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.

"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.

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.

> 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.

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

>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.

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.

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

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

> 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.

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.

> 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

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.

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.

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.

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

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

It’s the “MVP” of city design. :)

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

What's a superblock in Houston gonna do?

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.

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.

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

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.

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.)

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.

> 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.

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...

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.

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

I mean refactoring production software is no easy feat either

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.

> $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-...

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

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

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 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.

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.

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.

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

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

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?

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.

You must be new to HN