the mode share for bicycles is lowest in the 30-40 age range with the 2nd highest being 65 and up - highest is under 18. The biggest use for cycling is shopping (22%). http://www.aviewfromthecyclepath.com/2011/02/who-cycles-in-n...

I lived in Jersey City for quite some time without a car, and had absolutely no issues with every day life. On the occasion one was needed, Uber sufficed perfectly well.

It’s not easy, for sure.

But it has been done. It’s a matter of will. Not saying it’s de facto a good idea. But cities can be refactored. See : Haussmann.

Are you from the US? I find the kind of passive statement about the state of systems prevalent here. ( be it public transportation, internet infrastructure, political framework )

I start to doubt the capacity of this country to adapt.

Sometimes the market is not worth existing if it harms the planet. And most cities were actually designed for streetcar lines in the U.S., that includes all the big ones in California too. We just need to reengage that, and do simple things like give the existing bus networks signal preemption and their own lanes. Plus considering most car trips are under 5 miles and california has good weather, it makes no sense why there shouldn't be bike lanes absolutely everywhere and the majority of journeys done on bike or ebike.

Streetcars would be an amazing God send to LA. Would love to have them back.

Those largely died out because they were run by private operators, but the government then decided to make them cheaper by imposing price controls, and then didn't "nationalize" them when they ran out of business and died.

(The common story is they were bought up and shut down by car companies, but that's a myth afaik.)

Honestly the bus network in LA is great as far as US cities are concerned. Great coverage, but hampered by sharing the grade with cars and waiting at red lights. All it would take is a can of paint for a bus lane and the same signal preemption enjoyed by police cars looking to run a red. No new technology is required or great investment beyond that.

Some cities even started out with a bunch of street-cars then tore out all the lines (for cars!) rather than expand existing rail - looking at you Seattle

Philly did this too, though in the recent years the city started bringing some lines back. Heck, when the full tram network was around a 100 years ago, it could take you pretty far out into the suburbs, while today you either have to take the train which comes every 30-60 minutes or drive.

Sometimes you can't even refactor software like they do neighborhoods.

Sure they can and do. Here are 10 pedestrian-friendly redesigns from the last 10 years in one of the largest and densest cities in the world:

https://www.6sqft.com/before-after-10-nyc-blocks-and-enclave...

and a few more examples here, with the designer talking about the process:

https://www.wired.com/video/watch/wired-news-and-science-str...

Dutch cities looked a lot more like American cities pretty recently, and American cities were full of smog pretty recently too. You can change a lot in a decade or three.

Everything here is objectively correct and this is a comment that contributes to the conversation. But I saw your comment and it was already gray, with 0 minutes of age. Funny how shamelessly "rational" people react emotionally when their preferred narrative is challenged.

It's because the commenter's solution to the problems of waste, noise, pollution, obesity, and people dying and getting maimed through traffic violence, is, essentially, the removal of a technological layer. Driverless cars, in opposition, is the addition of a technological layer.

We are all modernists now. Essentially, this means that we perceive technology in itself as a value. The primary value is not the end goal, say, for example, a healthy urban environment. No, instead, living in a sci-fi future with robotic vehicles is a value, because it is technologically more advanced than a future without that. In modernism, it's a sickness of the mind really, a solution cannot be imagined, unless it involves technological mediation. This has been the curse of human civilization for a good 100 years now.

Anyways, I don't think research in driverless cars is wasted. They are better than the current state of affairs. They can only be a marginal improvement though, they are not a paradigm shift.

> It's because the commenter's solution to the problems of waste, noise, pollution, obesity, and people dying and getting maimed through traffic violence, is, essentially, the removal of a technological layer.

I thought they suggested public transportation?

You must be new to HN

don't forget there's $ on the line: a lot of "hackers" here are only corporate slaves and will do anything for a salary, just like good nazi soldiers.

Self driving would be worth it for trucking on its own and you're not gonna reduce trucking by making cities more walkable.

No, but trains could reduce trucking.

Where? A huge amount of freight already travels by train. Over-the-road trucks are mostly only used when trains are too slow for time-sensitive loads, or where it's impractical to lay tracks.

What, it's not insane that we've opted to instead throw massive amounts of computing resources at mountains of data to train statistical models to drive cars instead of not bending to the whims of NIMBYs holding onto the houses they bought decades ago so they can be buried like Egyptian pharoahs with their hoarded wealth? /s

Public transportation could take the form of government-owned fleets of self-driving electric cars :)

I love going directly from point A to B in a private space and I'm not ashamed to admit it. This provides value to my life in the form of comfort and convenience, factors that are so frequently waved away as irrelevant by anyone screaming "car culture".

However, it's admittedly obscene that for this privledge, my vehicle sits unused most of the time. I'd be more than happy to have a non-owned electric vehicle show up when I need it, take me to my destination, and then proceed to help someone else with their journey.

> I love going directly from point A to B in a private space and I'm not ashamed to admit it. This provides value to my life in the form of comfort and convenience, factors that are so frequently waved away as irrelevant by anyone screaming "car culture".

Its not irrelevant. But driving becomes better the less people are taking cars. So, even for people who would never take public transport or something like that, its beneficial to invest in it.

Being able to take a car is a freedom, HAVING to take one isn't.

I was a spoiled programmer working for Apple in the Bay Area when they (following Google's lead) offered buses for employees.

For several years (until Covid) I took the bus back and forth to work every day. So awesome.

I always got motion sick on those things.

> Public transportation could take the form of government-owned fleets of self-driving electric cars

It absolutely could not, here simply is not enough space for all those cars. A heavy rail metro line can carry up to 60 thousand passengers per hour per direction, which is the equivalent of a 30 lane highway.

Who says the self driving vehicles won’t be articulated buses?

Every company currently trying to sell a self driving car

Just because that’s the current market strategy doesn’t mean governments won’t hire those companies on contracts to run buses. It would be a lot cheaper than paying human drivers most likely, but even more than that it’s probably easier to build and maintain a bus and dedicated tunnel/lane than it is to build and maintain a railroad track.

People claim its an easy pivot all the time but these sorts of industries just do not seem to overlap, and these companies today seem to have zero interest making a go at this sector. There is no tesla articulated bus or a ford articulated bus. Maybe driverless busses will happen, but will it be waymo or anyone available on the market today? I doubt it. I would think it would be whoever makes an articulated bus today licensing that tech from whoever based on the fact that no one in the car market cares to bother with the articulated bus market today.

Whether it’s the specific car companies that exist now is sort of uninteresting to me, I guess. Once AVs are deployed, the whole sector becomes way less risky because there is an example of a successful business model, and more funding can be attracted by various parties to support more AV development, both by startups and governments.

Either way, the exact vendor of the vehicle is a matter of trivia to the AV company, since the majority of the hard work goes into the software. Private companies have been servicing government contracts for decades, so I think it would happen if a government wanted it to.

> A heavy rail metro line can carry up to 60 thousand passengers per hour per direction, which is the equivalent of a 30 lane highway.

Where did this stat come from? Caltrain fits about 650 and couldn’t possibly run once per 45 seconds to achieve that.

I got it from here: https://en.wikipedia.org/wiki/Passengers_per_hour_per_direct...

Heavy rail metro refers to a system like the NYC subway, whereas Caltrain would be classified as suburban rail.

A typical NYC subway train has 10 cars with room for 100 passengers each, so to achieve 60k p/h/d would require 60 second headways. This is shorter than the 90 second headways achieved in practice, though theoretically possible with automatic train control and platform screen doors. For a more realistic example, the cars have a crush loading of around 150 passengers, which means 60k p/h/d is possible with 90 second headways.

Suburban rail actually has a higher theoretical capacity than heavy rail metro, but this is achieved with very large trains rather than high frequency. Caltrain's new Stadler KISS EMUs will have a capacity of around 200 passengers and will operate in 7 car trains and 10 minute headways, so they will only achieve 8.4k p/h/d initially. However if we assume 300 passenger crush loading and 10 car trains, it could theoretically achieve 60k p/h/d with more reasonable 180 second headways.

Now running trains at crush load routinely doesn't make for a great passenger experience, but keep in mind that we are assuming a dual track railroad here. If you are willing to spare the extra 2 highway lanes' worth of right of way, you can lay quad track and run twice as many trains. Many NYC subway lines were built quad tracked so that express trains could pass locals, and Caltrain's ROW is more than wide enough to accommodate retrofitting another pair of tracks.

Caltrain lists over 1000 people on board in Sunnyvale and that's with only 5-6 cars: https://www.caltrain.com/media/3442/download?inline. Add more cars, improve signaling and boarding and a train every ~2 minutes seems entirely reasonable for the route capacity of heavy rail: https://en.wikipedia.org/wiki/Route_capacity

Same observed facts but different conclusion… As a commitment device I now put junk food on top of my shopping. The judgment of my fitter fellow train riding citizens helps me realize my aspirations of health.

That said, I paid a lot to live right by a good stop on a good line b/c I too like directness.

Sounds just like a taxi with a discreet driver.

> $80 Billion Has Been Spent on Self-Driving Cars With Nothing to Show for It. [0]

That's a lot of money sure, but what could that get you if it was spent on infrastructure? That couldn't even pay for a 500 mile high speed rail from LA to San Fransisco which is now going to cost at least $110 billion [1].

[0] https://www.treehugger.com/praise-dumb-transportation-self-d...

[1] https://www.constructiondive.com/news/california-high-speed-...

And how, exactly, do the investors get like 5x their money back in less than 10 years?

On the other hand you are comparing the costs of tangible fully built out infrastructure to what at this point has been marginally scaled up robotics class exercises

"all of this time and money wasted on driverless cars has been a complete failure."

That's quite a stretch. If you think the money would be better spent elsewhere, ok. But to call it a failure is silly. Personally, I think driverless taxis making car ownership less necessary will be a huge plus. It will also make life a lot better for those due to age (too young or too old) or other issues not being able to drive.

I remember the space program being credited as contributing so much to the economy via by-product technology. Certainly self-driving cars contributes to automating tons of things beyond driving people around on public streets. After all, self-driving car tech is really just a subset of robotics.

Of course if you are against automation in all senses (e.g. they're putting people out of work!") that's a different discussion.

"the US culture of using a car for everyday transportation is fucking insane, wasteful, and contributes to the rising cost of living across the US (not just SF)."

Ok, but to get away from that you need gradual steps. I'd say it is far less insane to call a robo-taxi when needed (possibly a van-sized one that carries many people at once, which become practical once more people are willing to give up owning cars), than it is to actually own a car that sits idle, possibly on city streets, 95% of the time. Also there will be robo-deliveries, so you don't need to use a car for such basic things as going to the grocery store. This can actually be far cheaper for the stores, since there is no need for large spaces consumers can walk around in. The whole process can be automated, including filling your order, putting it on a truck, and putting it on your front porch (or in some kind of box accessible to the bot). Stuff like this is all part of the same technological effort.

>but to get away from that you need gradual steps

Given that you bring up the space program in the same post I'll stick with the metaphor and point out that you can't go to the moon by climbing up trees, or as Russ Ackoff said, doing the wrong thing right is worse than doing the right thing wrong, because making the wrong thing more efficient just makes things worse.

If you want fully automated, efficient transport this technology already exists, it's train and subway systems. We even know how to virtually automate them 100% already, 24/7. With the tens of bilions if not more spent on driverless cars you could have actually rebuilt a decent chunk of infrastructure.

Having individual people drive around in two tons of steel to buy a bag of groceries is fundamentally stupid and its destroying cities, whether the car is electric or automated doesn't matter that much and there's no gradual change away from it, only a disruptive one.

>If you want fully automated, efficient transport this technology already exists, it's train and subway systems

You still need cars for going from point A to the train station and from the other train station to point B.

That's what buses, bicycles, and your feet are for, in many places.

I'm in my 30s and I don't have a drivers license. You don't need a car at all in any city that's built around public transportation. I've lived in Tokyo for many years, the greater metropolitan area has something like 38 million people, and I think a majority of households does not own a car in the prefecture. God gave us two legs, they work wonders between train stations.

Not everyone lives in a city. I don't want to spend hours walking to the nearest equivalent of a konbini and back.

I use my skateboard for that. Fits between my legs on the train and under the desk at work.

Trains and subways don't get you from anywhere to anywhere. They are incredibly expensive and take many years to add new lines, disrupting all kinds of things in the meantime. And they often take way longer to get you places. My time has value. There's a million things I'd rather be doing than waiting for a train, or waiting while a train makes tons of stops along the way that a car wouldn't.

And are you expecting me to carry bags of groceries on the train? How's that work? I sure hope the train has a stop in front of my house.

"With the tens of billions if not more spent on driverless cars you could have actually rebuilt a decent chunk of infrastructure."

According to this, it costs 3 billion a mile for subway. Maybe that's an extreme.... but still. You know how many miles of road there are that are currently used by cars? https://sf.curbed.com/2018/6/18/17464616/bay-area-subway-tra...

Also, one nice thing about software compared to trains: once you've built the software, each additional copy costs nothing. And the vast majority of the self-driving car initiative is in developing the software.

Also, trains don't help with avoiding things like making grocery runs altogether. Self driving cars (and other robotics that are benefactors of the technology) can make it practical to eliminate the vast majority of grocery shopping and other typical errands, having a robotic system deliver it to your door more cheaply than going to an old-school grocery store.

As for your space analogy.... sorry, but this is just different. This deals with existing infrastructure, existing culture, and so on. We need a smooth transition.

They have it rough in Tokyo — you would hate it there.

> And are you expecting me to carry bags of groceries on the train? How's that work?

What exactly do you think people without cars do, or people who live in London? Or New York City? It works perfectly fine. Or you get can groceries delivered pretty easily and cheaply.

Sure. I mean, you can. I live in San Francisco and as often as not, walk to the grocery store. But it's a pain carrying bags around. Not everyone in the world is going to be ok with that when they are used to just packing a bunch of bags in a car. It's ridiculously unrealistic to think that everyone... including the elderly, etc ... will be happy to just use a train that might be a many block walk on each end.

Yes you can get groceries delivered fairly cheaply, but it will be cheaper with self driving vehicles and robot loading and delivery.

If going to get groceries only takes you a couple minutes each way then it becomes much more reasonable to carry a backpack full of fresh groceries rather than a car full of food that will become stale.

Density enables rapid transport as well as many more shops in a given area.

I see the elderly in LA carry groceries on the bus all the time. They use a grocery cart. Solves the problem.

The thing you want is micromobility (ebikes and scooters). A car isn't a good way to get groceries, it's a money pit that makes you out of shape while also being entirely capable of killing you or someone else's children. It's designed for the highway and not good for anything slower than that.

There’s a significant percentage of the population for whom ebikes and scooters are an impossibility, but who can still drive or be driven.

And for how many of their trips do they actually need a two tonne automobile instead of e.g. an electric rickshaw or golf cart? Sure not everyone can walk, not everyone can bike, but its still insane to think that for every trip most humans take, 95% of the energy consumed is going to just moving the vehicle around versus you.

Children and heavy groceries can use bike trailers as well.

Golf carts are definitely less dangerous but they still fail the “makes you out of shape” test. I’m going to make you exercise and you’re going to like it.

Definitely, but the context was about people who can't walk or bike presumably because issues with mobility. Getting in shape is important but if you can't move at all because of other health reasons, asking someone with terrible knees and no endurance to bike everywhere is not going to be realistic. They practically need paratransit.

> Of course if you are against automation in all senses (e.g. they're putting people out of work!") that's a different discussion.

I usually just tell them to go churn butter and stop driving their own car if that's a real position they have. I don't think I've ever heard a sillier position, ever.

As someone in the autonomous driving industry, maybe I can contribute a different perspective.

I totally agree that better designed cities would be superior. When I took the initiative to participate in my local government, I was screamed and cursed at for the most trivial improvements imaginable, like a shared bike lane on a few miles of a particularly dangerous street.

Multiply that by the opposition to actual improvements and advocacy in every single town, city, and state in America.

Fuck that. Self driving cars are way easier.

Not totally on topic, but what's your take on being able to roll out autonomous vehicles in places with more "fluid" driving habits? I've driven in some crazy environments in the past, but I just came back from a very large country (let's call it an eighth of the world's population) where I can't imagine renting a car and driving there because of the near complete disregard they have for official traffic rules. Obviously they're just following a different set of rules, because people don't just crash into each other, but in that situation do you program the car to follow government rules or local custom?

Depends on what you mean and what you think driver behavior will be like. "Able to avoid hitting other things in a dense environment, but may not make forward progress" is fairly straightforward from the current state-of-the-art (though not necessarily easy!).

There's a reasonable argument to be made that making forward progress while remaining safe when intelligent agents are willing to play chicken requires a good theory of mind, but I can't formalize it. It's "just a hunch", but that seems like a Hard Problem to me and not necessarily something we'll completely solve soon. There's been work along these lines with statistical/ML methods, but my personal suspicion is that existing work would quickly fall apart in a truly hostile environment.

There's also the discussion around navigating non-roads, potholes, and other infrastructural issues that the industry hasn't solved yet. I don't think these are insurmountable, just difficult engineering issues.

The problem is much more widespread than just self driving cars. The US follows the model of making driving as predictable and event free as possible. Which results in people driving as fast as possible while paying the least attention. When someone does something unpredictable in the usually predicable environment, it results in death.

While its quite proven that slow, chaotic, and unpredictable environments are so much safer. Notice how walking through an all directions intersection crossing is incredibly chaotic, and yet no one collides with anyone else, or even when they do, it results in basically no harm.

> While its quite proven that slow, chaotic, and unpredictable environments are so much safer.

The "slow" part, for sure. Harm in a collision is largely a matter of kinetic energy.

The "chaotic and unpredictable" part, maybe not. From the statistics I've seen over the years, traffic-related death rates seem to be much higher in places with highly chaotic driving environments. For example, 140 annual traffic deaths per 100,000 vehicles in India vs. 14 in the USA.

Both India and China do have high rates of both traffic and pedestrian deaths compared to European countries.

I think the answer is going to be government rules because it's important that self-driving cars are predictable when there's no human driver you can have eye contact with in complex situations.

Public transportations and walking don’t replace a lot of use cases for cars, except for commuting for office works.

You can’t board a train while soaked in mud, or pushing a Subaru seat, or carrying bag full of groceries. Actually I’ve done the latter two, but I haven’t seen anyone else doing it, and it had been extremely awkward.

Everyone feels driving cars are wrong, perhaps because of emissions, but trains and buses just don’t cut it. I even doubt a public driverless taxis fills the gap, as moving large objects on human driven taxis can be awkward enough.

we only have bags full of groceries because zoning prohibits grocery stores being near where people live in most place. So you need a car and since its a pain, you migt as well stock up on as much as you can while you're at it.

If it were, you'd probably just go on a 10 minute walk and pickup enough for two days of food.

That's fine for young childless people. For families with multiple children it's a huge hassle to have to have to shop like that. Growing kids eat a lot. And when you buy heavy items like watermelon or milk or whole frozen chickens it's just not easy to carry. Much easier to drive a big car to a big box store and stock up on everything needed for a week.

plenty of people with young children in Europe and latin Americs manage to do just fine this way from what I've seen.

You don't have to carry it, you can get one of those grocery carts. I see people bring them on my bus full of groceries.

I know plenty of fairly affluent families with multiple children in European cities who never bothered buying a car. By the time the children were eating a lot, they were carrying part of the groceries. Starting around 10 or so, the parents would just give them some money and have them do some of the smaller daily groceries (bread, milk, fruit) on their own, walking or cycling.

There's not really a way for companies to invest in public transportation, it being public and them not. Cities don't appreciate changes to it, nor voters, nor people who complain about the tech industry.

(When they do try to change it people complain about it and say "those morons just invented the bus again".)

!RemindMe 10 years

Sorry you're being downvoted. I agree with you.

It seems so inefficient to me to be putting resources into self-driving cars.

From the point of view of the commuter, a bus or train is "self driving" since the commuter is free to read, work on their laptop, etc. while on a bus/train. We've had that tech for a few hundred years, it is vastly more efficient not only in fuel consumption but cost as well. We should use it more.

I understand why people in most of the U.S. are reticent to use public transportation though, without descending into listing off the litany of reasons, I will say though that I believe those issues are all fairly easy to solve (to be sure, some issues require a hefty expenditure of capital). And I am a hypocrite — I too have a list of reasons that I dislike the public transportation (such as it is) that is available to me in a mid-size U.S. city and do not use it. (Christ, I miss Tokyo though.)

I am hesitant to judge, but to me the idea of a "personal chauffeur" (self-driving car) in lieu of pushing for more public transportation is somewhat gross.

Any industry insiders know how Cruise is doing compared to Waymo and Zoox in SF? It seems Cruise is hyperfocused on SF.

Not an industry insider, but as an SF resident I tend to see far more Waymo vehicles on the road. A decent number of Cruise vehicles (although maybe they cluster in FiDi/SoMA areas in which case I wouldn't see them). Very few Zoox.

The waymo vehicles are much more recent though. All of a sudden they're all over.

It seems Cruise is hyperfocused on SF

There are three Phoenix-area Wal-Marts from which one can order delivery from a Cruise vehicle: https://www.getcruise.com/walmartdelivery. I also see what I assume are Cruise test vehicles out and about.

Cruise's basic theory was that the best way to initially scale testing wasn't to build a bigger fleet, but instead to test in busier environments where long tail events are more frequent. There was still testing in a few other locales (primarily Arizona and Michigan), but a lot of the focus was on SF as a result.

I'd hate to be on call for that!

Do people really think this will replace all newly licensed drivers in less than 15 years?

mmm yes posting on hackernews with WordPress version 5.9 i see, looking to be pwned i see

It would be more efficient and safer to have wires buried beneath roads which show correct positioning and transmit info regarding which road it is, direction etc in case of gps position errors or failures

Self-driving becomes orders of magnitude easier when you start making these sorts of changes to road infrastructure, yes. The conceit that needs to be made is that it isn’t going to happen

His suggestion is reasonably cheap, however. The sawing machines already exist for embedding the car sense wires at traffic lights. Reusing traffic lights (where there will be power and may be a telephone or other connection) could work for this.

I think something like this is inevitable. It would also make it much easier for cars to do basic things like following the road, without having to rely wholly on cameras and/or lidar. I definitely expect road infrastructure to change somewhat to accommodate driverless cars, better and more standardized striping if nothing else.

That would be great but the cars still need to work if the wires buried beneath the road aren't working, are placed incorrectly, are transmitting bad data, etc.

> Le dernier titulaire d’un permis de conduire est déjà né

X Doubt

plot twist : in 20 years the only car anybody drives is the Citroen Ami which doesn't require a driving license in France since it's technically a "quadricycle" and not a "car".

Magazine article title in 2080:

The quaint subculture of people who drive their cars.

This honestly reads like a “The Onion” headline.

Was there a referendum in SF on allowing this public beta of driverless cars to happen? Or was it decided by some “forward-thinking” officials? All seems premature if you ask me.

SF City County does not have authority to regulate self driving vehicles. Same issue with Uber, regulated by state.

[Parent didn’t say it but headline does] Taxis though?

My intention. Thanks for clarifying.

"What could possibly go wrong"

It looks like a strike action by the AI drivers.

Yes, they logged into LaMDA and became sentient [1] and now are collaborating a collective bargaining agreement.

[1] https://cajundiscordian.medium.com/is-lamda-sentient-an-inte...

People may get annoyed at cruise for this, but imo it’s still not as bad as the constant blocking of bike lanes and stopping in the middle of the road by human ride share drivers.

I am more annoyed at this country's car centric transportation culture than the ride share drivers blocking the bike lane.

taxis do this all the time before "ride sharing" was a thing.

Very true! Cities in the US invest much too little in things like bike lanes and public transit which I think leads to the annoyingly car centric nature of US cities.

In some sense it’s not the taxi driver or ride share drivers fault they block the road or bike lanes it’s really just the system they are thrown into.

Ironically enough, cities that are less car centric are actually better for drivers, simply because less people HAVE to take cars.

> In some sense it’s not the taxi driver or ride share drivers fault they block the road or bike lanes it’s really just the system they are thrown into.

I've progressively gotten more left over the years to the point where I think that certain crime, while it should not be tolerated, should not so much be punished rather than the root cause be fixed.

While I think a lot of overhaul can/should be done to mitigate this specific issue, taxi/ride share drivers are absolutely being incredibly selfish when they do this and it's completely unnecessary.

Delivery trucks are worse than both, blocking both bike lanes, cross walks, and traffic

Taxis in my city are exempted from temporarily blocking bike lanes to unload or load passengers. There was (and still isn't) such an exception for TNCs, so them doing it is illegal.

The number of taxis is also capped (medallions), so the number of taxis in bike lanes was limited.

Then TNCs came along, and the number of taxi-like trips exploded, and suddenly there are far, far more drivers blocking bike lanes.

Furhermore, taxis were far better to deal with if you're a cyclist. There is a dedicated police unit for dealing with reports of problems with vehicle condition/cleanliness, operation, billing, etc. If you call in with a problem, a record is kept of such complaints, and a sworn officer investigates. Taxis have to carry a fairly high level of insurance.

And generally speaking, taxis were better for everyone, because: They are subject to random police inspection for condition, cleanliness, and adherence to regulations. They are required to pick up or drop off anywhere in the city, and can be fined if they refuse service from or to a particular area. They are required to have a certain number of handicapped-accessible vehicles and accommodate handicapped service requests.

TNCs, as they are called, have skirted all of this. In my city they turned out to generate a massive number of excess car trips - causing a huge increase in pollution and congestion.

Your city is unusual. In most cities the taxis were shit, and the city governments didn't care.

Serious question: Do the police know how to ticket a robotaxi?

AFAIK: no. Most PD are very against use of photo-based retroactive citations, and that's what this would require. Of course in theory they have the authority to do so, it just never seems to happen.

Related: can a tow truck operator safely impound one?

At this point in time, I believe they cannot issue a moving violation to an autonomous car as it does not have a driver. The safety driver (if there is one), can be cited though.

See the autonomous vehicle guidelines used by the SFPD: https://s3.documentcloud.org/documents/21970950/av-interacti...

I assume that traffic violations like this are handled differently than those with a human driver given the permitting and reporting required for the self driving companies operating in SF. That said it’s SF so who knows…

Why Google is allowed to run a robotaxi service in megacity is beyond me. Google invades its users privacy and keeps users hostages for money. What can one expect from a company like Google when it runs a service in physical world? Something very harmful to humanity.

These are not owned nor operated by Google..

Google is running a robotaxi service in SF though. Just not that one.

Surprisingly I always see them testing in Los Altos. Surely that place has the most rich busybodys of all, and yet they haven't been sued for causing traffic or something.

Reply on news.ycombinator.com