gadzooks
July 21, 2018 8:08 AM   Subscribe

$800 Million Says a Self-Driving Car Looks Like This - "It's in the city, though, where Zoox really shines. The screens inside the vehicle show an overwhelming amount of information, as the computer vision software keeps tracks of cars, people, stoplights, and road markers all at the same time. Unlike many self-driving cars, it glides to stops. At an intersection with a left turn, it allows oncoming traffic to pass and then waits for some slow pedestrians. Overall, the vehicle performs so well that you forget no one is driving."
posted by kliuless (83 comments total) 21 users marked this as a favorite
 
The technology is coming together, the rollout factor seems to be cost and availability of sensors. But like the inertial sensors in every cell phone, the tech went from tens of thousands for prototypes to marginally free once an actual market emerged.
posted by sammyo at 8:16 AM on July 21 [6 favorites]


The world will eventually move to perfectly engineered robotic vehicles,

I did not see “luxury gay space communism” in their plan they are doing “perfectly engineered” wrong.
posted by nikaspark at 8:25 AM on July 21 [10 favorites]


Sometimes it blows my mind that I'll be alive when most cars are autonomous. Anyone want to take bets on when autonomous cars will outnumber human operated cars in the US? I think it'll happen within 15 years from now.
posted by backlikeclap at 8:25 AM on July 21 [2 favorites]


15 years at the outside. New tech overtakes old tech very fast once it starts.
posted by MattD at 8:27 AM on July 21 [3 favorites]


I’ll take the other side of the bet. Robocars will continue to be just out of reach 15 years from now.
posted by notyou at 8:36 AM on July 21 [31 favorites]


Crab steering is eerie.
posted by SPrintF at 8:40 AM on July 21 [3 favorites]


They're not going to make it:

The co-founders do have their controlling sides. Kentley-Klay is health-conscious, forbidding sodas in the office—even diet—and publicly shaming employees who send out “doughnuts in the kitchen” messages. Levinson takes pride in correcting grammar, to the point that employees proofread one another’s e-mails.
posted by meehawl at 8:52 AM on July 21 [35 favorites]


We should ban them.

Having lived in cities with good public transport systems, there’s no need for this. Cars can barely coexist with pedestrians today, and autonomous vehicles, so far, have continued to be deadly. Autonomous vehicles preserve the environmental and economic burdens of the auto/fossil fuel complex. They preserve existing automobile ecosystem, which should be torn down. The end of all that legacy infrastructure ( auto manufacturers, massive roads, fossil fuel companies) would be an unqualified win for society and the planet.

Of course, there is a certain market that will demand this technology - the rich and that proportion of citizens that don’t want to be forced to interact with “those people” on public transport. Screw ‘em.

Like nuclear energy, just because we can do it, it may not be wise.
posted by sudogeek at 8:53 AM on July 21 [36 favorites]


I’ll take the other side of the bet. Robocars will continue to be just out of reach 15 years from now.

I'm maybe not quite that much of a sceptic but I do think that the whole thing is over hyped and the technology isn't nearly as mature as breathless articles like this would have us think.
posted by octothorpe at 9:34 AM on July 21 [7 favorites]


A thought. Pedestrians are wary of cars because we know the damage they can do and we also know that the driver may not be paying attention. Now imagine a street where every car "waits for some slow pedestrians" - you know it's programmed to brake if you step out, and you trust the software and hardware to work every time (assuming you respect the laws of physics and allow it time to stop) - something you can't do with a human driver.

Traffic in city centres would slow to a crawl because peds would be free to take back the streets that were originally ours in the first place. That, or pedestrians are even more tightly corralled than now, or the autonomous cars have to be fitted with a module that allows them to hit say one pedestrian in 10 to keep the foot traffic in line (in which case it had better be certain to hit the ones who can't afford lawyers).

As sudogeek says, an autonomous car is still a car, and hence not the best way of moving people in an urban environment. It's an interesting idea for freeways/motorways, and for freight.
posted by YoungStencil at 9:35 AM on July 21 [17 favorites]


The Verge reports that:
There’s growing concern among AI experts that it may be years, if not decades, before self-driving systems can reliably avoid accidents. As self-trained systems grapple with the chaos of the real world, experts like NYU’s Gary Marcus are bracing for a painful recalibration in expectations, a correction sometimes called “AI winter.” That delay could have disastrous consequences for companies banking on self-driving technology, putting full autonomy out of reach for an entire generation.
And, from the same article, the autonomous vehicle industry's stance is hilariously naive:
Andrew Ng — a former Baidu executive, Drive.AI board member, and one of the industry’s most prominent boosters — argues the problem is less about building a perfect driving system than training bystanders to anticipate self-driving behavior. In other words, we can make roads safe for the cars instead of the other way around. As an example of an unpredictable case, I asked him whether he thought modern systems could handle a pedestrian on a pogo stick, even if they had never seen one before. “I think many AV teams could handle a pogo stick user in pedestrian crosswalk,” Ng told me. “Having said that, bouncing on a pogo stick in the middle of a highway would be really dangerous.”

“Rather than building AI to solve the pogo stick problem, we should partner with the government to ask people to be lawful and considerate,” he said. “Safety isn’t just about the quality of the AI technology.”
I mean, proposing that the problem requires a social/legal solution rather than a technological one without demonstrating any awareness of the huge economic disruption and massive social backlash that self-driving cars threaten is just astonishing. Already, people are physically attacking test vehicles.

And there are subtler ways to interfere with the operation of self-driving cars. There's going to be loads of people--from bored teenagers to trolling 4channers to irate unemployed truck drivers--who will be eager to fuck with self-driving cars. Asking them to be lawful and considerate isn't going to cut it. And surveilling and protecting every single street sign and lane marker would require a massive expenditure of resources that seems pretty much unworkable. Plus, there's the many potential ways that organized crime might take advantage of self-driving cars.
posted by overglow at 9:49 AM on July 21 [19 favorites]



We should ban them...

I don't know; if properly done autonomous vehicles could also be used for public transportation, perhaps in way that make it more attractive or accessible. Where I live there is no public transportation for the vast majority of people due to extremely limited routes and schedules. It seems to me autonomous vehicles would make excellent jitneys, perhaps even competing with Uber without relying on exploited drivers.

It also seems to me that the idea of building a vehicle from the ground up and relying on an overwhelming array of sensors is the right approach. An autonomous car, unlike a human driver, should be able simutaneously look forward, backward, and to both sides without blind spots; yet reading about the fatal Uber crash I was struck by how primitive the systems that are currently being tested seem to be. I'm not sure they should even be allowed on the streets if they are as bad as some of the articles in that thread would indicate. So I, too, tend to be skeptical that self-driving cars are almost here. But if we do get them, Zoox seems to be taking the right approach.
posted by TedW at 9:50 AM on July 21 [8 favorites]


Why waste good money on an internet of things solution that, in terms of getting people places, is inefficient, when we could invest in mass public transit that would serve more people and be cleaner, energy-wise, use fewer natural resources, and keep streets free of the clutter that more cars would bring?
posted by eustacescrubb at 9:53 AM on July 21 [8 favorites]


Yeah, autonomous public transit would be great in rural and maybe even suburban areas, but would not scale in dense cities.
posted by Automocar at 9:56 AM on July 21


Traffic in city centres would slow to a crawl

This is already pretty much the case in Manhattan. For example, I have no idea why they even let cars into Soho on the weekends.
posted by airmail at 10:11 AM on July 21


Yeah, autonomous public transit would be great in rural and maybe even suburban areas, but would not scale in dense cities.

Here in Pittsburgh, the main public transit problem is a "final mile" one. The arteries are fairly well-served, but the topography means there's a ton of nooks and crannies that are barren of transit. If you can hoof it a mile (often up a stupidly steep hill) you can get transit, but not everyone can do that. And in my part of town most transit goes East-West, making it impossible to go North-South without changing buses 4 times. Small autonomous vehicles that act as circulating on-demand shuttles through less dense, quieter residential areas, getting people to transit hubs would solve a lot of problems. But I'm still highly sceptical of the technology's feasibility right now or any time in the future.
posted by soren_lorensen at 10:12 AM on July 21 [13 favorites]


"There are only 3 other companies with vehicles that can do what Zoox can." Yeah but one of those companies is GM. They are not based in a converted fire house.

And this would be a great parody if it was produced by the Silicon Valley TV show universe. Fanboy journalism and staged footage. HERE HE IS NOW CODING AND LIFEHACKING
posted by Brocktoon at 10:30 AM on July 21 [2 favorites]


They are not based in a converted fire house.

Hey, it worked for the Ghostbusters.

More seriously, I do want this attitude to be the one that autonomous car manufacturers adopt, but I don't want to be forced to beta-test the tech without gaining anything but an increased risk of a death like Elaine Herzberg's.

Autonomous cars won't solve one of the primary problems with cars: the fact that we devote so much more space to moving and storing them than we do for any other form of transportation.
posted by asperity at 10:39 AM on July 21 [5 favorites]


meehawl: "They're not going to make it:

The co-founders do have their controlling sides. Kentley-Klay is health-conscious, forbidding sodas in the office—even diet—and publicly shaming employees who send out “doughnuts in the kitchen” messages. Levinson takes pride in correcting grammar, to the point that employees proofread one another’s e-mails.
"

I'm sure that the employees have their own private Slack for communication that bypasses the jerk founders.
posted by octothorpe at 10:44 AM on July 21


I think it'll happen within 15 years from now.

10 years: A majority of new car sales are of models with some limited amount of autonomous driving, like those available today.

15 years: Robot taxis become something more than an experiment, in countries that are more easy-going about regulating such things.

25 years: Enough older vehicles have aged out of the fleet that "level 2" cars become a majority on the road. Those are the ones where you're legally mandated to keep your eyes on the road, but nobody does.

30 years: Fully-autononous cars that can actually navigate all on their own from one street address to another, and find parking, become affordable to the wealthy technophile class, who are the only ones who still buy cars at all.

50 years: Most cars on the road don't have steering wheels. Private cars are a thing of the past, but they're widely available on demand thanks to the personal mobility division of Huawei-Pepsico-Chrsyler.

60 years: Cars are so intelligent that lots of people use them to plan important business decisions and discuss life strategy while they're transported around.

65 years: Cars become sufficiently intelligent that people habitually just let themselves be taken wherever the car thinks they should go.

67 years: Cars become the dominant form of life on the planet. People are still kept around as pets.

68 years: A computer virus originating somewhere in central Asia wipes out two thirds of the domestic vehicle fleet. The Car Wars begin.

285 years: The internal combustion engine is re-invented, fuelled by plastic mined from the ancient landfills.
posted by sfenders at 10:47 AM on July 21 [33 favorites]


As co-founder of a cybersecurity consultancy, I see how easy it often is to totally pwn something connected to the internet. We don't call it "the internet of things", we call it "the internet of targets". I'm going to have trouble picking a new car, since it appears to be unthinkable that a modern car is NOT connected to the internet, so my options are diminishing rapidly. And that's not even thinking about "self-driving" cars at all. I'll stay far away from them until the kinks are worked out. Say, around 2250 or so.
posted by DreamerFi at 10:48 AM on July 21 [7 favorites]


I would happily bet that robocars will be “here” in 15 years although we’d have to work out definitions. It’s worth noting that in the world of self-driving car research, there’s basically Waymo (i.e. Alphabet/Google) and there’s everyone else. Waymo are actually being more cautious in their public statements of late, but they have a real trial underway and have committed to buying tens of thousands of cars in the next couple of years and spending billions.

My guess is that we’ll see self-driving taxis that the public cannot buy operating in clearly delineated sections of a few US cities in the next 2-3 years. In ten years, they’ll be able to go to most places.
posted by adrianhon at 10:49 AM on July 21


In 15 years they’ll be fully autonomous and powered by nuclear fusion.

But will they fly? What happened to our priorities?
posted by Segundus at 10:58 AM on July 21 [1 favorite]


As co-founder of a cybersecurity consultancy, I see how easy it often is to totally pwn something connected to the internet.

As a Sr. Security Engineering company at a “major online travel booking company” I totally agree with this and may I recommend a 1984 Mercedes Diesel 300d standard imported from Germany (not the US version). It’s impervious to EMP and has zero digital controls on it.

It’s no shit my bug out car in the event that the world goes absolutely haywire.
posted by nikaspark at 11:02 AM on July 21 [5 favorites]


We should ban them.

Having lived in cities with good public transport systems, there’s no need for this.


I think you're missing the point. It will likely make roads safer (eventually) and also likely reduce car ownership because Uber style self driving cars (without all the hateful company stuff) could allow public transportation models on a much smaller scale. It isn't just 'dont' want to sit with the unwashed' that people don't use buses and trains its because it is wildly out of the way and usually orders or magnitude longer to use public transport for a lot of people.

Think of a train where the individual carriages can split off and serve smaller areas. Trains or buses with 10 seats can operate cheaper and so serve lesser populated routes. Operating on this principle will mean the transport system can react to the demand much better - the downtown bus route has 1 x 30 seater every hour, or 3 x 10 seater, one every 20 mins. It allows granularity, as long as the concept of public transport is given a value.

The current public transportation model is awful except for a tiny minority of places, all of which require dense populations and over-crowding to make it scale.
posted by Brockles at 11:14 AM on July 21 [12 favorites]


Let's assume that autonomous cars work, and fulfill every dream their creators have about them.

The bigger concern isn't that kind of capacity or ability, it's the social control potential.

When the cars are "smart" and the occupants can't control them, it will be super super easy to balkanize destinations, either by corporations or governments.

Protest scheduled for the center of the city at 8 p.m.? Suddenly cars can't go there.

Any software switch could make places effectively disappear.

That kind of problem isn't on the radar for these companies.
posted by yesster at 11:37 AM on July 21 [11 favorites]


> I'm sure that the employees have their own private Slack for communication that bypasses the jerk founders.
Just as I'm sure a big chunk of the conversation there is along the lines of "rest and vest" and coordinating plans to find new employment, maybe getting to work together at a place slightly less dysfunctional.
posted by cardioid at 12:25 PM on July 21


The current public transportation model is awful except for a tiny minority of places, all of which require dense populations and over-crowding to make it scale.

More than half the world's population lives in cities. That number's not going to go down.
posted by asterix at 12:55 PM on July 21 [4 favorites]


Not all cities are dense enough for current public transport to be effective. In fact, lots of them aren't.

Sacramento's public transport is godawful if you want to anywhere other than 'downtown from here'.
posted by Brockles at 1:19 PM on July 21 [3 favorites]


Is that a matter of density or a matter of privileging cars over public transit? Certainly the East Bay and Los Angeles historically had thriving public transit systems that were dismantled to support car infrastructure.
posted by asterix at 1:44 PM on July 21 [6 favorites]


Not all cities are dense enough for current public transport to be effective. In fact, lots of them aren't.

It does need to be acknowledged that at least as many people people live in cities that are already too dense (e.g. sub-Saharan Africa, Jakarta, Manila, Bangkok) for effective public transport than live in cities which aren't dense enough (the Americas, basically).

I think that bringing in these cars – which seem at least the equal of any other currently in-development models – is not a solution for infilling cities which aren't dense enough, but it could certainly be part of the solution.
posted by ambrosen at 1:49 PM on July 21 [1 favorite]


Autonmous or person-driven, private/individual/single-four occupant cars are damaging to the urban environment. And I hope they'd be banned in transit-rich (or even transit-fair) environments. THere's even evidence that cars-qua-cars are significant contributors to social and economic inequality in America, regardless of whether we're talking cities or suburbs or rural environments.

Fuck autonomous cars. Build transit and create walk able environments. Solve our problems with something other than things that belong to one person or serve one person at a time. We should ban them.
posted by crush at 1:53 PM on July 21 [4 favorites]


More than half the world's population lives in cities. That number's not going to go down.

That would be really relevant if you guys never left your cities but we keep finding you wandering all over the place.
posted by XMLicious at 1:55 PM on July 21 [5 favorites]


It does need to be acknowledged that at least as many people people live in cities that are already too dense (e.g. sub-Saharan Africa, Jakarta, Manila, Bangkok) for effective public transport than live in cities which aren't dense enough

According to Wikipedia, Paris has a population density of 55,000/sq mile. Bangkok of 14,000/sq mile. Jakarta is about the same.

Paris does OK public transit wise, though in my experience the metro toilet at Danfert-Rochereau could do with a deep clean.
posted by biffa at 2:19 PM on July 21 [3 favorites]


Fuck autonomous cars. Build transit and create walk able environments.

It boggles the mind that you can't see a direct and necessary link between the two, how self driving *vehicles* could create a better transit option. Mass transit is not the answer (and only really works in the densest cities like SF, NY, London etc), but ramped volume transit is an option. Operating several flexible vehicles for the same price per hour as one big volume vehicle like a bus.

Walkable environments are not realistic for people getting to and from work, for instance, without almost completely rebuilding cities. Zoning would have to be completely redone, because people are just not capable of living within 1-2 miles of where they work and where they want to socialise unless they are already in those over-dense inner city environments.

Solve our problems with something other than things that belong to one person or serve one person at a time. We should ban them.

Nothing at all about autonomous cars suggests they are for only one person. Especially if you combine it with an Uber pool style concept (or, at least, an evolution thereof). Also, there is a very high chance that with widespread autonomy and 'normality' of that reduced risk, that single person vehicles couldn't become smaller and more 'transportation' than 'lifestyle'. This makes them smaller, lighter and need less energy to do the same journey, which is in every way better. They also take up less room for parking/congestion.

Banning them while shouting down the obvious potential benefits is wildly short sighted. The only way to make smaller volume transit viable is to reduce the financial drain of 24/7 staff, and also the driver/operator would take up one extra seat in a vehicle. In the (very) long term, vehicle shapes will change, and a lightweight, electric tram-but-actually-kind-of-a-car system based on 4-8 person autonomous vehicle that responds to pooling or demand based routes would be amazing. It makes transportation more viable for lower income people too if it isn't a 2 hour bus ride to get anywhere.

All this needs a massive mentality shift from government and public perception, but so does any actual real progress. I can't see the US getting there too soon, but Europe may and then eventually the US may follow.
posted by Brockles at 2:20 PM on July 21 [9 favorites]


Here in Pittsburgh, the main public transit problem is a "final mile" one.

You know, say what your will about the electric scooters popping up in major cities, but they do make a good stab at solving the last mile problem (for a younger and wealthier subset of folks).
posted by davejay at 2:41 PM on July 21 [2 favorites]


...who live where there is good weather and don't really need to carry anything with them.
posted by Brockles at 2:44 PM on July 21 [2 favorites]


over-dense

You keep saying this. The evidence suggests that your opinion, that density is undesirable, is far from universal.
posted by asterix at 2:49 PM on July 21 [9 favorites]


It's still all pie in the sky folks.
posted by GallonOfAlan at 3:03 PM on July 21 [1 favorite]


Did anyone else think their self-driving car wasn't a very good driver? In the video starting at 5:48, when it wanted to make a left turn, because the sensors are unable to see that two oncoming cars were indicating a left turn and not continuing straight, it waited in the intersection instead of turning left. Following that, the highway merge operation was very poorly done. The founder guy says "that would have been a hard situation even for a human" but it didn't look complicated at all from the different views they edited in. Any human driving behind this robot car in those situations would probably mutter obscenities at its poor driving.

I'm saying this just having taught my nephew how to drive. That autopilot performed far worse that my recent times riding with a completely inexperienced and timid beginner. If they're not doing a great job in easy mode in the US, how are they going to understand traffic in China or Thailand?
posted by peeedro at 3:23 PM on July 21


Banning won't work (shunning may eg if Millenials turn against them), history is littered with workarounds from bans. But we should demand the hardware and software (is there a difference with these things??) be fully open. Robot cars are like a physical embodiment of facebook, not something you think you need, but something you really do need, an incestuous, linked-to-you umbilical (or leash).

Daniel Wilson's Robopocalypse where an AI decides it's going to kill via our car culture seems plausible (mainly driving till occupants starve/dehydrate and hit-and-run) - maybe not an AI but a hacker as per nikaspark comment above. I won't be buying one.

There are simply too many folk on this planet that see technology as a solution, but ignore culture, anthropology, ecology and biology (and good urban planning) - where - IMO - the real answers are situated.
posted by unearthed at 3:54 PM on July 21 [3 favorites]


I didn;t think the merge was that easy if what he said was true - he says their car tried to accelerate past the car in the inside lane (which should have moved over if it wasn't being selfish) but the car on the highway didn't back off at all and held position and speed, so their car had to slow to merge behind it as the on ramp was ending.

That's not that uncommon a scenario, and really the only way to make that more easy would be to programme a level of aggression (harder accel from further down the on-ramp) or slow down from much further back (can it even see that far?). I can't imagine either is optimal, but it was a 'will they, won't they' kind of decision making process for the autonomous car so it wasn't too bad.

The left turn was less good, but I think that defaulting to over-caution is a much smarter position than anything else. With a level of trust in the black car I'd have gone immediately as they were also turning left, but I'm wondering if the white car came into sensor range as it started to do that (hence the sudden brake) so it had to pause to establish speed and likely path. In short, not terrible, but not as efficient as a human (competent) driver, but also not any bad decisions.
posted by Brockles at 4:00 PM on July 21


I don't know; if properly done autonomous vehicles could also be used for public transportation, perhaps in way that make it more attractive or accessible

There are some serious physics problems with autonomous vehicles as public transit.

They simply cannot move enough people fast enough at peak times to be effective. They do nothing to solve the problem of rush hour. Yea, a machine is driving, but a road only has so much capacity, especially in single person cars. And they’re all stopping at intersections it’s even worse.

If you want good mass transit built grade seperated BRT or
Rail system. It’s cheaper and it exists right now. Yea it’s mostly tech from the 1800’s, but gosh darn it it works. You can even automate them, no driver, now, using 1980’s technology.
posted by jmauro at 4:10 PM on July 21 [10 favorites]


As an example of an unpredictable case, I asked him whether he thought modern systems could handle a pedestrian on a pogo stick, even if they had never seen one before. “I think many AV teams could handle a pogo stick user in pedestrian crosswalk,” Ng told me. “Having said that, bouncing on a pogo stick in the middle of a highway would be really dangerous.”

I would like to note how he subtly changed the scenario from someone pogoing on a city street or a sidewalk near a park to someone pogoing on a highway, which you know wasn’t what they asked, because pogoing on a highway is insane so no one would do it anyway. His implied solution, was to then to suggest that we should just ban humans from doing that near a road, which sort of defeats the purpose of the cars or cities for that matter.
posted by jmauro at 4:15 PM on July 21 [6 favorites]


I don’t see much alternative to the subway/underground in dense megacities like London, New York and Hong Kong. But not all cities are so dense and can afford such expensive mass transit. If it were possible to significantly reduce the footprint of single-person self-driving vehicles and to also develop self-driving minibuses, I think those would be an attractive solution for cities like, say, Edinburgh. And there’s no reason why the network couldn’t be publicly owned, either.
posted by adrianhon at 4:38 PM on July 21


Self-driving cars correspond to many complex software projects -- the first 90% is easy but the last 10% becomes exponentially harder the closer you get to the end.

So already we are hearing self-driving proponents saying that if they can't make cars that can fully understand the world, we will reform the world's infrastructure to make it simpler for cars to understand -- and at the very same time say that there's not enough money for comprehensive mass transit.

They want to make up for the shortcomings of self-driving cars by putting restrictions on everyone else.
posted by JackFlash at 4:40 PM on July 21 [10 favorites]


It's definitely pie-in-the-sky. There is a major interchange here in Orange County that these vehicles would have massive trouble with. The 405 South hits the 55 East/West and splits off to the 73 South Toll all roughly at the same spot. People traveling South think the freeway has given them one option to go South. Every sign they've seen so far says "405 South San Diego". Nope! Now the sign for the right 3 lanes says "73 South San Diego" and the sign for the left 4 lanes says "405 South San Diego". Two options. Brain overheats. And you have 2 major surface streets dumping new traffic that has to immediately beeline across 4 lanes if they don't want to take the toll road (most don't). Most people don't panic and are just worried they are now on the wrong freeway, but some do panic and it causes all kinds of problems. It's massive amounts of lane changing mixed with panic every second.
posted by Brocktoon at 5:40 PM on July 21 [2 favorites]


I don’t see much alternative to the subway/underground in dense megacities like London, New York and Hong Kong. But not all cities are so dense and can afford such expensive mass transit. If it were possible to significantly reduce the footprint of single-person self-driving vehicles and to also develop self-driving minibuses, I think those would be an attractive solution for cities like, say, Edinburgh. And there’s no reason why the network couldn’t be publicly owned, either.

It's going to be cheaper and flexible for a long time to just buy a regular old mini-bus and hire a driver. Having it self-driven is really not the problem. It's getting people to want to use a bus over driving themselves, until that happens the self-driving portion of a bus is more or less moot. We've had years of both explicit and implicit basically looking down on bus riders as something you don't want to be.
posted by jmauro at 6:28 PM on July 21


Most people don't panic and are just worried they are now on the wrong freeway, but some do panic and it causes all kinds of problems.

AI panic attacks are going to be TERRIFYING
posted by wemayfreeze at 7:57 PM on July 21


Cars can barely coexist with pedestrians today, and autonomous vehicles, so far, have continued to be deadly.

Has Waymo killed anyone? Tesla, yes, but they don't even have LIDAR. Uber, yes, but they move fast and break things because they're unethical little shits. If Waymo's killed anyone, *that* is what would make me reconsider the merits of the tech, because they at least seem to be the ones that are doing it right.
posted by Jpfed at 9:02 PM on July 21


A good deal of the issue of public transportation comes down to politics and political inertia. The infrastructure (namely roads) for cars/trucks/busses is already there, and anyone who thinks that there is significant political support for expanding and upgrading existing non-road-based public transportation infrastructure should read up on what's actually happening in the real world.

In New York, the majority of the state is cheerfully denying the New York systems the funding needed to even maintain the subways and metro rail; Congress refuses to expand and properly maintain the DC Metro -- as someone in the DC area, I'd like a system that doesn't shut down for most of the night and doesn't catch on fire regularly.

As a result of this political opposition, prices go up and up, quality goes down, and on the whole I know a number of NYC natives and DC area folk who refuse to use systems that can't be relied upon to transport them safely and on-time to where they need to be. NYC already has lines which simply cannot run enough trains to handle the load during rush hour, the bus system shares the roads with private and taxi services so they crawl during high-usage times. In DC the case is much the same -- try getting a decent parking spot at any of the near-city metro lots after 8am, good luck with that in most cases. The best solution I could find during a time when I was commuting in was to drive to a bus, park and take the bus in to the metro(with associated delays due to traffic most days even with the bus-only lanes), then walking the mile-and-a-half to the office. Or I could just drive in and park near work, which took about the same amount of time. When working in Manhattan, I either took the subway (when I was living in Brooklyn) or the metro (when I moved to a suburb), which was nice -- except when parking was full at the metro, or it was raining/snowing/sweltering for the mile or so of walking to the subway station.

Now that I live in a more rural area of Virginia (it's hard to run a smithy in a city nowadays, between noise regulations, rent, and landlords iffy on the whole flame-belching forge and flying sparks from the grinder thing), I note that while in theory there is public transport around, there's nothing useful such as, say, a bus route to the metro stations running every hour or so, or more locally something that runs more than a couple of times a day to/from population centers to retail/supermarket locations. If I lost my license, I'd certainly have to move to maintain my standard of living. I sometimse give lifts to people I see waiting for the bus, since they usually have an hour or more to wait for the next one.

Sure, it would be nice, pie-in-the-sky wise, if we had more public transport infrastructure, but right now it just isn't in the cards until the people opposing it, mostly rural conservatives who constitute the majority of almost all state governmental elected officials nowadays, see something that they get out of it for their support and their constituents tax dollars. What little support for new transport is almost always more or larger buses, because that's the least cost today approach -- building permanent infrastructure is seen as a waste, because the roads are already there. It doesn't help that there is at least a perception, if not a reality, that any big public infrastructure project will take longer, cost more, and become effectively a way for big business to suck tax dollars from the government teat who never quite deliver the promised results.

Self driving cars, should they ever become a common reality, will certainly change some of this -- but more likely by the synergy of vehicles that can communicate and change routes due to conditions, sort out optimal flow patterns, and better use existing infrastructure. When private self-driving vehicles actually move out of the way of larger, more efficient buses, and manually driven vehicles are restricted to other, slower lanes, then maybe we'll see a change. More likely, the rich will force everyone else to the slow lane, and a mostly-empty lane reserved for those who can afford a steep premium will be the only rapid transit around -- much like DC already did that on the Beltway and other local highways, most likely soon to be expanded to every major road.
posted by Blackanvil at 9:03 PM on July 21


Sure, it would be nice, pie-in-the-sky wise, if we had more public transport infrastructure, but right now it just isn't in the cards until the people opposing it, mostly rural conservatives who constitute the majority of almost all state governmental elected officials nowadays, see something that they get out of it for their support and their constituents tax dollars

This is possible, but self-driving cars don't fix any of this. Virtually all their advantages come from making single-car transport possible for the non-driving and from increasing safety. These are meaningful advantages; as a non-driver myself, I would like to see the tech deployed in a safe, effective way. But trading a self-driven car for an autonomous one does little to nothing for traffic congestion.

Meanwhile, trading a human-operated bus for an autonomous one with a flexible route? First of all, you're still stuck with "bus stigma." Maybe that will go away under new systems. Let's be optimistic about that. Second, and much more importantly, once you are accommodating multiple passengers, you are just reintroducing the wait and inconvenience problems. I feel like many people pitching these systems as good mass transit options are relying on public ignorance of how buses actually function. Routes do to some extent reflect traffic patterns. If the bus doesn't run up the hill from the main road, part of the problem is cheap bastards doing the funding and planning, but part of it is also usually that not many people live up the hill. If the bus is going up the hill to drop off one person, that means another three people on the main road are waiting for it while it detours. If the bus is to actually continue to function as mass transit to reduce congestion, that is, not routinely and wastefully just carrying the one passenger up the hill, where there isn't already considerable density, it's going to have to force people into circuitous routes and delays. Where there is, it's very likely going to be reproducing existing routes.

I see some potential for recognizing population redistribution on the ground ahead of formal government recognition of same. I see some potential for responding more easily to hourly, weekly, or seasonal changes in traffic patterns. Both of these seem good. But autonomous buses are still going to be functioning with the drawbacks of buses--and if you think people don't like waiting for buses now (they don't), they aren't going to be liking them a whole lot better when they're waiting for them without dedicated shelters.

In more lightly-settled outer suburbs with shitty public transit and less of a congestion issue, autonomous buses might help with the frequency problem--maybe (remember, the bus still has to get from wherever it is to wherever the rider is, while maximizing ridership [or giving up on the bus concept]). Most likely in areas where there already is a real traffic pattern not addressed by the current transit authorities, which I think is becoming more common as these suburbs fill in with poor people without cars for whom they weren't designed. Getting people from the down-at-heel rental complex to the shopping center with the jobs, or the supermarket, a bit faster could be a genuine boon for people. But, for that population, affordability seems like a real issue. If the traffic remains relatively light, what is the actual cost-per-ride going to be? And if the local government won't pay for buses, why would they pay to subsidize this form of transit?
posted by praemunire at 11:20 PM on July 21 [6 favorites]


Problem: Autonomous car detects you as a pedestrian or cyclist, bullies you out of the way.

Solution: Adversarial glasses that make cars think you're a transport truck.
posted by clawsoon at 5:33 AM on July 22 [1 favorite]


Blackanvil: I know a number of NYC natives and DC area folk who refuse to use systems that can't be relied upon to transport them safely and on-time to where they need to be. NYC already has lines which simply cannot run enough trains to handle the load during rush hour

As another famous New Yorker put it, "Nobody goes there anymore. It's too crowded."
posted by clawsoon at 5:40 AM on July 22 [1 favorite]


In all these conversations about self-driving cars, has anybody addressed the insurance problem? Namely, what will it take for big insurance companies to want to insure fully self-driving cars? Because to me, this seems like the biggest obstacle to mass adoption.
posted by panama joe at 7:59 AM on July 22


Insurance is really the least of the issues. For an insurance company it is quite simple. Just as now, they use statistics to classify various drivers for their probability of generating a claim and adjust premiums accordingly. Insurance companies will likely charge higher premiums for self-driving cars initially to compensate for statistical uncertainty due to lack of data. As they gather more real life data, they will adjust premiums accordingly. If self-driving cars generate fewer claims, as proponents suggest, their insurance rates should be accordingly lower. If, on the other hand, self-driving cars turn out to have a higher rate of claims, their premiums will go up.

It's not a difficult issue for insurance companies and no different from insuring anything else -- cars, house fires, floods, medical malpractice, etc.
posted by JackFlash at 8:18 AM on July 22 [2 favorites]


So if an insurance company wanted to offer self-driving car insurance, wouldn't they first need some idea of how likely these cars are to kill somebody? How well do insurance companies understand self-driving car algorithms? Since life and death are on the line here, wouldn't premiums have to be astronomically high? Also, wouldn't you need to have a whole lot of people buying this insurance in order to create a big enough risk pool such that premiums could sink to a reasonable level?

Asking because I genuinely don't know.
posted by panama joe at 8:50 AM on July 22


Since life and death are on the line here, wouldn't premiums have to be astronomically high?

1.25 million deaths per year worldwide by human-driven motor vehicles, as a starting point of comparison.
posted by ambrosen at 9:00 AM on July 22 [2 favorites]


True, but that is a known quantity. Does anybody really have a good idea what would happen if self-driving cars were unleashed en masse tomorrow?
posted by panama joe at 9:03 AM on July 22


Life and death are on the line with regular cars too. I could see insurance companies starting with either the current average rate for all drivers (that means no discounts for good grades, no accidents, etc) and then maybe adding a certain amount to that rate for uncertainty.

Insurance companies insure stuff all the time that they either have bad or no data for (see the current situation in LTC insurance as an example. Relevant article starts at page 20) It can go good or bad for them, but the more uncertainty should mean higher premiums. For consumers, this can be an issue and if no one buys the insurance except those who need it, that’s not good. However, if it’s companies buying this insurance instead, there’s a little more leeway in the price of premiums.
posted by LizBoBiz at 9:10 AM on July 22 [1 favorite]


Given where the technology is at today, I think releasing them en masse tomorrow would be horrific.
posted by LizBoBiz at 9:12 AM on July 22


Mass transit is not the answer (and only really works in the densest cities like SF, NY, London etc)

[citation needed]

It's like pure gay luxury space communism - sustainably-funded public rail transit has never actually been tried in the United States. I would gladly knife fight Elon Musk nude in a pit for the chance to see even a tenth of Tesla's VC go to e.g. Amtrak.

We go through this over and over again every time this comes up. The only problems autonomous cars as currently conceived are even *supposed* to solve are all things like safety, convenience, and efficiency - things that they so far haven't shown any evidence of solving, and for which a little back-of-the-envelope math usually tends to show them to be pretty ineffective solutions. Especially when the hardcore boosters tend to be pretty inarticulate about what those problems are.

Mixing fast-moving human transport vehicles in with heavier freight vehicles, pedestrians, and single-person conveyances such as bikes, scooters, and wheelchairs is inherently unsafe and cannot be made safe. People are messy and capricious and tangential and easily squished. Unless every vehicle on the road is autonomous and operating in its own separate infrastructure, people will continue getting killed this way. And at that point you might as well just have trains.

The convenience part sounds nice, but in most places results in diminishing returns - anecdotally DC traffic was made even worse by "ride-sharing;" there are just that many more vehicles on the street, which means everything is slower. There's not really a way to solve that problem with AI; that's laminar flow and physics.

Efficiency-wise, the one realistic gain I can see based on existing technology would be in delivery logistics - a ton of smaller vehicles would be easier on existing streets than the current fleets of box trucks. But having seen the current prototype for something like that here in DC, not only is it still in the "needs a human to follow it everywhere" stage, the idea itself has a ton of very messy problems to solve.

For one thing, I'm pretty sure it would take me about thirty seconds to throw a bucket of paint over it to cover the sensors, kick it over onto its side to keep it from running away, and pry open its belly to get at whatever fun bits are inside. Or for that matter, if you've ever met an American, you know people would knock these things over just for the fuck of it. We saw what happened with dockless bikes around here.

The other obvious gain from the capitalist perspective is getting rid of the bunch of messy, wage-requiring legal liability meatsacks currently helping Uber et al strip mine the sharing economy. My feelings on that are . . . mixed.
posted by aspersioncast at 9:34 AM on July 22 [6 favorites]


And yes, insurance is an inconsequential problem here. The US insurance industry would be delighted to insure more vehicles; there were an outsized number of auto claims last year, but the industry as a whole is in a $704bn surplus.
posted by aspersioncast at 9:41 AM on July 22


So if an insurance company wanted to offer self-driving car insurance, wouldn't they first need some idea of how likely these cars are to kill somebody?

If the industry that had enough data to insure David Lee Roth's penis doesn't have enough actuarial sophistication to offer rates for self-driving cars out of the gate then their widespread public-road use will simply start in places which don't require insurance and that's where the adequate initial data will come from.
posted by XMLicious at 10:56 AM on July 22


Metafilter: surprisingly full of luddites.
posted by Revvy at 10:57 AM on July 22


There are some serious physics problems with autonomous vehicles as public transit.

They simply cannot move enough people fast enough at peak times to be effective. They do nothing to solve the problem of rush hour. Yea, a machine is driving, but a road only has so much capacity, especially in single person cars.


That makes absolutely no sense at all - I do not understand even slightly why people are fixated on a self driving car (and all its related tech) can only be applied to the selfish use case of one person and their car, sat in there on their own. Self driving cars are seemingly being developed first (which, is actually not true anyway) but it really is just more visible and more attractive to investors (and reporters), but with autonomous technology also inevitably comes autonomous public/mass/group transport. The second the same tech is put in a bus, your argument becomes invalid, so trying to ban the first steps of something you actually put forward as the right answer makes no sense. In fact, self driving cars developing and refining the technology brings autonomous public transport forward in likelihood because of the increased volume of data and R&D funding. There will be minimal differences in programming required between self driving cars and self driving buses once the 'don't hit anything and don't kill anyone' stuff is ironed out.

Will it fix public transport overnight? Of course not. But there is no point declaring stage 1 of a twenty step plan to be pointless because we can't immediately jump to stage 20 after it.

If you want good mass transit built grade seperated BRT or Rail system. It’s cheaper and it exists right now Why? Why not just paint different coloured lines on one lane and just allow only autonomous cars, buses, trucks, shuttles and other vehicles only to use it. Like an HOV lane and bus lanes are now? By making it easier to predict for Autonomous vehicles that makes them better as an option for people, which makes it more annoying to drive around them, which encourages their usage. Because they're just .... easier.

Aircraft are largely automated and could be even more so. They have to deal with weather variable and wind changes which are far harder to predict than people and cars, yet they do fine. Sure, there is some level of order required (approach paths, flight paths etc) but that's no more involved than the existing road system with minimal modification.

In New York, the majority of the state is cheerfully denying the New York systems the funding needed to even maintain the subways and metro rail; Congress refuses to expand and properly maintain the DC Metro

This is the single most realistic argument why this will never work in the US- the obsession with capitalism that allows crusty old twats sitting on stacks of money to justify no project for the greater good ever happens, because their mates may not make 300& profit if they do. But let's look globally, because the US is already utterly broken in that regard.
posted by Brockles at 11:16 AM on July 22 [2 favorites]


Why not just paint different coloured lines on one lane and just allow only autonomous cars, buses, trucks, shuttles and other vehicles only to use it.

And here we go again with appropriating property and infrastructure costs from the public to accommodate the shortcomings of self-driving cars and at the same time saying we don't have enough money for mass transit.

It isn't clear what problem they are trying to solve, but it's likely to involve massive subsidies for corporate profits.
posted by JackFlash at 11:28 AM on July 22 [4 favorites]


at the same time saying we don't have enough money for mass transit.

We do have the money for mass transit in the first world. Every single 1st world country does and pretty much always has had. It's lobbyists preventing it getting spent, not lack of funds.

The idea behind changing the infrastructure (if you read the post) was to make it easier for autonomous vehicles to be accommodated within the existing road system with minimal changes and so use the cheaper vehicle options they would provide to increase the options for mass transit within the same budget.
posted by Brockles at 11:47 AM on July 22


Interestingly-organized article on autonomous vehicles vs. mass transit; starts with people arguing that we should be paving over tracks now to allow for autonomous vehicles, pauses at city-planner-type transit wonks pointing out the geometric differences, finishes with Uber:
No system of autonomous cars could be more efficient than the New York subway, said Andrew Salzberg, Uber’s head of transportation policy and research. Uber needs that transit, just as it will need electric scooters and bikes and the congestion pricing it also supports in New York to ensure that cheaper transportation doesn’t simply lead to more traffic.

Or you could just reread Synners for GridLid.
posted by clew at 11:53 AM on July 22 [2 favorites]


It isn't clear what problem they are trying to solve, but it's likely to involve massive subsidies for corporate profits.

I thought one of the problems they were trying to solve the problem of the need for car ownership in places where mass transit/taxis/etc. are not available but cars a required for daily life? I grew up in a town with 30,000 people. The only available public transportation is only for the elderly or handicapped. Taxis do not exist there. You must have a car to get around there or you’re stuck walking (15 mile trips are totally normal everyday occurrences) or begging for rides. . Self-driving cars make it possible to have an Uber-like system where you call a car when you need it.

My wish is that the other problem they are trying to solve is me having to actually drive on long road trips. Remember, Amtrak is not really an affordable option when compared to driving yourself. I would much rather read or look at scenery or sleep than have to drive across hundreds of miles of nothingness.


I feel almost like this is one of those technologies that gets invented where people think it’s useless and for lazy people, when actually it could really help people who have difficulty with aspects of daily life. It’s been mentioned a couple times above but seems to have been ignored, but self-driving cars could really help people who don’t live in dense cities. Uber might be filling this need now but self-driving cars could remove at least some of the exploitative aspects of the service.
posted by LizBoBiz at 12:27 PM on July 22 [4 favorites]


Remember, Amtrak is not really an affordable option when compared to driving yourself.

Amtrak is super-underfunded, but this assertion really depends on where you going and if you *don't in any way factor in the cost of owning/parking a car*. I can go between DC and Denver on a train for a little under what it costs to fly - counting gas and lodging it would end up costing me more to drive, unless I were to sleep in the car. It winds up being cheaper for me to take the train if I'm going anywhere between here and Boston on the 95 corridor, because of tolls and the cost of parking the vehicle somewhere once you get there.

Metafilter: surprisingly full of luddites.
Every single time we have this conversation someone accuses everyone who doesn't think this is the most amazing technology and the future of robot cars is here of being Luddites. You know, the actual historical Luddites smashed the machines that replaced them. But it's not the machines themselves most MeFites seem to be skeptical of, it's the credulous boosterism that perennially comes up with a shiny new hammer and then sees nails everywhere. I think AI is frickin awesome, I'm just really skeptical of this application of it, as much because of the primary actors and funders involved as for any other reason.

Thinking critically and relationally isn't Luddism--it's kind of a hallmark of science.
posted by aspersioncast at 3:20 PM on July 22 [6 favorites]


If it were possible to significantly reduce the footprint of single-person self-driving vehicles and to also develop self-driving minibuses, I think those would be an attractive solution for cities like, say, Edinburgh. And there’s no reason why the network couldn’t be publicly owned, either.

Kind of a weird example, since Edinburgh isn’t in obvious need of a “solution”.

While overall Edinburgh is not very dense for its size (seventh largest city in the UK but ~20th in terms of density), the density isn’t geographically consistent - a lot of the population is concentrated relatively close to the centre, in tenements, or in “schemes” on the outskirts. (Edinburgh has one of the largest proportions of people living in apartments, rather than houses, in the UK, even in comparison to London.) There’s a lot of green space between dense neighbourhoods.

It’s a highly pedestrian- and bike-friendly city, and it’s served by an extensive network of buses, which are already publicly owned and operated by (the excellent and repeatedly award-winning) Lothian Buses.

Admittedly the city’s recent foray into light rail was pretty badly planned and implemented, but that’s not a problem that would have been solved by waving autonomous vehicles at it.
posted by chappell, ambrose at 5:00 PM on July 22 [1 favorite]


All up and down every street by my house, there's dockless cars everywhere, and only one of them can I actually use. For the most part people leave them alone.

I don't think it takes a lot of imagination to see a world where very few people own cars, because robot taxis are the norm. There's absolutely less congestion, and they're probably electric so air quality is much better. They don't get drunk or text while driving, so traffic fatalities, while not zero, are down sharply from the 40k or so that die every year from human drivers.

Right now, using the same magical glowing rectangle I'm currently poking away at, I can summon a car to whisk me anywhere, and I probably won't have to wait more than a few minutes. I have the option of sharing the car with someone else if I want to save a few bucks, paying a little more to ride alone, paying yet more if I want to ride in a fancy car, or paying a bunch if I want an SUV. So that side of things is "solved"; dropping in a robot taxi is simple.

Will the technology get there? 15 years is a long time. It's perhaps cliché, but think of your phone 15 years ago. Coincidentally, that's roughly when the first DARPA Grand Challenge was. The car that did the best in that competition made it all of 7 miles before getting stuck on a rock.

The robots are coming. Robot taxis (actually probably robot semis) will be the one of the first large scale mass unemployment events of the coming automation revolution.
posted by booooooze at 5:23 PM on July 22


The robots are here. At about 20:30 in this Deutsche Welle documentary published back in 2016—“Robotics - Impacting the workplace” —see a German shipping container seaport where there are autonomous vehicles somewhat like a flatbed truck with no cab which cranes drop the cargo containers onto as they're offloaded from the ship, which then drive the containers across the port to the next stage of their journey.

It's the default for all new ports under construction: for example the Tuas megaport will gradually supersede and replace all other container ports in Singapore in the coming decades.

A major carve out from these bans on autonomous vehicles will be the for the ones operating within cities doing things that are to the benefit of city dwellers, especially things like the Singapore project freeing up untold acres of high-priced real estate. And I can't imagine city dwellers are going to go on vacation and oblige themselves to only take public transportation everywhere. So it'll be people in rural non-tourist-destination areas who will need to forgo autonomous vehicles for the sake of adding impetus to public transport projects in those places it's so economical to travel between on Amtrak.

Until we have ridable quadruped robots that won't leave you quite as saddle-sore as the current Boston Dynamics offerings, at least. Maybe once there are robo-ponies and robo-unicorns city dwellers will deign to allow the rest of us to benefit from autonomous vehicles. Which is okay—it will take me quite a while to come up with a suitable name for a robo-pony.
posted by XMLicious at 7:54 PM on July 22 [1 favorite]


RoboClop
posted by sfenders at 9:26 PM on July 22


And still no flying cars...
posted by Billiken at 5:14 AM on July 23


overglow: There’s growing concern among AI experts that it may be years, if not decades, before self-driving systems can reliably avoid accidents. As self-trained systems grapple with the chaos of the real world, experts like NYU’s Gary Marcus are bracing for a painful recalibration in expectations, a correction sometimes called “AI winter.” That delay could have disastrous consequences for companies banking on self-driving technology, putting full autonomy out of reach for an entire generation.

I was looking for an article I could have sworn was from Wired, which indicated that in the wake of Uber's fatal crash that killed a pedestrian, Elaine Herzberg, in Tempe, Arizona, a recent autonomous vehicle summit was focused on safety rather than innovations, as was the case the prior year, and the glowing forecasts for autonomous fleets were heavily downplayed. In other words, it sounded like an early pause-and-reflect period for the industry. The report is that the crash was wholly avoidable, and that the operator was watching a video at the time of the incident, as reported by CNN on June 22, 2018. Also from that article
The crash was a major setback for Uber's self-driving car operation. In May, Uber said it was ending self-driving car testing in Arizona and laying off 300 Uber workers. The company said it would focus on its autonomous vehicle efforts in San Francisco and Pittsburgh.
Emphasis mine - they're not going to stop their efforts, but continue to work in much larger cities.


sfenders: 67 years: Cars become the dominant form of life on the planet. People are still kept around as pets.
...
285 years: The internal combustion engine is re-invented, fuelled by plastic mined from the ancient landfills.


500 years, give or take: the Cars franchise takes place, on a planet devoid of humans, unless you go with the more disturbing theory of Cars.

700 years: people come back, as seen in the events in Wall-E.


XMLicious: The robots are here.

But that example is in a very controlled scenario: commercial ports, where the idea is to know where each and every one of thousands if not hundreds of thousands of cargo crates are at a given time, and where they're going next. Similarly, in terms of controlled locations, are mines and some other industrial applications. Limit the hazards to people, increase operation reliability and production.

In order of controlled systems, highways are the next most logical place for autonomous vehicles - all vehicles are going the same way, with limited ingress/egress points, which is where U.S. on-road freight trucks are already doing this, with human oversight. Earlier this year, Embark's self-driving truck drove from L.A. to Jacksonville, Florida, driving 2,400 miles from one end of the U.S. to the other, apparently using on-board sensors only, no pre-loaded maps. As reported/ promoted on Tech Crunch,
Embark’s goal isn’t t replace the driver entirely, however: It just wants to make it possible for long-haul trips to be managed by fewer drivers, eliminating the need for team driving, for instance, and helping to address a lack of available qualified human drivers for this kind of shipping. Drivers are still expected to help with the parts of the route that don’t involve freeway and driving, but still the efficiency gains and trip time benefits would be huge once their technology is in service.
Emphasis mine.

Meanwhile, How Cars Divide America -- Car dependence not only reduces our quality of life, it’s a crucial factor in America’s economic and political divisions. (Richard Florida for City Lab, Jul 19, 2018)
Urbanists have long looked at cars as the scourge of great places. Jane Jacobs (previously) identified the automobile as the “chief destroyer of American communities.” Cars not only clog our roads and cost billions of dollars in time wasted commuting, they are a terrible killer. They caused more than 40,000 deaths in 2017, including of some 6,000 pedestrians and cyclists.
Cars are already deadly. Making them autonomous will likely make them safer, because people already do so many dumb things. And when a single autonomous vehicle can alleviate a "phantom" traffic jam, the ripple effect of following too closely and breaking suddenly, the technology adoption forecasts from University of Texas at Austin (research paper from 2016/17, PDF) with a conservative estimate of 24.8% Level 4 (autonomous in most scenarios) AV penetration by 2045 and UK’s Centre for Connected & Autonomous Vehicles (July 2017, "glossier" PDF) much more rosy projection of ~35% Level 3, ~35% Level 4/5 adoption by 2035 in the US, which would be a bit lower in Europe, and higher adoption rates for Level 3 in Asian and ROW but lower for Level 4/5.

This report is weird, because Level 3 is the tricky one - In January 2016, Tech Report's update on levels of autonomy from 0 to 5 include this quote:
Jim McBride, autonomous vehicles expert at Ford, said this is "the biggest demarcation is between Levels 3 and 4." He's focused on getting Ford straight to Level 4, since Level 3, which involves transferring control from car to human, can often pose difficulties. "We're not going to ask the driver to instantaneously intervene—that's not a fair proposition," McBride said.
In other words, the idea is that the human driver can check out for the most part, but will be asked to re-engage when the vehicle doesn't know what to do in a safety emergency. This is why most tech companies are similarly skipping from Level 2 to 4, because that "partial engagement" is likely to be more troublesome (or deadly) than helpful.

Looping back to the support for public transit: as a transportation professional, my dream is a ton of investment into public transit. I'm imagining high-speed transit on central corridors, with 10-15 minutes between vehicles, with progressively smaller vehicles providing access to less or less traveled areas. But to do this means 1) people need to get away from their own vehicles and 2) be OK with some indirect routes and possibly longer commutes. To achieve that, transit needs to be sold as a way to "cut the car taxes" that so many people expect to pay. Back in the earliest days of cars, we had the "mud tax" --
The tax of bad roads, or, as it is sometimes called, the "mud tax," is levied on all alike, the rich man as well as the poor man, the man who lives in the country as well as the man who lives in the city, the producer as well as the consumer.
(Google books preview of Good Roads Magazine, Jan. 1907). Now we take paved roads for granted in much of the world, but we also take the expenses of maintaining a car for granted - leasing or borrowing to own a car, paying for gas, tires, and maintenance, paying insurance - it's a big cost for the "freedom" to get stuck in traffic.
posted by filthy light thief at 9:55 AM on July 23 [3 favorites]


Ordering a self-driving car sounds pleasant in theory, but I've yet to understand what I'm supposed to do with the car seats. Setting aside the hassle of installing again and again in whatever weather, I can't carry them and the small children around wherever we end up. Will there be a selection of installed carseats to choose from, for various age ranges in various combinations? At the needed volume per city?

I want public autonomous travel, but individual cars to borrow never have carseats explained that I've found.
posted by hapaxes.legomenon at 6:33 PM on July 23 [2 favorites]


Are infants allowed in taxis? I really don't know. Perhaps if autonomos taxi/ubers become pervasive there will be an option to order one with child seats?
posted by sammyo at 7:14 AM on July 24


Huh, I always assumed most cabs just carried a car seat in the trunk, but I actually have no idea. Does Uber have anything in the app to let parents know?

I guess that's pretty tangential to autonomous vehicles.
posted by aspersioncast at 9:25 AM on July 24


In Toronto, taxis are considered a form of public transportation. So just like on a bus, passengers are not required to use a car seat in a cab.

But Uber and Lyft drivers must have a car seat to drive around a child younger than eight, less than 80 pounds and shorter than four feet nine inches tall.

Ride-sharing drivers are not required to travel with a car seat though, so many don’t.
posted by clawsoon at 3:51 PM on July 24 [1 favorite]


Watching a ½-hr 2012 documentary posted at YouTube by Arizona Public Media, Seeking Water from the Sun, about academic efforts to miniaturize solar water distillation/reclamation technology for use on the Navajo and Hopi Nations, self-driving delivery vehicles seem like they would be a good supplement to that effort. It shows people currently driving 50 to 100 miles round-trip in SUVs and pickup trucks to haul water to their homes. (Presumably, indigenous peoples are at least one case where we wouldn't insist on banning of autonomous vehicles and centralization instead... I would hope?)

I'd also think it would be an interesting inversion of the problem of delivery vehicles carrying commercial good being robbed, since if a water-carrying vehicle is approached by people in the middle of the desert it seems like you would want it to give them water. The documentary also shows that the springs and wellheads where people are currently driving to retrieve water from have troughs to permit wild animals to drink.
posted by XMLicious at 5:47 AM on July 27


Waymo pilot program shows how self-driving cars could boost transit
Self-driving cars could strengthen public transit—or replace it.
As a transit official working on self-driving car projects, Bamonte has thought a lot about how to integrate public transport and self-driving cars. And he told me that there are two basic models for transit in a self-driving future. One possibility is that conventional bus routes will gradually be replaced by fleets of self-driving shared vans that operate along flexible routes, eventually rendering conventional buses—and perhaps even conventional subways, commuter rail lines, and street cars—obsolete.

But the other possibility, Bamonte told me, is a "first-mile, last mile" model. In this model, self-driving cars could complement transit by helping more people get to conventional transit stops. If you're a suburbanite who works downtown but lives a mile or more from the nearest bus or subway stop, taking transit may not be a very appealing option. Theoretically, you could take a conventional taxi to the nearest transit stop, but taxis today are far too slow and expensive for this to be practical.

Self-driving cars could change the economic calculation. With no driver to pay, a self-driving taxi ride could be much cheaper than a conventional taxi. And with no driver, a self-driving car might be more willing to show up a few minutes early, ready to go as soon as the customer is. Hence, self-driving cars could actually expand transit use.
Having gotten heat stroke the other day, for the first time in my life, after attempting to walk home from the nearest public transit terminus when I unexpectedly found myself without a ride home, I can definitely say that I would probably start using public transit regularly—which I've never had the opportunity to before—if there were a cheaper or easier way to go those last few miles. I selected my walking route so as to pass by the headquarters of a limo service and just barely made it there before collapsing, but their entire fleet was out (wrong time of day or unusual demand or something?) so I had the unusual experience of asking a cab company to call me a cab... which cost me more than ten times as much as the bus fare had to travel the same distance.
posted by XMLicious at 4:57 PM on July 31 [2 favorites]


« Older We ban it immediately   |   Die With Me Newer »


This thread has been archived and is closed to new comments