pick one: driverless cars OR high-quality, human-scaled communities
October 6, 2018 2:52 PM   Subscribe

Safe, efficient self-driving cars could block walkable, livable communities. Ready or not, here they come.
posted by aniola (112 comments total) 20 users marked this as a favorite
 
I am still not convinced we'll have fully automated self-driving cars any time soon. The number of edge and corner cases in driving is still too high. I'd love to see one of these self-driving cars try to make a go of it in an urban environment during a winter Nor'easter. Let's see how well these things work when you can't see lane markings, when there's poor visibility, high winds, black ice, and other crap that will fuck with sensors and algorithms.
posted by SansPoint at 3:09 PM on October 6 [35 favorites]


Not seeing it as either or. To me, walkable communities seem more likely if you have a way of getting from your community, to work, to the store, to the airport when you need it. Then, driverless car is your best answer. In this scenario, I don't need to hunt for parking at the mall or worry about my car at the airport. And the avenues between the communities can be simplified to make the vehicles operate more safely. "Lumps" of neighborhood connected by a tissue of autonomous roads seem ideal to me.
posted by SPrintF at 3:26 PM on October 6 [12 favorites]


Wow. Even our futurists are broken these days.
posted by Wizzle at 3:26 PM on October 6 [4 favorites]


We've had the technology to solve modern transit problems for, what, 150 years? Every single complaint people have about public transit could be addressed if municipal transit agencies were given adequate funding. Every one of them, except that you won't be able to go door-to-door. But the expense, inconvenience, lack of personal safety, uncleanliness, and everything else people associate with public transit? Those are all issues that could be solved, but won't, because the people with money are going after the cooler-sounding novelty solution instead of the one that will support far more people.

And I can guarantee that the novelty of a self-driving car will wear off. Your commute isn't going to stop being your commute just because you can read and work in your autonomous car. Man, you can already read and work on the damn bus or subway, and people do.

This is the first I've heard of "autonomous-only lanes," even though the author is critical of the idea, given how little funding roads get nowadays. How about special "bus-only" lanes? If you want to increase efficiency, increase the number of people being moved per vehicle.

As always, tons of money pours into the future technology that may eventually solve our problems, because the existing technology isn't exciting enough. It's like the ol' Segway thing: you can have gyroscopes and complex electronics that ensure your 2-wheeled vehicle stays upright, or you could install a third wheel. But oh look, the gear differential was designed to produce a musical harmony.
posted by shapes that haunt the dusk at 3:30 PM on October 6 [82 favorites]


Am I missing something, or is this argument absolute nonsense? It seems to run:

1) Autonomous cars will make the streets safer for other road users. (I guess this is true long-term, although I'm not certain about how long-term)
2) This will make other road users so reckless as to cause gridlock (I don't know why this will happen. It's not like they won't almost all also be users of the network of autonomous cars, if only in the form of Uber or Lyft or whatever).
3) This would probably be controlled through some regulation of other road users (I guess? Although I thought jaywalking was already illegal in lots of places in the US? I am finding it hard to imagine how he thinks people are going to be using the roads that isn't already prohibited in some way)
4) Having regulations like that goes against the idealised spirit of the public space. (Well, I guess, but that's life outside an anarchist collective on its own private island)
5) Therefore, for some reason unexplained, autonomous vehicles will prevent the creation of more liveable communities.

Am I being really dumb? It has happened before...
posted by howfar at 3:37 PM on October 6 [14 favorites]


This is a terribly silly argument that makes all kinds of unsupported assumptions about the state of future self-driving cars and traffic systems.
posted by Foci for Analysis at 3:42 PM on October 6 [6 favorites]


I still want to know how autonomous cars are going to solve the current issue of significant congestion problems in major cities. You replace a human driven car with an autonomous one and you still have the same result as far as the number of cars on the road goes. The only thing I’ve heard people say in regard to this is “people will carpool” and there’s no information at all to back that up.

Busses and light rail significantly reduce congestion. I’ve posted multiple times in these topics the statistics that Portland’s TriMet service has released in regard to how many cars the public transit system here gets off the road, and the reduction in congestion that comes with that. As more and more people move here, an increase of funding for our public transit is vital if we are to continue being a public city where one can bike, walk, and transit around town in an efficient manner. We have strides to be made, we just need the backing.
posted by gucci mane at 3:45 PM on October 6 [10 favorites]


Self driving cars are the personal jet pack of this moment
posted by The Whelk at 3:47 PM on October 6 [29 favorites]


I still have a hard time getting my head around a driverless car that can navigate around rural areas and rough terrain (because we stubbornly refuse to hold an actual Infrastructure Week, and even paved roads are crap in my state), and severe weather, and remote-but-numerous areas where there's zero internet or data reception and you pretty much need a paper map to get around.

It's almost like the debate is taking place among people who have a hard time imagining life outside of cities. Or whose idea of "country" is "the interstate between the suburbs and that place where we pick apples."
posted by witchen at 3:48 PM on October 6 [9 favorites]


This.... article is nonsense. Weird suppositions and even from there, weird conclusions that those weird suppositions don't even really support.
posted by tclark at 3:49 PM on October 6 [4 favorites]


It's like Apple devices removing all your downloads and sending them to the cloud, with the assumption that we can stream whatever we need whenever we need it. Everything is designed for people who have wifi access 24/7. But that's not how most of us live.
posted by witchen at 3:50 PM on October 6 [8 favorites]


The other thing I don't get is when people think this is going to happen. Oh, self-driving cars are around the corner? What about for the family in Stockton who live on minimum wage? Everyone always talks about this as if every car on the road will be driving itself, but the future is never equally distributed. I can barely afford the bus, so why are people thinking I'll have an autonomous vehicle or be taking driverless Uber everywhere?

And let's say the technology does eventually catch up to its constantly-extended timeline ("any day now"). Is it going to be implemented nation-wide? Worldwide? All at once? How's that going to work? If San Francisco is the first city to become autonomous-only, what happens when someone drives in from Tracy? Which, incidentally, some transit workers do every day, because the transit agencies are so underfunded that they can't pay a living wage in SF.

It really seems like the people making this stuff happen aren't in touch with the realities of life for the majority of the country and the world.

I'm not anti-technology. But I am opposed to a technology that just swaps out elements of the hugely imperfect system we have with something shinier. I'm sure autonomous cars will eventually be better drivers than I'll ever be, and I'm sure some lives will be saved or improved, but not nearly as many as there would be if we'd look past the novelty of new tech and focus on what people and communities actually need, instead of what we want them to have.
posted by shapes that haunt the dusk at 3:55 PM on October 6 [15 favorites]


So, this is pretty much the research world that I inhabit, and yeah, the article's arguments are fundamentally broken. Just to start with, even if we get high L4 or even real L5 autonomy tomorrow, which is what the author seems to be thinking is just around the corner, we'd still looking at decades - I've seen estimates on the order of thirty years of mixed autonomous / semi-autonomous / manual roadways, and that's just with regard to the passenger vehicle fleet. In the absence of the Autonomous Vehicle Fairy, this doesn't strike me as a likely future.

What we're much more likely to get if you look at (1) the current state of the research and technology here and (2) what automakers are actually thinking about selling anytime soon is something closer to L1 (driver assistance) and L2 (limited autonomy). Real autonomy - L3 and up - in urban environments is absolute hell from a problem-solving standpoint. It's pretty easy, as far as these things go, to do autonomous driving in mixed roadway environments on the highway, because it's access-limited and you can make a lot of assumptions about the environment. Urban roadways? With anything that looks like either current-generation or next-generation technology that actually exists? Hell no. Too dangerous, too complicated.

What likely is coming for urban roadways (really, for everywhere) is a lot more driver assistance (L1 autonomy). Much too much of what makes the roadway unsafe for everyone - cyclists, pedestrians, drivers - is human error behind the wheel. All of the human actors here can make errors, but the one in putative control of two tons of machine is the one we can easily augment. So, let's make the car capable of helping the driver, and avoiding other road users and hazards. For example, can we tell the driver where they should pay attention? If they don't seem to respond, should the car take over, even if it'll piss the driver off, to maximize safety? We're asking those kind of questions. As a bonus, it's nowhere near as hard as full L4/L5 autonomy, and it's getting cheaper and easier to, say, have the car looking out for the kid who doesn't look and runs into the road and will put on the brakes if the driver doesn't start to. In much the same way that all new cars have backup cameras (after, I believe the 2017 or 2018 model year), I'd expect to see L1 start being a lot more prevalent a few years down the line.

So: while various automakers (cough-Tesla-cough) really want you to think that you'll be able to summon an autonomous car to pick you up in a year, that's a long, long way off, and the future that's actually on route is both saner, more useful and more realistic.
posted by Making You Bored For Science at 4:12 PM on October 6 [30 favorites]


Will self-driving cars be held to high environmental/efficiency/safety/sustainability/labor standards? If companies are planning to use them because of likely higher profits, shouldn't they be able to handle regulations that were 'too expensive' before?
posted by amtho at 4:13 PM on October 6 [2 favorites]


Gucci mane: The trick appears to be autonomous cars flow more freely. The most obvious example: No more slowing down to look at accidents. But even a few autonomous cars in a flow with proportionally human drivers appears to result in more freely flowing traffic.
posted by aurelian at 4:17 PM on October 6 [3 favorites]


SPrintF: ""Lumps" of neighborhood connected by a tissue of autonomous roads seem ideal to me."

Sounds like hell if you are poor.

Making You Bored For Science: " So, let's make the car capable of helping the driver, and avoiding other road users and hazards."

Man I'd love to see semi autonomous cars that actively enforce (by taking control) the legal requirement (here anyways) of giving cyclists 1m of space when passing.
posted by Mitheral at 4:20 PM on October 6 [13 favorites]


The issue of autonomous cars reminds me of issues of police violence. Bear with me for a second here. One of the best critiques of body cameras for police that I read was, "Be very suspicious of 'solutions' that just give more resources to the police."

Similarly, a solution to the problem of cars (too many of them, too much common space devoted to them, and they're too deadly) that just invests more in cars seems extremely foolish - especially since (as shapes that haunt the dusk helpfully points out) that we have had the technology to solve the problem since before cars existed.
posted by entropone at 4:25 PM on October 6 [13 favorites]


Also, autonomous cars are only autonomous of a driver. They can all talk to each other, and to the ground (and satellites). There are a set of variable speed limit signs in Seattle (that no one observes), with the idea of slowing down how quickly flow arrives to back up. Autonomous cars would follow the signs/limits.

But because of the communication between vehicles, they can hypothetically reduce following distance, meaning you can pack more vehicles in a given set of pavement.
posted by aurelian at 4:26 PM on October 6 [3 favorites]


I also don't understand how more proponents of car culture don't realize that investing in efficient transit options benefits them by removing some of the unnecessary cars from the road. It just... it baffles.
posted by entropone at 4:27 PM on October 6 [22 favorites]


Man I'd love to see semi autonomous cars that actively enforce (by taking control) the legal requirement (here anyways) of giving cyclists 1m of space when passing.

That would be great. It's a total pipe dream (can you imagine the howls of outrage at the mere suggestion?) but I'd also love to see cars be forced to follow speed limits in urbanized areas. The street next to me has a nice low speed limit, but the way it was designed encourages higher speeds so everyone goes faster, including me when I am driving.

But ignoring the reality that they are a long way from reality, I see the addition of autonomous vehicles (once the kinks are worked out) being only to the good for neighborhood safety and walkability -- with the caveat that they are not a replacement for mass transit, for which resources should be increased.
posted by Dip Flash at 4:32 PM on October 6 [1 favorite]


Oh, self-driving cars are around the corner?

At about 20:30 in this Deutsche Welle documentary published in 2016—“Robotics - Impacting the workplace” —see a German shipping container seaport where the containers are all hauled around by autonomous vehicles, like flatbed trucks with no cab. Self-driving cars are not some sort of vaporware.

It's almost like the debate is taking place among people who have a hard time imagining life outside of cities.

Strangely enough, the first OP link was actually written by a professor at the University of Nebraska. It's really bizarre, though... it basically boils down to an argument that pedestrians and bicyclists are so inherently unable to share the road that absent the possibility of death they'll bring all transportation infrastructure everywhere to a standstill with their feckless lollygagging. They'll probably uncontrollably void their bowels all over the place too like horses.

But the author asserts this will primarily affect cities, so whatever. People who live in cities can do whatever weird stuff to each other they collectively want to. I, personally, will continue to snag passing cars, autonomous or otherwise, to pull myself around on my skateboard in between bouts of time travel.
posted by XMLicious at 4:35 PM on October 6 [12 favorites]


autonomous cars aren't designed to make things safer for all people. they're designed for the maximal efficiency of car users first and foremost. And that's dangerous. And it's not like the tech sector that's pushing this exactly has a reputation for getting this shit right the first time.
posted by entropone at 4:36 PM on October 6 [11 favorites]


But because of the communication between vehicles, they can hypothetically reduce following distance, meaning you can pack more vehicles in a given set of pavement.

Following distance has to do with both reaction speed and braking distance, right? I'm sure you could tighten it up in terms of reaction time, but braking distance is going to be an ever-present, variable factor depending on the vehicle load and road surface.

Even with reduced following distance, or even if cars were able to drive bumper-to-bumper at speed, you'd still see far less efficiency than if you had more high-capacity vehicles like buses. The population is continuing to grow, especially in cities, and any gain from having cars packed tighter will be relatively short-lived, like adding a lane to the highway. Part of the reason we have congestion in the first place is that cities keep outgrowing existing transit options.
posted by shapes that haunt the dusk at 4:37 PM on October 6 [5 favorites]


Sanspoint, you make the best point out of the gate (ironic, that. *). All we do in NE is bitch about how folks forget to drive in bad weather or else how it sucks being stuck on stalled public trans in bad weather. God help us if we create cars that either only go manual or only lock up in bad weather.

*epoNOTsterical? I'm sure mefi has come up with a better opposite for that word
posted by es_de_bah at 4:38 PM on October 6


Sure, I can hear the howls.

I can also see the insurance companies raising your premium by 5x or 6x if you don't.

That's what's going to drive (ahem) the transition to autonomous vehicles. Only the 1% will be able to afford the sky high insurance of human-driven cars.
posted by aurelian at 4:38 PM on October 6 [10 favorites]


I'm not even trying to argue the efficiency of cars over buses - I'll take that as a given. (I rode buses - in LA - until I was 23.) I'm trying to address the issue of, "How is this an improvement over what we have now?" and present what I've heard.

And I also don't agree with everything. I do find it fascinating in a Dilbertesque engineering way, though.

(Except for the point about insurance companies. Really. Actuarially, they're going to go that way.)
posted by aurelian at 4:49 PM on October 6 [1 favorite]


I appreciate you answering my question aurelian :3

I really like shapes that haunt the dusk‘s comment above about the adoption rate of autonomous cars being unequal depending on circumstances. Smartphones are so ubiquitous now, but my family wasn’t able to afford them until the iPhone 4S was released, which was four years later. I imagine a car would take a much longer time for people to begin using. A lot of people still drive cars that are hand me downs. I for one have a car that was a hand me down, and it’s a 1994 Miata. I will probably never in my life be able to afford a new car.
posted by gucci mane at 5:05 PM on October 6 [2 favorites]


Shapes: It looks like you get 10 feet for every tenth of a second of reaction time at 70 mph. But the reaction times would probably go to milliseconds.

Yes, there's a limit to what you'd pick up in following space, but I not sure it matters given the scale of the difference of reaction times. Also, the car slowing ahead would tell your car it's slowing, so there wouldn't be any delay in assessing the situation in the first place.

I'm not disagreeing there's a limit - of course there is - but it's at least one more car length. So you could double passenger traffic.

No, what I'm worried about are commercial trucks. Yeesh.
posted by aurelian at 5:17 PM on October 6 [1 favorite]


First driver-less car that hits a busload of kids = last driver-less car ever. No insurance company will write those policies.
posted by Freedomboy at 5:22 PM on October 6 [1 favorite]


Freedomboy - not if legislators immunize driverless car owners and builders from punitive damages for forseeable low probability catastrophes as long as the technology is overall safer. Which is the case with every technology where large scale casualty events occur infrequently.
posted by MattD at 5:33 PM on October 6 [2 favorites]


In cities one huge savings is going to be parking. Even if everyone owns their own autonomous car, nobody ever needs to park. The car drives you where you need to go, then sends itself back home and comes back to get you. Even if it does need to park, it can go out to some higher density autonomous only parkade/charger rather than using street parking. Imagine city centers without any on-street parking at all; not only do you get more open road space (or far more viable bike lanes), but you don't have people trying to park blocking lanes anymore, which can be amazingly disruptive when things are busy.
posted by Grimgrin at 5:34 PM on October 6 [2 favorites]


The car drives you where you need to go, then sends itself back home and comes back to get you.

You’ve just increased traffic congestion by 100%.
posted by Automocar at 5:37 PM on October 6 [30 favorites]


Freedomboy: Mostly depends on if it's greater or lesser than the current 128 school bus related deaths per year we currently have.
posted by aurelian at 5:40 PM on October 6 [3 favorites]


Grimgrin: More likely, Uber (or something like Uber) owns most of the cars, and they park in regional facilities until they're called up again in rotation.

(googles)

Yeah, GM is starting a service, Maven Gig, where they cut out Uber as the middleman.
posted by aurelian at 5:46 PM on October 6


Token ring, but for cars. Collisions being bad, and all.
posted by aurelian at 5:47 PM on October 6 [3 favorites]


The reaction times absolutely go to milliseconds, and cases where you can use vehicle-to-vehicle communication to facilitate a safe stop are good, but aren't the really scary ones. The scary ones are the unanticipated events that the driver (or the autonomous vehicle) needs to respond to now. One of the big meta-analyses on this gives a figure of 1.3s for reaction time here - that's not time to come to a complete stop, which is pretty cleanly a physics problem (how fast are you going, how hard did you engage the brakes, and what's the road surface like), it's how long it took you to start stomping the brakes. If automation can cut that down, say, by having the vehicle notice a hazard faster than the human can, then yes, you can accept a closer following distance because you'll still be able to slow the vehicle in time.
posted by Making You Bored For Science at 5:54 PM on October 6 [2 favorites]


Automocar: I was under the impression (and I could be mistaken as I haven't done a lot of research) that maintaining a steady flow was more important for reducing congestion than the number of cars on the road. I.E. Jams and congestion are caused by stops and disruptions propagating through traffic flows rather than by there being too many cars on the road. Which is why I thought this would be an overall good thing.

I will cop to the fact that I'm in a situation where I walk everywhere, and find cars parked everywhere annoying for a variety of reasons. So mentally for me it was "what if the cars just fucked off to wherever it is cars go when they're not left littering the damn street".

Aurelian: "We understand you like disrupting business models, so we're certain you'll appreciate the irony of what we're about to do to you."
posted by Grimgrin at 5:58 PM on October 6 [1 favorite]


having the vehicle notice a hazard faster than the human can

For certain classes of hazard, this is already real. I've driven a GM SuperCruise-equipped car, and it warned me of slowing traffic ahead before I could perceive the reduced distance myself.
posted by aramaic at 6:01 PM on October 6 [1 favorite]


autonomous trucks are scary because of the collateral damage. if people aren't on the highways and byways then gas stations, restaurants, motels, truck stops and all businesses that rely on being "in between" destinations are at risk - massive numbers of jobs lost. technologically long haul trucking seems an easier problem to solve.
posted by askmehow at 6:01 PM on October 6 [2 favorites]


A solution looking for an problem, if ever there was one.
posted by 4ster at 6:02 PM on October 6 [5 favorites]


Having sat through a talk about this exact topic this week ("spontaneous" traffic jams and how to end them): yep, Grimgrin, you're right - for traffic jams without an obvious cause, it's the sum total of minor slowing behavior slowing everything up. The neat thing is that we can probably ameliorate them if we monitor following and leading distance and optimize that, and it doesn't take a big vehicle-to-vehicle network... think of it as Advanced Cruise Control on steroids.

aramaic - that's really cool. I haven't gotten to play with Super Cruise in person (and I've got a lot of reservations about GM's autonomous driving work), but that's a case where we're usually bad at detecting small changes, and augmenting the driver like that is exactly what good automation, in my opinion, should be doing.
posted by Making You Bored For Science at 6:04 PM on October 6 [1 favorite]


I can also see the insurance companies raising your premium by 5x or 6x if you don't.

That's what's going to drive (ahem) the transition to autonomous vehicles. Only the 1% will be able to afford the sky high insurance of human-driven cars.


I think the reality is that a lot of people would just lose out. A lot of people can barely afford repairs on the cars they own, let alone a new late-model car. The biggest impacts would be to those who can't afford a new car, but who still rely on a car to get to work (because in many places there is simply no other way to get around). A lot of other people would see their commutes doubling or tripling on public transit (unless maybe we put some money into public transit). Will autonomous cars go in hand with major changes that address rising costs of living that force people to live further away from their places of employment? Will they go in hand with more funding for alternative modes of transportation?

This is how the future is unevenly distributed. Any huge change like this will affect different groups of people in dramatically different ways, and it doesn't seem like that gets as much attention as it should. gucci mane's phone example is great: there are still people who have never owned a smartphone because they've never been in a situation where it made economic sense to get one (I only got my first smartphone 2 years ago, and even then it was an economic stretch). Why would we expect a different situation with things that cost at least 40x as much?
posted by shapes that haunt the dusk at 6:05 PM on October 6 [5 favorites]


Lumps" of neighborhood connected by a tissue of autonomous roads seem ideal to me."

So.... trains?
posted by pompomtom at 6:26 PM on October 6 [12 favorites]


Am I missing something, or is this argument absolute nonsense? It seems to run:

The version that makes more sense is:
1.) it turns out that it is impractical to design and program autonomous cars that are safe around cyclists and pedestrians
2.) this is "solved" by keeping cyclists and pedestrians off the streets as much as possible

I mean - that's sort of what happened with the introduction of cars in the first place.

This article is kind of confused about what argument it's making.
posted by atoxyl at 6:31 PM on October 6 [3 favorites]


The article tangentially mentions that

Separated bike lanes are one possible way to prevent conflicts between cyclists and vehicles.

and this sounds like it would both solve the Jevons-y problem that the article poses and improve safety. Separated lanes for all!
posted by Jpfed at 6:37 PM on October 6


Grimgrin: Yeah, I just don't see Uber taking the Netflix solution and building their own cars.

But what's funny is that Netflix and Uber clearly had similar models: Netflix using DVDs until streaming was possible, Uber abusing drivers until autonomous vehicles are available. I guess the bet was, none of the car manufacturers would get in the rideshare business. Oops. Because Netflix really did add value when it had every DVD not at Scarecrow. That kind of consolidation mattered. I don't see how Uber adds any value, though, especially with their PR troubles.
posted by aurelian at 6:46 PM on October 6 [1 favorite]


Yeah, the key to the argument is not really emphasised very well in the argument itself, but the point is that our suburban streets are designed under the assumption that pedestrians stay off it because a car will hit you if you wander onto the road. People wander onto the road all the time, kids play on the road, it's a useful bit of public space that we normally avoid unless we're pretty confident it's quiet.

So let's assume that autonomous vehicles can hit that theoretical L5 autonomy, and they don't hit kids running into traffic. The obvious second-order effect is that more kids and pets will run into traffic because running onto the road with a car coming isn't necessarily going to be an accident or even a problem. People can wander onto the road as necessary without fear of getting hit, and use that public space more effectively. But that means that autonomous cars can't ever guarantee they're safe until they get out into freeways where pedestrians aren't allowed, so they're going to have to always travel at a speed that means they can stop.

That highlights a tension in road usage that really doesn't have anything to do with autonomous vehicles: a road can be for cars, or it can be for people, but it can't be meaningfully shared without fatally compromising it for both. And the problem with transit is that we used to have people-focused transit plans (at the turn of the 19th century, basically every major city in the world had an extensive streetcar network) and modernists and car manufacturers convinced the world to replace it with vehicle-focused transit (as shown in the classic documentary Who Framed Roger Rabbit). Autonomous vehicles aren't going to solve this because vehicle-focused transit has fundamental problems with space, efficiency and access (not to mention the emissions!)

(I also question whether an autonomous car depot is really as exciting an idea as people are thinking. Part of the reason why people like cars is that it's your car, and all the mess inside it is yours. You don't want to order a car and have someone else's mess arrive, and that's basically guaranteed.)
posted by Merus at 6:50 PM on October 6 [12 favorites]


I think the reality is that a lot of people would just lose out

Well, this is the US we're mostly talking about, so of course some groups of people will lose out. It's sorta the foundational premise of the country, and frankly to many voters the fact that someone's going to lose out makes the technology more appealing not less. I actually think this is one of the reasons the technology will take off.

Will they go in hand with more funding for alternative modes of transportation?

Of course not. It will actually result in reduced funding for alternative modes of transportation, because the answer to all transit-related problems will be:
1) just buy a self-driving car, you loser ...or...
2) just use a self-driving car service you ultra-loser
posted by aramaic at 6:52 PM on October 6 [3 favorites]


I still want to know how autonomous cars are going to solve the current issue of significant congestion problems in major cities. You replace a human driven car with an autonomous one and you still have the same result as far as the number of cars on the road goes.

There will still be traffic, but autonomous cars will allow drivers to be more productive in their cars while commuting to work in traffic, thereby lengthening the period of the day we're all expected to be available (Oh, you thought those hours in traffic would lessen your work day?)

I feel like the more enthusiastic someone is about the potential for self-driving cars to reshape our society, the more I think they're keeping the quiet parts of their vision to themselves.
posted by RonButNotStupid at 6:57 PM on October 6 [2 favorites]


At about 20:30 in this Deutsche Welle documentary published in 2016—“Robotics - Impacting the workplace” —see a German shipping container seaport where the containers are all hauled around by autonomous vehicles, like flatbed trucks with no cab. Self-driving cars are not some sort of vaporware.

This is a controlled environment, exclusively populated by trained personnel - in real neighborhoods we have cyclists, five-year-old kids chasing after stray sportsballs, and adults sensibly crossing quiet streets away from ped x-ings.
posted by sebastienbailard at 7:01 PM on October 6 [2 favorites]


aurelian: "That's what's going to drive (ahem) the transition to autonomous vehicles. Only the 1% will be able to afford the sky high insurance of human-driven cars."

Insurance doesn't work like that; it is self funding not punitive. Even if we replace 50% of the human driven cars with autonomous vehicles the rates on human driven vehicles won't go up (with the big assumption that safer drivers don't disproportionately choose autonomous vehicles). Heck they might even go down if autonomous vehicles can avoid accidents that human drivers cause. Autonomous vehicles might have lower rates than human driven cars but the rates on human driven cars shouldn't rise faster than they would without autonomous vehicles on the road.
posted by Mitheral at 7:27 PM on October 6 [3 favorites]


Autonomous cars are incompatible with walkable, livable neighborhoods but not for any exciting new reasons, just for the same reasons that regular cars have destroyed walkable, livable neighboorhoods nearly everywhere.

If the transportation of the future is personal cars, electric or not, autonomous or not, we blew it.
posted by vibratory manner of working at 7:32 PM on October 6 [12 favorites]


Traffic jams and traffic congestion are two different things, though. Traffic isn’t bad in cities because people are slowing down erratically, as on a highway—it’s bad because there’s a finite amount of space in which to drive cars. Autonomous cars solve highway traffic, in theory, but do nothing to ameliorate traffic congestion in cities, and in fact have the potential to make it a lot worse.
posted by Automocar at 7:33 PM on October 6 [10 favorites]


XMLicious: "see a German shipping container seaport where the containers are all hauled around by autonomous vehicles, like flatbed trucks with no cab. Self-driving cars are not some sort of vaporware."

This sort of thing is basically just a scaled up version of pick and place electronics assembly; the engineering hurdles for such a project are pretty straight forward and there isn't much in the way of uncontrolled events to worry about. And even then speeds are relatively low and having a vehicle just stop when there is a problem is a much more acceptable solution then in the case of open streets. We probably could have had such automated ports 20 years ago if we'd cared to.
posted by Mitheral at 7:34 PM on October 6


"That highlights a tension in road usage that really doesn't have anything to do with autonomous vehicles: a road can be for cars, or it can be for people, but it can't be meaningfully shared without fatally compromising it for both."

That's generally true, but it's certainly not absolutely true. See squares in Europe that are shared by bikes, people, cars, and streetcars. Also the street I live on is frequently traveled on by humans safely; cars move slowly, and people move to the sidewalks when a car approaches.
posted by el io at 7:35 PM on October 6


I am still not convinced we'll have fully automated self-driving cars any time soon. The number of edge and corner cases in driving is still too high.

I totally agree. I still think it's probably at least twenty years of working out kinks and dealing with various accidents before they get to that point. So hopefully they figure out by the time I can no longer drive, but I'm not gonna plan that I won't be driving within the next few years either.
posted by jenfullmoon at 7:40 PM on October 6


> Let's see how well these things work when you can't see lane markings, when there's poor visibility, high winds, black ice, and other crap that will fuck with sensors and algorithms.

Oh, well we have none of those things in Silicon Valley. You'll just have to get used to living in rural New York City I guess.
posted by pwnguin at 7:51 PM on October 6 [1 favorite]


aniola: "Ready or not, here they come."

I can see how this can work for the cases where you go from reserved parking spot to reserved parking spot but how the heck does a vehicle without driver inputs handle ad-hoc parking or navigation. For example event parking where the parking "lot" is some big field with someone manually directing cars into rows. Or construction sites. Or handle directions from a flag person/police officer. Or pick up your kids when you drive by them coming home from school. Or pulling over to a safe spot on the shoulder if you blow a tire, hit a deer or have other mechanical difficulty.

Sure it can auto park or even automatically align to a trailer but how do you choose which trailer? How do you direct it to stop along the curb next to your buddy when you are picking them up from the airport?

How does it handle passing that horse and buggy or slow moving farm implement on a two lane rural highway where safely passing might involve crossing a double yellow. Or heck someone broken down in your lane on such a highway.
posted by Mitheral at 8:05 PM on October 6


We've had the technology to solve modern transit problems for, what, 150 years? Every single complaint people have about public transit could be addressed if municipal transit agencies were given adequate funding.

Mmmm... no. Why?

Every one of them, except that you won't be able to go door-to-door. But the expense, inconvenience, lack of personal safety, uncleanliness, and everything else people associate with public transit? Those are all issues that could be solved, but won't, because the people with money are going after the cooler-sounding novelty solution instead of the one that will support far more people.

If you're going to wave away the appeal/desire/necessity of "door-to-door" transportation, you simply are not gong to solve any problems, and you simply are not going to be taken seriously. "If only we adequately funded transit agencies..." Yeah, if only. If only we funded a lot of things

And I can guarantee that the novelty of a self-driving car will wear off. Your commute isn't going to stop being your commute just because you can read and work in your autonomous car. Man, you can already read and work on the damn bus or subway, and people do.

What in the world makes you think autonomous cars will be adopted because of novelty? I can think of some very appealing reasons I'd love to have such a contraption, none of which have anything to do with being the cool kid in class with the newest sneakers.


This is the first I've heard of "autonomous-only lanes," even though the author is critical of the idea, given how little funding roads get nowadays. How about special "bus-only" lanes? If you want to increase efficiency, increase the number of people being moved per vehicle.


These things exist already.

The other thing I don't get is when people think this is going to happen. Oh, self-driving cars are around the corner? What about for the family in Stockton who live on minimum wage? Everyone always talks about this as if every car on the road will be driving itself, but the future is never equally distributed. I can barely afford the bus, so why are people thinking I'll have an autonomous vehicle or be taking driverless Uber everywhere?

Interesting question, but not really relevant. If you can barely afford the bus anywhere in the world, you're at a disadvantage. Autonomous cars not only are not a solution to poverty, they're not often touted as one. If you're going to propose that "if we all can't have something, then nobody can", then I'm probably not the only one who'll be glad you aren't making the rules.

autonomous cars aren't designed to make things safer for all people. they're designed for the maximal efficiency of car users first and foremost. And that's dangerous. And it's not like the tech sector that's pushing this exactly has a reputation for getting this shit right the first time.

Really? This doesn't even make sense. Designing "for the maximal efficiency of car users first and foremost", if nothing else, makes things safer for all people. Unless you're proposing that "maximal efficiency of car users first and foremost" means turning pedestrians and bicyclists into paving material or something, with absolutely no consequence. Yes, that would, um, be dangerous. But you need serious citiation.

This sort of thing is basically just a scaled up version of pick and place electronics assembly;

Sure, but all transportation is basically scaled up pick and place assembly, as far as autonomous vehicles are concerned.

As for the article, well, it seems kind of half baked, no?
posted by 2N2222 at 8:47 PM on October 6 [3 favorites]


I work in the tech industry, and I do machine learning. Don't get into a self-driving car for at least the next decade. There is literally no reason to trust any company peddling self-driving technology, because it is mostly marketing hype and is way less safe for drivers, pedestrians, and cyclists than human-driven transit in real-world conditions. Humans suck at software. We're really bad at building it as individuals, and as organizations we become even worse at it. If you want to reduce traffic fatalities, enforce speed limits more uniformly, move to traffic circles instead of stop lights for most intersections, and vote for municipal government candidates that champion light rail. That would make roads safer and less congested, without having to expose people to the risk of death due to a bad software update or unpredictable model behavior.

Seriously, stay the fuck away from cars that claim to be self-driving. Don't trust companies that claim to sell panacea technology. That has literally never worked out for anyone.
posted by cirgue at 8:59 PM on October 6 [19 favorites]


If automation can cut that down, say, by having the vehicle notice a hazard faster than the human can, then yes, you can accept a closer following distance because you'll still be able to slow the vehicle in time.

The problem with that is that human drivers already follow much more closely than their reflexes or conditions allow. It's more likely that autonomous cars would enforce a more consistent following distance, not a closer one. Which, sure, I really don't give a shit what drivers do on controlled-access highways if they're narrow enough to be crossed in a reasonable time by pedestrians, with plenty of safe and convenient crossings. (If they're not? Narrow them. Rip them out. Stop putting uncrossable canyons in cities.)

Hell, we've got the tech to automate
following speed limits already. I'll believe car manufacturers' goal is safety when they start doing that, and no sooner. Cities could make it easier on them by adopting uniform speed limits. Survivable ones, none of this 30mph crap.
posted by asperity at 9:00 PM on October 6 [1 favorite]


This is a controlled environment, exclusively populated by trained personnel - in real neighborhoods we have cyclists, five-year-old kids chasing after stray sportsballs, and adults sensibly crossing quiet streets away from ped x-ings.

Yeah so none of that is a reason to still act like self-driving cars are some inventor's fantasy spoken only of by charlatans who promise they'll exist "any day now" in a "constantly-extended timeline." It's the people who held forth on imaginary reasons that nature abhors autonomous vehicles, and thus they'll never ever be possible, who have been continuously moving the goalposts over the last dozen-years-plus and redefining "autonomous" to grasp after some claim that they're impossible or don't already exist.

No one claimed that by 2016 everyone and her dog would have Willy Wonka's magic elevator in car form, but then merely delivered things like the driverless-vehicle-filled container port. It's the other way around—people on MeFi and elsewhere have constantly declared impossible technology which in many cases already existed and they're the ones revising their timelines.

There are technical and legal hurdles to fulfilling specific goals on shared roads in the United States, some of which may never be overcome, but acting like that means autonomous vehicles don't exist is like acting as though, because we don't all have personal flying cars, heavier-than-air flight is impossible.
posted by XMLicious at 9:04 PM on October 6 [1 favorite]


It's more likely that autonomous cars would enforce a more consistent following distance, not a closer one.

That would be...great?

I’ve been driving for almost 20 years now, and I still do not know 100% what the distance between me to the next car is if it suddenly crashes and I need to stop immediately. I know what a safe distance if the car slows or brakes gently, because that’s most of what I encounter, but I don’t know how far I need to be if I slam on the brakes because I never slam on the brakes unless there’s something immediately in front of me.

I would love it if the car made the same binging noise it does for seatbelts or lane shifts and said “unsafe following distance at your speed”. That would be super exciting and helpful! I just rented a car this week and have been marveling at the lane shift protection.

I think maybe part of the problem is we assume that tech improvements must Solve The Problem, rather than just make it a little bit better.
posted by corb at 9:21 PM on October 6 [3 favorites]


I am still not convinced we'll have fully automated self-driving cars any time soon. The number of edge and corner cases in driving is still too high. I'd love to see one of these self-driving cars try to make a go of it in an urban environment during a winter Nor'easter.--SansPoint

Well, they are going live before the end of the year, so maybe that will convince you. You are not convinced because you are setting too high a bar. It isn't all or nothing. They most definitely won't be deployed in snowy areas for quite a while. But they will be deployed in Arizona very shortly.
posted by eye of newt at 9:54 PM on October 6 [2 favorites]


Most of the studies I've seen show an _increase_ in traffic from automated cars. I really don't care if they flow better than manually operated. They'll still tear up infrastructure based on vehicle weight / miles travelled and we live in a country where fuel taxes don't cover infrastructure costs. My County/municipality already can't afford to repave enough to keep up. It'll only get worse.
posted by BrotherCaine at 10:14 PM on October 6 [4 favorites]


If you're going to wave away the appeal/desire/necessity of "door-to-door" transportation, you simply are not gong to solve any problems, and you simply are not going to be taken seriously.

I lived comfortably without door-to-door transportation for a solid twenty years in a city where very many did the same.

"If only we adequately funded transit agencies..." Yeah, if only. If only we funded a lot of things

this is what they did in the city in question. They funded transit and other non-automobile options, and it's worked. And there are still many people opting for the "door-to-door" option (and paying for it). The whole system works more or less. A variety of options serving a variety of preferences and needs.
posted by philip-random at 10:49 PM on October 6 [3 favorites]


See squares in Europe that are shared by bikes, people, cars, and streetcars.

Name *one* square with significant amounts of car traffic where this is actually successful.

(The UK currently has a ban in place on new “shared space” schemes like this because they’ve turned out to be incredibly dangerous in practice)
posted by grahamparks at 10:53 PM on October 6 [2 favorites]


In a thread from a few years ago nonspecialist says that in Taipei pedestrians have the right of way over vehicles by default. I swear I've seen video in some documentary of a bunch of cars immobilized in the middle of foot traffic when a crowd decided to cross the street all at once in downtown Taipei. But, I've never heard of the same sort of thing anywhere else.
posted by XMLicious at 11:19 PM on October 6


I’ve been driving for almost 20 years now, and I still do not know 100% what the distance between me to the next car is if it suddenly crashes and I need to stop immediately.

Count out a two-second or greater interval between you and the car in front of you, often. This will give you rather a lot more distance than is customary, but that's why rear-ending collisions are also customary. Plus it's a good check on your attention to the road. If we're too busy concentrating on other things while driving, we're too distracted to be driving.
posted by asperity at 11:23 PM on October 6 [9 favorites]


If you're going to propose that "if we all can't have something, then nobody can", then I'm probably not the only one who'll be glad you aren't making the rules.

What a weirdly hostile comment. And that was completely not my point at all?

I was talking about the assumption that almost all road traffic will be efficient, safe autonomous cars. People say it will happen, but no one talks about how we’ll get there, or who will get it and who won’t. The timelines for universal adoption are way longer than people think. If one of the selling points of your technology is how great things will be when everyone has it, I’d like to know how everyone is supposed to get it.
posted by shapes that haunt the dusk at 12:05 AM on October 7 [2 favorites]


I agree with shapes that haunt the dusk above. Driverless cars is the fantasy that you can have cities with millions of people without public transport.

The worst traffic I ever saw was rush hour in Istanbul, Turkey. Six or eight lanes of cars, all completely at standstill, bumper to bumper. It wouldn't have been any less crowded if all of these cars were a driverless car.
posted by Termite at 12:17 AM on October 7 [3 favorites]


I can also see the insurance companies raising your premium by 5x or 6x if you don't.

That's what's going to drive (ahem) the transition to autonomous vehicles. Only the 1% will be able to afford the sky high insurance of human-driven cars.


I've been saying this for years. This is America, not Star Trek or the Jetsons.

If you have a driverless car then you could spend that time watching ads if they underwrote the cost of the car, right? And you might want to go to Kroger, but your car could give you a discount if you go to Safeway instead. And you might want to go to that protest downtown, but "I'm sorry, but I can't take you there right now per police orders/too much traffic/it's too dangerous."

I don't believe the argument for a second that insurance rates wouldn't go up for driving you're own car.

You wouldn't really need to own your own car, they could be shared. And if they're underwritten and shared what grounds do you have to argue if it won't take you where you ask? You don't really own it, you just pay someone to use it. Car sharing is a thing now, but not being able to control it changes things.

It would be much more economical and safer to have them drive dedicated routes and lanes and carry more people. But now you have buses and trains.

The list goes on, you could do this all day. I seriously can't think of too many things with more unintended consequences. Well, they may be unintended for the consumer, but the companies involved have thought of a lot more of this than I have.

Of course many of these drawbacks may be built into cars, driverless or not. But it's sure a lot easier to control people with driverless cars.

No one is doing this for your benefit. That's not even part of the equation.
posted by bongo_x at 12:52 AM on October 7 [8 favorites]


I've recently had the opportunity to compare Tbilisi and Kiev. Both post-Soviet capitals of republics with a complicated history of corruption and at times open conflict with Russia. Both starting from a similar level - a working public transport system inherited from the Soviets and an explosion of car ownership once the cars (often used, imported) were available. Both hilly with a river in the middle.

In Tbilisi there's a marked difference between the average looks of people on the street and in the metro - the ones in the metro are clearly the poorer class, people who literally can't afford a car. Everyone else is in their own cars or in the (dirt cheap) taxis, even if they're missing their bumpers, have windshields held together with duct tape or have the driver's seat in the wrong place for the right-hand traffic, with the driver unable to read the Japanese dashboard notices. Driving is a nightmare - about the only thing people respect are red lights, everything else (lanes, traffic directions, sidewalks, right of way) might as well not exist. The trams and city buses were mostly dismantled in favour of a hodgepodge of private buses that sometimes stop in the middle of intersections to let people off, leaving them to wade through two lanes of traffic to the semi-safety of the sidewalk.

Kiev has all sorts of social classes in the metro at rush hour, including guys in suits and women dressed to the nines for the office. Traffic obeys standard European rules, cars wouldn't fail inspection at visual examination alone, and buses and light rail look rather efficient. Just over 20% of journeys in the city are made by car, with average car ownership of 0.6 cars per household.

Guess which country had mostly American advisers in the crucial development phase. It really, really shows in the car-culture and lack of infrastructure investment.
posted by I claim sanctuary at 1:01 AM on October 7 [20 favorites]


Ok so here's a thing I surmise (I do a lot of unsupported speculation but it's the internet welcome to the future).

China, with it's big data/AI fetish and authoritarian bent, is gonna come out hard on automated public transport waaaaaay before anybody else. Not autonomous cars, autonomous buses that can stop and keep you there if they don't like you, 'cause they got card readers, who takes cash anymore, and they're tied to your identity, so, if you board one and you're on a whatever government list you're facked, and that will just be the rationale to go full-bore with automated public transport.

Which probably will inspire an automated/augmented smart-car thing where there's a national alert system that routes you around stopped traffic.

Which in turn will inspire the rest of the world to go augmented driver or public transit + hackable open-sourcey + low-tech folding bikes and stuff, and then we're gonna hit a point in 30 years where roads are still shared, where there's a tribe who loves their auto-drivers, where the plebs swear by public transport, and there's a smaller tribe of manual-only fetishists, I point again to folding bikes and bike share things. This is probably going to all happen as the rest of the world mimics the Singapore/China/Russia model of housing and packs everyone into prefab slab apartment buildings, because global warming and population shifts = you're gonna have to build and subsidize the damn things quick + aging housing stock and slums getting increasingly intolerable.

And this is all gonna lead to automated public transit. We're going back to trolleys-except-they're-autonomous-ish-buses + auto-drive car share service lots which you reach through a still manually-driven rickshaw-or-taxi-or-motorbike depending on your location and service for the middle classes + rich douchebags in three-SUV security convoys for whom they will invent special security classes that allow them to run down "hazardous" pedestrians and non-rights-buying plebs in share-services. And Amsterdam will still be proud of its bicycle paths, because fuck all that. Other specific cities may take up that model as well.

China will be so creepy it heads chokes off full automation well before the tech catches up, buying room for alternatives. Subways and trains will continue as ever, probably replacing cheap air travel along lots of mid-distance routes. Snowy/rough climates will end up looking like Cuba, with their chicken-wire 1950's Chevys, because meh, who told you to live in places without perfect weather? That thing about global warming is probably leading to some huge migrations plus less going out in inclement weather anyway.

I think that's about where we're headed.
posted by saysthis at 3:11 AM on October 7 [3 favorites]


autonomous buses that can stop and keep you there if they don't like you, 'cause they got card readers, who takes cash anymore, and they're tied to your identity, so, if you board one and you're on a whatever government list you're facked, and that will just be the rationale to go full-bore with automated public transport

This is already implemented, for train travel at least, within China's “social credit score” system (1, 2, 3.) A more thorough picture of how it fits into the overall technology-augmented dystopian environment is described in the recent completely terrifying FPP about current conditions in the Uyghur regions of China where checkpoints control even the movement of pedestrians between different areas within cities. In Xinjiang at least there's both ID card and biometric identification in use.

It doesn't have anything in particular to do with self-driving vehicles, though I'm sure access to them will be tied in to the same surveillance and control systems.
posted by XMLicious at 3:43 AM on October 7 [3 favorites]


Interesting to read through the many good points raised. I think it's far enough away from becoming a reality that we can't accurately predict when/how it will occur. Some of the impacts will depend on how it takes shape. As with a lot of technological development, there's likely to be some sort of 'punctuated' phase where there will be an unexpected breakthrough or shift. I think I'm old enough that it won't happen in my lifetime; at least I sincerely hope so. I really like driving.
posted by dancing leaves at 3:54 AM on October 7 [1 favorite]


autonomous buses that can stop and keep you there if they don't like you, 'cause they got card readers, who takes cash anymore, and they're tied to your identity, so, if you board one and you're on a whatever government list you're facked, and that will just be the rationale to go full-bore with automated public transport

This is already implemented, for train travel at least, within China's “social credit score” system (1, 2, 3.) A more thorough picture of how it fits into the overall technology-augmented dystopian environment is described in the recent completely terrifying FPP about current conditions in the Uyghur regions of China where checkpoints control even the movement of pedestrians between different areas within cities. In Xinjiang at least there's both ID card and biometric identification in use.

It doesn't have anything in particular to do with self-driving vehicles, though I'm sure access to them will be tied in to the same surveillance and control systems.
posted by XMLicious at 7:43 PM on October 7 [+] [!]


I didn't wanna get too much into the dystopian aspect of it all, but yeah, it already is tied in there, and it's only gonna get worse as automation gets tied to real ID's. I know it's gonna get worse before it gets better, but we're not China, and I suspect somebody is gonna step up real soon and say, "wait, freedom of movement is a thing here, let's safeguard this" with laws, and I'm expecting a whole lot of political debate over that in other countries as China rolls out what is possible without it. Not saying we're saints by any means, but once you have a legit rollout of dystopia going on, it adds a lot of immediacy to inventing alternatives, and it strips the techbro-jetpack-futurism sheen off these systems real quick. The future includes a mishmash off all these things, and I think we need to get serious about creating standards for electronic ID verification with segmentation and privacy protections. I already live in China, I'm a lost cause, but it's not too late for y'all.
posted by saysthis at 3:59 AM on October 7 [4 favorites]


  1. Self-driving cars are claimed to be ready for the roads right now
  2. The most reliable way to prove you are human online is to select regions of an image that contain a stop sign.
posted by rum-soaked space hobo at 4:26 AM on October 7 [21 favorites]


Yeah, the key to the argument is not really emphasised very well in the argument itself, but the point is that our suburban streets are designed under the assumption that pedestrians stay off it because a car will hit you if you wander onto the road. People wander onto the road all the time, kids play on the road, it's a useful bit of public space that we normally avoid unless we're pretty confident it's quiet.

And it's worth comparing this with railroads where we already have a certain degree of automation. In most circumstances it's illegal for pedestrians to be on the tracks or in the right or way because the desired level of operation (velocity * mass) makes it mostly impossible for a train to avoid collisions. And even with public information campaigns, fencing, and awareness, people still get struck unnecessarily.
posted by RonButNotStupid at 5:07 AM on October 7


Am I missing something, or is this argument absolute nonsense? It seems to run:

THe truth is far simpler and far more boring:

The right wing and the municipal gerontocracy are using the existence of autonomous cars as an excuse not to expand transit or bike lanes.

Pure stupid political rhetoric, and it is not the fault of autonomous car developers.
posted by ocschwar at 6:15 AM on October 7 [3 favorites]


And I can guarantee that the novelty of a self-driving car will wear off.

Almost every day, I see people driving like total assholes because they think they have something to prove. They ignore stop signs and pedestrians in parking lots, ignore motorcycles and bikes on streets, ignore school busses with their STOP signs out, disdain traffic laws in general. They speed up and thread the needle to pass you just so they can be first to wait at a red light, or so they can stop in front of you to make the left turn they knew they were going to make anyway; grandmas drive 25 in a 40 zone and testosterone-poisoned kids tailgate them in their Mustangs and lurch like they're going to ram them. I watched one idiot run into an old woman in a motorized wheelchair a few weeks ago because he looked left without looking forward (and I've been rear-ended twice that way myself).

And almost every day I think "self-driving cars would stop all this bullshit... if they can work at all."

I'm not opposed to better public transit. It's not an either-or proposition.
posted by Foosnark at 6:55 AM on October 7 [8 favorites]


Almost every day, I see people driving like total assholes because they think they have something to prove.

And there's an entire segment of the car industry dedicated to giving them the opportunity to prove something.

When fully-automated driving becomes a reality, is BMW (or Porsche, or any other maker of "performance" automobiles) really going to use an AI with the same behavior as the one in a Honda Civic? Or is it more likely that they'll just use a "performance" AI that changes lanes and drives more aggressively to satisfy the egos they're catering to?
posted by RonButNotStupid at 7:46 AM on October 7 [3 favorites]


Safe, efficient self-driving electric cars could encourage walkable, livable communities.

For a start, electric cars are quiet. A city full of humming electric cars is a lot quieter than a city full of internally combusting cars.

And zero tailpipe emissions, in addition to eliminating harmful and smelly local gases from the system as it is currently designed, would allow us to cover over roads without worrying about ventilation. In cities and other built-up downtown areas, I bet this leads to the road surface level being covered block by block and abandoned to smokeless autonomous vehicles (buses, trams, trains, taxis, and private vehicles) running underneath.

You could cover almost every street in Manhattan and turn the next level up into one safe, quiet, sunny system of parks, paths, and walkways with zero motorized vehicles bigger than a scooter or wheelchair, with maybe a horse lane in certain neighborhoods. Then the biggest, busiest cities in the world could each become a lot of walkable, livable interconnected communities with various electric vehicles circulating underneath for longer distances.

In any case, private cars, whether autonomous or not, are here to stay. Car people (the people who run every city and every country in the world) are never going to voluntarily give up the privilege of traveling from door to door on a secure mobile couch. Every effort you make to shift money from that and towards making them carry all their stuff in their arms to a bus stop and wait for a bus will lose. All you can do is find ways to make private cars safe, ecological, and fun. Self-driving electric cars will do that.

As for worries about the accuracy of navigation, things are always getting better. If acceptance of autonomous cars is a matter of them being better at driving than the average human is, and not a matter of them meeting some impossible standard of perfection, I'm sure the day is coming. In an open road system with people wandering around unpredictably, a few people will still be squashed by robot-driven cars, but it will be fewer than the number who would have been squashed by drunks driving regular cars but were instead safely (and legally) passed by those same beer-guzzling, dope-smoking people in robot-driven cars headed for the next pub or party.
posted by pracowity at 9:14 AM on October 7 [1 favorite]


Also, when all cars are networked and can make decisions where to go you have the means to privatize the roads. So to the person claiming up-thread that autonomous cars will never work because the your city is poor and the roads are just to shitty, well the roads will be great and perfect but will probably now be owned and maintained by Verizon or Koch or something.
posted by uandt at 10:39 AM on October 7


For a start, electric cars are quiet. A city full of humming electric cars is a lot quieter than a city full of internally combusting cars.

Except at very low speeds, electric cars make almost exactly the same amount of noise externally as petrol ones. There are also various studies that suggest a lot of the air pollution around busy roads comes from resuspension of road dust, much of that dust coming from brake, tyre and road wear. So a road used only by electric cars would by no means pleasant to live alongside.

All you can do is find ways to make private cars safe, ecological, and fun. Self-driving electric cars will do that.

Let me tell you about a form of urban transport that's safe, ecological and fun that also happens to be free to use and makes you healthier. Or would be those things if there weren't so many cars.
posted by grahamparks at 11:02 AM on October 7 [5 favorites]


whinnies - "horse lane in certain neighborhoods"
very punny pracowcity
posted by unearthed at 11:40 AM on October 7


Let me tell you about a form of urban transport that's safe, ecological and fun

Yeah, I know all about bike transport. I ride one to work and back every day through sun, rain, or snow, all year round.

But we're not talking about me. My neighbors, who have pretty much the same transportation options and needs, all drive cars. To them, cars are necessary. Part of modern life, like phones. If you told them to give up their cars, they would laugh at you. If you told them to give up their cars, I would laugh at you for bothering.

The best you might hope for is to get them into a self-driving electric car that comes at their call and takes them and their stuff comfortably from door A to door B while they do things other than guide the car. Then you could work on reducing pollution and improving safety centrally.
posted by pracowity at 11:54 AM on October 7 [1 favorite]


We're not going to get people off cars by telling them to ditch their cars. We get people off cars by stopping the massive subsidies that make personal cars so attractive. We stop prioritizing cars in designing out cities. We improve transit and biking infrastructure. The war on cars works, if you can follow through. When transit is more convenient than cars, people switch. Not all of them, but that's okay.
posted by vibratory manner of working at 12:29 PM on October 7 [18 favorites]


We stop prioritizing cars in designing out cities. We improve transit and biking infrastructure. The war on cars works, if you can follow through. When transit is more convenient than cars, people switch. Not all of them, but that's okay.

this is effectively what Vancouver has been doing for a few decades now, and it's accelerating. Every car unfriendly decision is, of course, complained about by some, but such decisions keep getting made. And it's not as if it's been hurting real estate values.
posted by philip-random at 12:55 PM on October 7 [2 favorites]


The most reliable way to prove you are human online is to select regions of an image that contain a stop sign.

I'm guessing that stop signs stand out dramatically on millimeter wave radar though. As well, you can just map the stops and use location data. We're still a long way from successful autonomous vehicles for a host of other reasons though.

If you told them to give up their cars, they would laugh at you. If you told them to give up their cars, I would laugh at you for bothering.

I think the in between area between cars and bikes of electric assist velocars and ebikes may offer some incentive for some drivers. I know it's made a small amount of converts where I live. Not as much as creating a low stress network of bike infrastructure does, but something.
posted by BrotherCaine at 1:39 PM on October 7 [1 favorite]


I’ve been driving for almost 20 years now, and I still do not know 100% what the distance between me to the next car is if it suddenly crashes and I need to stop immediately. I know what a safe distance if the car slows or brakes gently, because that’s most of what I encounter, but I don’t know how far I need to be if I slam on the brakes because I never slam on the brakes unless there’s something immediately in front of me.

I would love it if the car made the same binging noise it does for seatbelts or lane shifts and said “unsafe following distance at your speed”. That would be super exciting and helpful! I just rented a car this week and have been marveling at the lane shift protection.


On a dry road with good visibility, a reasonable following distance is... 2 seconds. In other words, your car should pass a given point no less than 2 seconds after the same point is passed by the car ahead.

And this measure works at any speed. When the car ahead passes a signpost, start counting: one-steamboat, two-steamboat ... and now your bumper should be passing that same signpost.

Don't need no steenkin computer to follow at a safe distance. Given how most people drive, a safe-distance alert will be disconnected by most users in a couple of weeks.
posted by Artful Codger at 1:42 PM on October 7 [4 favorites]


vibratory manner of working: Which is great, right up until Rob Ford (or local equivalent) gets elected mayor. Seriously, any idea that has an implicit "First, we assume that we have control of all layers of government indefinitely" at the front is a bad idea.
posted by Grimgrin at 2:11 PM on October 7 [1 favorite]


Self-driving cars seem attractive to some city planners because they hold out the promise of improving driving safety and traffic problems, with almost no improvements to infrastructure. Which is pretty naive.

The touted example of the automated container-moving vehicles at container-ports is unique, and a key factor in its success is infrastructure: laneways marked electronically and a centralized computer that orchestrates all movement. And of course a tightly-controlled environment with almost no un-tracked moving objects sharing the space. The current concept of self-driving cars doesn't have those factored in.

Autonomous vehicles will be no substitute for good urban design which REDUCES the number of private vehicles circulating, in favour of pedestrians, bikes, and public transit.
posted by Artful Codger at 2:18 PM on October 7 [4 favorites]


any idea that has an implicit "First, we assume that we have control of all layers of government indefinitely" at the front is a bad idea.

I don't think that that's an implicit assumption here, any more than it is in any political program. Sure, it's easier to implement a thing if you have complete control, but you do what you can when you can, and push for more, and not everything can get rolled back by your opponents when they have the upper hand. That's the case for pretty much anything you want to achieve at that level. Or are you of the opinion that trying to do anything with the government is a bad idea?

It sucks, it's slow, there will be setbacks, but trying is better than not trying.
posted by vibratory manner of working at 3:19 PM on October 7 [1 favorite]


The touted example of the automated container-moving vehicles at container-ports is unique,

Oh that documentary film footage taken however many years ago is just something I had handy from a comment in a previous thread because I saved a link in 2016 when it showed up on Deutsche Welle's Youtube feed. Just by checking Wikipedia you can see that there are many more examples of autonomous ground vehicles in use, not to mention aircraft, trains, ships, and spacecraft going back well into the 20th century.

and a key factor in its success is infrastructure

Unlike human-operated ground vehicles, where all you have to do is level and grade immense tracts of land, carve mountainsides away, dig tunnels, and build bridges, even when they run on tracks rather than roads?

Something tells me that as soon as we have autonomous vehicles which require less infrastructure than human-operated vehicles, BigDog-type things that can walk between trees in a forest and ford rivers on foot and cross canyons on a tightrope and hop around like mountain goats, the weight and cost of infrastructure is going to turn out to actually be character-building or something like that and it will instead be declared a deficiency of autonomous vehicles that they don't require as much infrastructure.
posted by XMLicious at 4:14 PM on October 7


As a committed non-car person. No driver's licence. The car in my driveway is not my car. I use public transport - planes, trains, buses, streetcars. And I walk. I probably should start cycling more but my little city has an awful lot of hills, so that may need more time so that I have the necessary fitness.

My part of the world has lost hundreds of small towns and villages over the last 60-80 years because - CARS. Commuting makes "sense" when you can devote 8-16 hours a week to driving from one place to work in another in reasonable comfort, so you stop living close to work. My guess is that if the journey becomes even more comfortable, especially if I can get a bit of a snooze on the way, why not devote 16-32 hours to commuting? I like those short plane commutes - a 2 hour flight is just long enough for a refreshing nap. Should work for car commuters too. So now the commuter travels even further for work.

If you think we have a problem with urban sprawl now - some weird variant of Moore's Law is going to scale it up even more.
posted by Barbara Spitzer at 11:46 PM on October 7 [3 favorites]


If you think we have a problem with urban sprawl now

Especially because you'll be able to work in the car while it drives. People will set up little workstations in their cars and do a couple hours on the morning drive, a couple hours more on the evening drive, and the face-to-face stuff in between. It might instead shorten your workday, though, if you can keep the same commute and include your drive time in your working hours, and you might be able to commute to a fairly distant college that would otherwise be impossible for you to attend because you have responsibilities back home.

Depending on what the law allows, people might even start sending kids off to school in driverless cars that hand them off to receiving adults at the school end. You might not want to send a kindergartner off alone in a car, but maybe a teenager who wouldn't otherwise be old enough to drive?

Motels will have to offer something besides temporary beds (maybe quick-stops for toilet and shower breaks?) if you can just sleep in your car while it drives you across the country.

Package and pizza delivery might amount to a driverless van that parks in front of your home and opens the compartment for your delivery after you give it your information.

Lots of strange things are going to happen if robots take to the streets.
posted by pracowity at 8:24 AM on October 8 [1 favorite]


METAFILTER: Lots of strange things are going to happen if robots take to the streets.
posted by philip-random at 9:41 AM on October 8 [3 favorites]


"and a key factor in [the success of automated containerports] is infrastructure"

Unlike human-operated ground vehicles, where all you have to do is level and grade immense tracts of land, carve mountainsides away, dig tunnels, and build bridges, even when they run on tracks rather than roads?

Absolutely, yes. Unlike those.

In an automated container port, all the routes are all laid out and marked electronically; there's no need for AI and complex vision algorithms to detect and follow conventional roads. All automated container transporters are peers and all under the control of a central computer system. And there's very very few pesky unregulated obstacles like pedestrians, bikes, non-automated vehicles. That kind of infrastructure.

The proponents of self-driving cars maintain that they can achieve personal vehicle nirvana without any of that. Pretty naive.

That Arizona fatality where the car's system failed to recognize a pedestrian wheeling a bike across the road - epic fail, and any system that could not pass that most basic of tests should not have been allowed on public streets.
posted by Artful Codger at 9:51 AM on October 8 [3 favorites]


all the routes are all laid out and marked electronically

Again, as opposed to laying out and building an entire road system? All steps of which are done electronically for new and reconstructed roads today, by the way?

here's no need for AI and complex vision algorithms to detect and follow conventional roads.

Just a system of driver education and licensure oriented around standardized signage and lines which have to be painted on the road surface itself.

any system that could not pass that most basic of tests should not have been allowed on public streets

So systems which involve a human driver should not be allowed on public streets, because you could definitely find a human-driven car which has fatally failed that test. Way, way, more than one in fact. You're basically making the argument for them at this point.

Designating one example as supposedly having been “touted” in an attempt to use it as a piñata for a disproof of “personal vehicle nirvana” rather than actually saying anything about the technology in general is goofy and masturbatory. It's abject self-fahrvergnügening.

But the people in these threads as well as many of the past FPPs on the subject which intelligently try to anticipate and explore real-world consequences and problems of this actually-existent technology convey a great deal of insight and in this particular thread the comments on the whole are more interesting than the OP articles themselves.
posted by XMLicious at 5:33 PM on October 8


Either I completely suck at English, or you've just about taken everything I've said out of context, or worse. I mean, it could be the former, but you'll have to try harder to convince me of that.

"all the routes are all laid out and marked electronically"

Again, as opposed to laying out and building an entire road system? All steps of which are done electronically for new and reconstructed roads today, by the way?

No, I was trying to state that the routes of an automated container port are physically festooned with electronic markers in the pavement for the vehicles to follow, and that (I imagine) also can report as vehicles pass over them.

"There's no need for AI and complex vision algorithms to detect and follow conventional roads."

Just a system of driver education and licensure oriented around standardized signage and lines which have to be painted on the road surface itself.

Sorry, what does that have to do with automated container ports? Lemme try again: in an automated container port [t]here's no need for AI and complex vision algorithms to detect and follow conventional roads, or to expect unregulated traffic.

Also, notice that they're rather slow?

"any system that could not pass that most basic of tests [- not hitting something in front of them] should not have been allowed on public streets"

So systems which involve a human driver should not be allowed on public streets, because you could definitely find a human-driven car which has fatally failed that test. Way, way, more than one in fact. You're basically making the argument for them at this point.

Absolutely disagree. Self-driving cars are pointless if they fail something as basic as not hitting a person wheeling a bike across an otherwise empty street. A human would only commit that error through inattention or malice - the most basic things that self-driving vehicles are supposed to eliminate.

If an AI system can't recognize and brake for a human-sized jaywalker, do you really want it on the public roads? Either Uber BSed the officials about the system's readiness, or Arizona didn't do due diligence.

Designating one example as supposedly having been “touted” in an attempt to use it as a piñata for a disproof of “personal vehicle nirvana” rather than actually saying anything about the technology in general is goofy and masturbatory. It's abject self-fahrvergnügening.

Pithy retorts to decontextualized lines is somewhat self-pleasuring, no? Technology is my field; I can go there when appropriate. But my point is more about policy/expectations, and the rose-coloured distortions that seem to predominate when people discuss the reality of self-driving cars.

Everyone knows they're inevitable, but there's so much to do before they are mainstream. and, by themselves, they will not significantly improve the problem of urban congestion and efficient transportation.
posted by Artful Codger at 7:24 PM on October 8


Absolutely disagree. Self-driving cars are pointless if they fail something as basic as not hitting a person wheeling a bike across an otherwise empty street....
Everyone knows they're inevitable, but there's so much to do before they are mainstream. and, by themselves, they will not significantly improve the problem of urban congestion and efficient transportation
.--Artful Codger

One problem for public perception and reputation of self-driving cars is the terrible, reckless job Uber did in their testing, resulting in someone's death.

In contrast, Google has been testing self-driving cars on public roads for many years, many of them didn't even have a steering wheel or pedals! You may not have heard of this because it has been so uneventful.

(I work near Google, so I see them and drive along side them every day).

There isn't much more for Google to do except open them up to the public, which they are doing in the next couple of months in Arizona.

Can't speak for urban congestion, but in the suburbs I think they will be much more efficient getting you from point A to point B than the mostly empty buses I see driving around (except on a few major routes). If you do happen to go by bus, you will usually end up travelling twice as many miles in those near-empty buses and take three times as long to get there. Lots of transportation authorities actually understand this and from time-to-time they try out on-call bus systems. With self-driving cars, there is a chance to actually make such a system work.
posted by eye of newt at 9:59 PM on October 8


eye of newt - I admit I'm mainly interested in the urban scenario, because I think that's where transportation bottlenecks becomes acute and affect more people, as well as the economy.

For the suburban scenario... yes I guess the potential benefits of self-driving cars could be more fully enjoyed... but by whom? Mainly the better off folks who can afford to live in the 'burbs and to buy a self-driving vehicle, or subscribe to a service. So we're back to my main complaint that self-driving cars are mostly a shiny thing that will at best improve things only for the well-off, and that they draw attention away from the harder problems of urban design and of mass transit that works for everyone.
posted by Artful Codger at 4:56 AM on October 9 [2 favorites]


For those with an hour, I highly recommend this vision of what autonomous electric vehicles are going to be like, and why it’s not a distant future.

The key: Think of transportation as a service. Very few people will own their own EVs, just as very few people own their own airplanes. The engines will be electric. The electricty will almost certainly be from renewables, which are already cheaper in most settings than fossil fuels.

Yes, people have a romance with their own vehicles. That will dry up pretty quickly if car ownership is A) significantly more expensive than hailing an EV as needed; and B) inconvenient (maintenance, especially if auto shops and gas stations start to become scarce).

Rather than individuals owning cars which are in use 4% of the time, think of cars in use almost all the time, until their batteries run out. That means a much smaller fleet of vehicles overall.

What this will look like for traffic overall is debatable, but it does mean a hugely reduced parking infrastructure. Los Angeles’ total parking lot area is larger than the area of many large cities.
posted by argybarg at 5:36 AM on October 9 [1 favorite]


For the suburban scenario... yes I guess the potential benefits of self-driving cars could be more fully enjoyed... but by whom? Mainly the better off folks who can afford to live in the 'burbs and to buy a self-driving vehicle, or subscribe to a service. So we're back to my main complaint that self-driving cars are mostly a shiny thing that will at best improve things only for the well-off, and that they draw attention away from the harder problems of urban design and of mass transit that works for everyone.--Artful Codger

Here in Silicon Valley (I'm sure NYC is similar), it is cheaper to live in the suburbs than in the city. There's even a reverse commute--young professionals would rather live in the city but most of the jobs (including Apple, Facebook, Linkedin, etc) are actually in the more suburban areas.

It doesn't have to be expensive. The on-call-bus systems they experimented with cost users no more than a regular bus, and it would be more efficient and even less expensive to run with self-driving cars.
posted by eye of newt at 8:08 AM on October 9


Rather than individuals owning cars which are in use 4% of the time, think of cars in use almost all the time, until their batteries run out.

At some point someone's going to figure out how to run a fleet while swapping out for fully charged batteries at charging stations for close to 100% uptime.
posted by BrotherCaine at 8:47 AM on October 9


I notice so many of these visualizations of a fully automated self-driving car future omit pedestrians. If I just want to cross the street, who gets right of way? What if I’m riding a bike? What if I’m in wheelchair?

That the self-driving car future promo videos ignore the idea of multiple mode transportation is telling. The system only works if the system is replaced entirely by one thing. I shouldn’t need to pay for a self driving car subscription to go where my feet can take me.
posted by SansPoint at 10:09 AM on October 9 [2 favorites]


Another point. Let’s assume your average self-driving car has the same average capacity of your average dumb car: 4-6 people tops.

A bus, self-driving or otherwise, takes up the space of two to three cars, and even if the cars are fully packed, the bus can still hold more than twice the number of passengers as the cars. How efficient can a self-driving car system be when it’s still carrying an average of two to three people per trip? If you’re planning an Uber Line situation, just make a bus instead!
posted by SansPoint at 10:12 AM on October 9 [2 favorites]


argybarg: "Rather than individuals owning cars which are in use 4% of the time, think of cars in use almost all the time, until their batteries run out. That means a much smaller fleet of vehicles overall.
"

But car use isn't evenly distributed rather being very peaky. Regardless of whether cars are individually owned or parts of fleets owned by google/uber/GM you'll need enough cars to transport everyone to work on Wednesday morning. Heck if fully autonomous cars are affordable you might even see an increase in required fleet size because of people tolerating longer commutes and users without the ability to drive being able to make use of personal transportation (teenagers, disabled people, drunks, etc.).
posted by Mitheral at 12:51 PM on October 9 [2 favorites]


And the traffic thing is actually kind of a problem because the alternative that keeps being suggested here - use mass transit options that are proven technology, already exist, and work when deployed - also solves your parking problem and your traffic problem by getting people off the road in the first place.

Hell, you build actual biking infrastructure and you get point-to-point transit as well.
posted by Merus at 4:20 AM on October 10 [3 favorites]


I won't believe in truly safe, autonomous vehicles until we have numeric cadastres and time-based location datum. Sure the vehicles have LIDAR and a bunch of sensors and cameras, and their routes can be programmed manually, but they also rely on GNSS that is fixed at the centre of the earth, and our continents are constantly shifting. 2.5cm might not seem like very much, but when it's the difference between staying safely on-course or clipping the gutter, it can make a pretty big difference. And out in the regions, where datum is even shittier? Fuggedaboudit.
posted by turbid dahlia at 4:41 PM on October 10


Not to mention that global warming will make GPS increasingly inaccurate to begin with.
posted by SansPoint at 6:43 PM on October 10 [1 favorite]


« Older Humble Publishing   |   On this strange and mournful day Newer »


You are not currently logged in. Log in or create a new account to post comments.