My immediate thought is a dystopian version of this article:
If all cars were autonomous, society, the economy and the environment would all suffer. Here is how…
* You will be forced to watch non-stop ads in these automotive cars, similar to how some taxi services work today. You can turn these off for a fee of course.
* Hackers or bugs in softawe may easily cause global traffic jams
* Congress will pass laws controlling when, where and how you may travel
* You may be blacklisted from traveling in any autonomous by corporations, governments, police etc.
* The aggregate amount of automotive travel will increase exponentially, causing more carbon emissions and even worse traffic since more people will have access to travel
* Sprawl will get worse, as people move further away from urban areas since they can do other activities in cars.
* Investment in mass transit will diminish, since funds will move towards automatic cars.
Exactly. Don't forget the inevitable regulations on the software too, or even restricting/banning non-autonomous cars completely. I'm glad there are others out there who also see autonomous cars as a threat to freedom.
It is said that driving is one of the most dangerous activities we do, but I think it's no coincidence that it's also one of the most liberating. It's a very pleasurable feeling to be able to get into a car and drive, controlling it as if it were an extension of your body.
The phrase "If all cars were autonomous" has a similar feeling to others like "if there was no crime" --- sounds good on the surface, but also has deeply authoritarian and dystopian implications.
> It is said that driving is one of the most dangerous activities we do, but I think it's no coincidence that it's also one of the most liberating
I experienced this to an amplified degree the first time I rode a motorcycle. I now commute almost everyday on a motorcycle and my commute has transformed for a soul sucking hell to one of my favorite parts of the day.
> It is said that driving is one of the most dangerous activities we do
In the USA cars kill a small city every single year: https://upload.wikimedia.org/wikipedia/commons/d/d1/Motor_ve... This is roughly on par with gun deaths (which are ~2/3 suicides). Cars, not designed to kill people, kill roughly as many people per year in the USA as things designed to kill people.
And if you want that feeling of freedom, ride a bicycle. Not as "powerful", but it surpasses the "extension of body" feeling in every way possible.
That said, granted, the very layout of how people live in the usa is much different than in other countries. You guys have, compared to most of europe, vastly inflated distances between everything outside of city cores, making cars an evil necessity for society to keep functioning.
> You may be blacklisted from traveling in any autonomous by corporations, governments, police etc.
Well, I mean, you can already be blacklisted from travelling in regular cars: your driver’s license can be revoked.
Or are you referring more to capability than legality? Because even in a world with entirely autonomous cars, I would find it hard to believe that all such cars would be “non-rooted”, for the same reason we don’t live in a world where 100% of computers run iOS. Large companies (e.g. logistics companies) need total control and predictability over their vehicle fleet, just like large enterprises need total control and predictability over their servers. So there would likely always be a category of “dumb autonomous” vehicles like all-terrain pick-up trucks, sold to farmers and loggers and quarry workers et al, that you could buy; just like you can buy a regular PC today. You, as its “OEM” owner, would have complete control over what cloud traffic-control systems it submits to, if any. It might be illegal to configure it in certain ways and then drive it on the road, but that’s not the point here.
Car companies will fight this tooth and nail. For an example, see John Deere who use patent and cyber crime law to prevent their product owners from fixing/modifying their equipment.
Or, look at Tesla and how they have fought releasing service manuals or allowing owners to make any kind of modifications/repairs to their cars. Don't be surprised if you have someone coming into this thread to also tell us how these are "good things".
And you can forget about ever visiting your favorite fishing hole again. You know the one, where you have to drive a mile or two down that tricky dirt road that's not on the GPS. Not allowed.
Cars are such an emotive topic because they are a powerful democratizing force. Any ordinary citizen can have at their disposal a machine that can transport them, their family/friends and their stuff, anywhere they want to go at any time, can be packed in advance and unpacked at leisure when you get back, can act as a base or a shelter at the destination, can extend a bubble of personal space on long or stressful journeys that no other means of transport can, and so on and so on. The self-driving car utopia breaks all of this. You can experience it right now, we have voice-activated self-driving on-demand cars right now, running on artificial-artificial-intelligence: they're called taxis. Taxis have not supplanted personal vehicle ownership by a long shot, therefore neither will self-driving cars, unless personal vehicles are banned by government edict - and a government would love to determine where and when you are allowed to go. Will it be worth it?
Taxis have not supplanted personal vehicle ownership by a long shot, therefore neither will self-driving cars, unless personal vehicles are banned by government edict
Yeah, that's why washing machines have never become popular; after all, we could already hire maids to wash clothes for us, yet most people washed their own clothes.
If that were true in this market noone would ever buy a Mercedes when a Nissan would be a fraction of the price. Turns out the real world isn't so simple as you think.
The equivalent of the ultimate promise of the self-driving car that we have right now isn’t a taxi; it’s a limousine driven by someone you have permanently on your personal retainer, like a landed noble’s man-at-arms.
Such a person might favour the law over you, but they won’t like doing it, because they try to hold your beliefs as their own.
Or, for a simpler comparison: a horse. Your horse, from your personal stable.
Another thing to consider is, if you're just calling a magic box on wheels to take you from A to B then disappear again, why would you can who it was made by, what style it was, and so on?I wonder if the manufacturers have fully thought this through from their own perspective... I mean I can't imagine ticking a box for "... and it must be a Volvo" when I summon one, I simply wouldn't care!
I'd care. for a long trip > 15
minutes I'd want a quiet and comfortable car. Quiet so I can relax and/or listen to audio without having to crank up the volume over the noise of the car. comfortable might include seats big enough I can put my notebook on my lap and work. comfortable might also mean a luxury style suspension so the ride is not hard.
Electric autonomous vehicles will already be quiet without much effort compared to today's predominant ICE cars. And comfort, well... the absence of a steering wheel, the center console and all that will free up space for some nicer seats.
And today a lot of cars waste a lot of space on cargo space in situations where you're not carrying any cargo. With hailable fleets they could have a mix of vehicles with and without trunks, the trunkless ones offering more interior space for the passengers.
I don't see those general feature changes to be specific to manufacturer brands, they are more a feature set optimized for sale towards ride hailing services because they might not work as well for individually-owned cars.
Hackers or bugs in software can already kill you - a modern car is a computer with an engine attached. You think that pedal is any more advanced than a scroll mouse?
Congress already passes laws controlling when, where and how you may travel and you may be effectively blacklisted today by insurance companies, courts or police officers for a range of reasons. Do you honestly think that you are entitled to roads wherever you want and your license can never be revoked and insurers can't charge what they want?
Half of the total lifetime emissions of a car are in the driving of it. Autonomous vehicles open up the opportunity for an Uber-esque experience where you're less likely to buy a car and just rent it for when you need it: most people actively need a car for less than 10% of the total time that they own one for. That means significant environmental gains will come from lower demand for car manufacturing. How is that a problem?
Sprawl "getting worse" is another way of saying housing becomes more affordable as it does not need to be so concentrated. City centres will become more comfortable, communities will become more connected.
And finally, if there is one area that will cause a boon for investment by autonomous cars, it's mass transit: driverless buses, trams and trains, will all allow for more money to be spent on more vehicles and the net price of using these services will drop as the cost of employing, training and managing driving staff is reduced.
We can go around and around on this: it's happening, come to peace with it, and try and make sure we maximise the positives and reduce the negatives.
Note though, I did not counter the advertising thing. Tesla's T&Cs on their cars already suggest the owner does not really own the car. I think that's a problem. RMS in one of his more popular talks says he doesn't want software inside his microwave to Free because he would never want to modify it: I think a good argument can be made that if I buy a car that drives itself, I should be able to see and modify the source code.
Agreed. There’s an article of a remote hijack of I think it was a Jeep using Ulink or whatever it’s remote interface was. Scarey with its limited capabilities.
To be fair, I won’t even blame the auto industry for having vulnerabilities in their software. At this point I’ve come to accept that any bit of technology has flaws and can be exploited. No matter how great their dev team is. Comical as it was and unrealistic for the time period, the latest fast and furious movie showed a potential scenario of a fully autonomous traffic.
It’s going to be the same reason there’s no flying cars. The tech has been around for easily 20 years. But if you watch enough Air Disasters, autopilot and other computer controlled airplane functions still fail catastrophically. To this day. The average person is not capable of flying an airplane. Shit, the average person still isn’t great at driving a car. Sure that brings to argument that then all cars should be self driving. But the added risks, I don’t think so. The current sample data of self driving cars is still small and there aren’t active hackers looking to truly exploit yet. It’s like when people said that Macs were immune from viruses back in the day. No, people made viruses for windows because they can cause more wide spread damage since they were more widespread. Few had macs thus, why bother?
But it doesn’t matter. Self driving cars will happen and something bad will happen in that time. There’s too much money to be made. This isn’t being done to save the world. It’s a control scheme as you pointed out. It’s funny how the older you get the more you realize that when someone starts to spout they’re trying to saving the world, they’re covering up some type of scheme to make money and make the world a shittier place.
> It’s funny how the older you get the more you realize that when someone starts to spout they’re trying to saving the world, they’re covering up some type of scheme to make money and make the world a shittier place.
I agree with all the points the OP made for self-driving car downfalls, but I still think that the people working on the tech have the best intentions.
It's just that we as humans do a bad job of having foresight and the critical analysis skills necessary to truly assess the whole of a technology, and not just the good parts. Maybe that reflects well on humans as it hints at an optimistic nature?
I see the potential benefits of self-driving cars, but am worried that the medium-term to long-term negative points will outweigh the positive. There should really be a council of techies, representatives, legal experts, and most importantly average citizens that sit down and heavily critique this technology to address all the negative points before it becomes widespread.
I think the most interesting part is how we humans reason when it comes to risk. A hacker causing accidents is so scary that autonomous cars sound like a bad idea, yet the risk of dying to a hacker while in an autonomous car would almost certainly be lower than dying to a drunk or distracted driver right now.
37 000 people in the US died in vehicle accidents last year and around 2 million were injured. Obviously not all accidents would be preventable by autonomous cars, but I wonder how many of those people would be alive or uninjured if the cars could have been autonomous.
Yea, suicide rates are just as bad too. Same with medical malpractice and just getting an antibiotic resistant infection. Bad things will always happen. An issue here is control. I have an amount of control on the road as an active driver. Truly once self driving cars become a thing, no one will be a passive driver realistically. Everyone will have their nose in their phone. I grew up in Florida and learned to drive in Florida. The rest of the USA has rookie numbers when it comes to deaths on the road. Defensive driving is why I’ve dodged 2 drunk drivers. I won’t give up control of a vehicle to a computer. The reason why flying commercial flights is scary, lack of control even though it’s statistically safer. But it’s also statistically a complete shit storm when something does go wrong. Engine failure in a car and a plane have different stats on result for a reason. Point being, I doubt annual deaths on the road will differ all that much. We will all die in new ways because of self driving cars.
That and folks who drive drunk are actively not making a smart decision already. “I don’t need the car to drive me home. I’m fine” that’ll be the new last words. Anyone who goes they’re too drunk to drive, don’t already, without having a self driving car.
Another interesting point that I don't think is brought up at all is how self-driving tech will impact children. The legal driving age requirements will likely become obsolete.
Can school children take a self-driving car to school every day? Will this increase traffic significantly? How long until a hacker takes control of a car and abducts the child inside? How safe do parents really feel letting their kids be in this machine on their own? etc.
While making it sound like they care about your, productivity or the world in general what matters to Volvo is their profits and the preceding brainwash to get the right traction and acceptance.
[Digress: Volvo's pitch to legitimize joining forces with Uber to fully automate driving.]
Do you expect Volvo to make more profits in an era where fewer people own cars?
I don't think car manufacturers are rubbing their hands at the thought of becoming suppliers to be squeezed by Uber et all, rather than sellers of aspirational items to billions of individuals. I see them as resigning themselves to that future.
This is something I often thought about as a child. I didn't imagine self driving cars as they are being developed today. Rather, I thought about what it would be like if all cars were controlled by a common system. I would imagine cars all taking off together in unison from a stop at traffic lights, cars being merged together from two separate lanes without bringing everything to a crawl.
The beauty of a fully automated and coordinated system is that vehicles would only stop at the end of a journey. Intersections would be a graceful interleaving of vehicles in all directions. This would conserve a lot of energy that is currently wasted on braking and acceleration, and the passengers have less frustration of being stopped when it is their strong desire to be _going_ somewhere.
That video is just full of false assumptions. For example at 0:25 he attributes the delays between cars starting at traffic lights to bad reaction times of humans.
No, you start with a short delay because otherwise there is no way to reestablish the safety distance to the car in front of you. Which you still want to have with automated driving because self-driving cars won't solve sudden mechanical failures.
Replace pedestrians with self-walking autonomous robots! This way they can sync with the cars and cross the street without a single car having to stop or them getting hit.
I find it really disturbing that some people think that non-walkable cities full of speeding robot cars without any regard for pedestrians sound like a bright future.
More pedestrian overpasses? Or maybe the same amount, with all the cars flowing down just a few trunk roads, because nobody is trying to “avoid traffic” by taking a side road.
Give pedestrians a route over or under the roads. This is done in many places throughout the world already, and it works well.
But even if you allow crosswalks and stoplights, cars can use buffer space in traffic (assuming the road is not saturated) to minimize the stops/starts.
Totally agree. This is not something I would have considered as a child, but it seems clear that some decentralization is needed. I think the proper solution is definitely some kind of p2p protocol with cars exchanging information.
Yeah, there would be advantages in the cars exchanging data, but you could have that while still leaving the final decisions up to the individual car (rather than being dictated by a centralized controller).
That's more or less what ants and bees do -- they exchange information with each other, but each unit is still quasi-independent.
It doesn't have to be centralized. If you think of the cars as data packets and the roads as cat-6 cables, then you can see how redundancy and alternate routing can be built in.
While I take your point, many of the system behaviours of IP networking rely on inferences from failure. Information gathering needs to be a bit more proactive and connected without it. Eg determining a jam shouldn't be a matter of noticing that cars sent through a particular area are getting stuck ("dropped") but something that can estimate viable throughput before triggering phase changes like jams.
> I would imagine cars all taking off together in unison from a stop at traffic lights
Why do you even need traffic lights anymore? Autonomous cars can brake and stop for people who need to go through, and synchronise with one another to organise crossing.
Even if all cars were fully connected and automated, I don't see traffic lights going away outside specific types of areas (rural areas, etc.). In cities, pedestrian infrastructure is pretty much built around the presence of intersections and pedestrian walkways. To keep pedestrians moving relatively quickly, you need those intersections. And while there are options (elevated pedestrian walkways, etc.), those have their own problems (size, expense, ADA compliance, etc.). If you enforce delays for people to cross, which would be the cheapest option, you've just got a stop light by another name. Make it too annoying for people, and they'll run across when they think there's room. With as cautious as autonomous cars have to be with pedestrians, you'd probably wind up with more traffic delays than what we have now.
We already have traffic controller systems, but I'd imagine that vehicle communication would allow them to be even more sophisticated.
> If you enforce delays for people to cross, which would be the cheapest option, you've just got a stop light by another name.
If you go to Germany, or Belgium, cars generally stop whenever somebody gets close to the border between the sidewalk and the road, at which point the pedestrian just goes across (then wait 5 minutes and crosses back because they weren't actually trying to cross the road, they were just a foreigner not expecting this). And these are human-driven cars.
This can just be universal policy for self-driving cars. It will necessarily be anyway since they will have to deal with people just running across without warning.
> And while there are options (elevated pedestrian walkways, etc.)
That would be very dumb, one of the benefits of self-driving cars is that you can significantly scale back the car-centric infrastructure and especially affordances towards drivers, there is no risk of drivers being bothered or annoyed or taking rash decisions, the cars just follow policy which is whatever you want to be, every merge is now a proper zipper at the very end of the merge lane & all that.
And wherever that's an option you can just bury the road entirely and keep pedestrians in the sun up top as they did for the centre of the planned city of Louvain la Neuve in Belgium:
> one of the main points of the urban design of Louvain-la-Neuve was to make it people rather than automobile centred. As a consequence, the city center is built on a gigantic concrete slab, with all motorized traffic travelling underground. This allows most of the ground level of the city center to be car free.
(when they need new accommodations, they expand the slab and build the buildings at the same time)
> I'd imagine that vehicle communication would allow them to be even more sophisticated.
If all cars are autonomous it is more efficient to have two cards approaching at the same time negotiate a speed at which they can safely pass through the intersection without stopping or colliding. Stopping would be a thing reserved for when you get to your destination or a pedestrian gets in the way. It's more efficient that way. The energy needed to accelerate is less than the energy needed to maintain speed.
Which of course bring sup the idea of hacking the negotiation so that you always have the favorable outcome. The same way some people try to hack the sensors that turn lights green for emergency vehicles.
What's more efficient? What's more resilient? An army all walking at the same pace taking their cues on orders or a crowd of autonomous individual, all adjusting pace etc... Based on their local condition (terrain, speed of who's in front...).
It probably depends on what you mean by a centralized system. Right now, autonomous vehicles are being developed with a focus on independent operation. Which makes sense, since early examples will find themselves having to integrate into roads lacking car-to-x (or similar) infrastructure[0] shared with vehicles that lack the ability to share anything. There'll be a transition, but older cars will pretty much always be around. Though I wonder if we'll see retrofit kits making use of OBD-II ports for older cars to give them at least some communications abilities at some point, or even some sort of GPS "box" for even older cars lacking OBD ports altogether.
Even when car-to-x and vehicle-to-vehicle communications are standardized with widespread use (and they'll both be usable even with non-autonomous vehicles), I doubt we'll see the sort of centralization with hyper-optimized traffic patterning in the animation linked by philrw. Cars will still be able to share information, and make use of it to change their behavior for minimal disruption of traffic flows. They'll talk to infrastructure, and probably even make use of relays well ahead of key infrastructure elements (ramps, bridges, construction zones, etc.). But I wouldn't consider it "centralized" in that situation.
For example, a car is entering an on-ramp. Local infrastructure could communicate that fact, as well as other information about the car to traffic approaching the on-ramp. Is it a heavily-loaded truck, with slower acceleration? Or is the vehicle 'dumb' (maybe it's an older car with no communication abilities)? Approaching cars would leave a greater gap ahead of them, switch lanes, etc. Even without giving control over to a central traffic system, we'd see significant improvements in traffic flow.
The other challenge for highly-centralized control is the sheer amount of data that's collected by autonomous cars. They apparently generate upwards of 25Gb/data an hour[1] from their various sensors. Figuring out what part of that data needs to be kept, normalized, and then sent to a centralized system is going to be tough. And it'll be hell on the wireless spectrum. Of course, that's assuming V2V spectrum isn't carved up and used for internet access.[2] In which case, things will get...trickier.
Last week, I got an e-mail from Subaru telling me my windshield washer fluid was low (it was). The e-mail provided information on how to refill it, and, a link to make an appointment for service if I felt something else needed attention.
If, instead, my brake pads were worn and the car were fully autonomous, I could have scheduled it for service while I am at work and it could have driven itself in, got new brakes and driven itself back to my office. No loaner car, no time off from work, no hassle waiting in line for a service adviser, ....
File under the 'productivity' column, and I expect cars will be better maintained than now, since we procrastinators will be able to simply click a link and forget about it.
They miss the point of not owning a car (partially mentioned in both "economy" and "environment"): the car is usually the #1 or #2 (after house, if you own one) household expense; it's a depreciating asset with a terrible utilization rate (typically less than 5%). Bought by individuals, who typically have much worse credit than corporations.
If you pay by the drive for a car service operated by a large company that borrows at the prime rate, your cost per mile is liable to be much lower. Given the huge costs of car ownership, it's hard to believe any residual emotional bond of owning a car could trump the economic benefits.
And as for electric: knowing aggregate demand means you never have "range anxiety", and the charging infrastructure can be much less obtrusive.
Social consequences? Well, when you take that Google-operated car to Manresa it will likely detour past MacDonalds and offer you a 2-for-one discount on Big Macs if you go there instead. And there will be no ad blockers...
Oh, and you want to take that car to where there will be an alt-right rally? Oops, no cars allowed to drive there (either by government edict or because the car owners don't want to risk their cars being destroyed).
The self-driving fleets can't arrive soon enough for me, but I am aware of some of the malign risks as well...
No, they're assuming you still own your car - it's just that your car would be autonomous. Only the last section ("More flexible ownership") acknowledges the possibility that you might not own the autonomous car.
Self-driving cars and subscription-ownership cars are two circles on a Venn diagram that don't have to exactly overlap. I'm a great example - I want a self-driving car of my own, but I don't want to share it with anybody else. (I've played the Zipcar game in Chicago and I'm over that.)
> Self-driving cars and subscription-ownership cars are two circles on a Venn diagram that don't have to exactly overlap. I'm a great example - I want a self-driving car of my own, but I don't want to share it with anybody else. (I've played the Zipcar game in Chicago and I'm over that.)
That's transitional and based on the hysteresis of looking at the current system. I remember when PCs started to be introduced; they replaced dedicated word processing systems (and, via Visicalc, some central data processing) but it was pretty well accepted that they would never make it into the corner offices because "executives don't type". In 1982 I worked at a lab where the director had his secretary retype his email onto paper; he would read it, mark it, and give it back for filing, and would dictate responses (she was overjoyed when I showed her how to print the mail out).
I have a longer reply to a different reply to my message which emphasizes the primacy of the economic drivers (and mentions some corner cases where it might be worth buying one)
> I want a self-driving car of my own, but I don't want to share it with anybody else.
That's assuming that you will even be able to buy a self-driving car as an individual. (Or at all.) I find it highly plausible that car companies will expand to fill the entire vertical of car design, manufacturing and as-a-service operation.
The service-oriented model was not really viable until now because the model of individuals buying cars was pretty much the only accepted option. But car companies have already started the vertical integration (many of them all have their own carsharing subsidiaries).
You miss the point of owning a car: it's space that's yours. You only use your bed 1/3 of the time, and your toilet for less than 5%. Why not hot-sheet it and use a shared toilet down the street? Because you're willing to pay more for personal space, where you can leave your stuff and maintain your preferred level of hygiene.
I agree that certain cases will remain, such as for, say, a plumber, though in theory the ubiquity of autonomous vehicles means that a plumbing supply company could provide JIT service.
But although I don't blindly believe in homo economicus, I wasn't kidding when I said that the car is an (or for many the most) enormous expense for most people, and the size of that expense will aggressively drive the adoption rate. The ability to have a box of Kleenex or your gym bag will become a nice to have when the price of that convenience doubles or triples your transportation costs. The difference in price might allow you to afford gym membership in the first place.
It's also a generational thing. My 19 yo kid and his friends are pretty uninterested in cars. A bunch of them don't even have licenses (which is common in some big cities like NYC but uncommon out here in the sub urbs). They care about phones though.
In the case of the toilet and bed, those are typically bundled, but even then some people do make the economic choice to have only a shower rather than a bathtub just like they make such a choice for a normal car vs paying extra for automatic transmission.
(there are also transitional and special cases; it's hard to hail a car where there is no phone service!)
Sharing of bathrooms, washing machines, etc is actually quite common among people who want to live in the city centres (expensive) while having low income - e.g. students. Reusing the space used for sleeping is much less common, and comes with more trade-offs, but it's not unheard of either (high beds and similar come to mind [1] [2]). Not sure if this is different in the US...
I stumbled upon this when searching for "what if all cars were self driving". I really feel that a great deal of the problems with making self driving cars more prevalent are related to the human-driven cars that are on the road along side these autonomous vehicles. Having only autonomous vehicles on the roads seems to have some powerful advantages, including lighter vehicles. I would love to see a region or city go totally driverless to empirically test whether autonomous vehicles really can provide the added benefits mentioned in the article.
Much lighter. All the people driving alone could be in cars one-quarter the size if they didn't have to worry about being crushed by Canyoneros. Aside from the cost savings, half-width cars can offer a traffic volume / speed improvement by sharing a lane. (Ultimately though lane markings can go away along with traffic lights and signs.)
One often ignored thing in this conversation is we often use the ability to carry more. I keep my emergency kit in the back (I'm part of a CERT team) as well as spare chargers and cables and stuff. I keep telling myself I should stock a spare change of clothes too. Obviously sometimes I drive with friends or family, load up my car with groceries, occasionally move furniture or computers.
The car is a relatively reasonable standard for covering an individual's regular use cases for transportation.
I wonder if some kind of detachable "trailers" might be the answer here. For normal driving, you'd have a single-passenger pod (with maybe enough cargo space for a bag of groceries or whatever), but you could hook on extra passenger or cargo pods as needed.
There is a small portent here of something I think we should be cautious of:
> The space that is currently used for car parks could be used for something else.
We should be careful that this space does not uniformly end up being used for yet more skyscrapers and office buildings, but open space and parks instead. I feel like this is a major decision moment for ending up with beautiful cities or dark cyberpunk monstrosities.
Even if we still need car parks, cars parking themselves can park really close together (no need to open doors), and block each other in (they'll move autonomously to let another car out). If you go multi-story, you could have some floors with a really low ceiling where regular (non-SUV) cars could park at a lower price. I'd bet you'd need a lot less space for the same number of cars.
2. I don't want to have to travel 30+ minutes to see more than a single tree in one place. Frequent access to nature (or a sufficiently advanced facsimile) is life quality.
3. If a city is sufficiently large, traveling to its outside may be prohibitively expensive for poor people who are thus barred from experiencing nature. (This concern affects developing countries and sci-fi dystopias more than current developed countries.)
It’s not like #3 hasn’t happened before. Central Park is so far north in Manhattan because it was originally meant to be inaccessible to those without the means to afford a carriage.
Everyone is waiting for autonomous cars to be near-perfectly safe before accepting them.
The thing is, let's say that autonomous cars are only 0.25% safer than the average human driver. That means switching to them will reduce accidents by 0.25%. That's a significant amount considering how many traffic accidents there are.
Considering that traffic accidents result in the actual death of human beings, it is morally imperative that we require as much autonomous driving and reduce as much human driving as possible. Even the far-from-perfect autonomous cars we have now are safer than continuing to let humans drive.
Something you're neglecting to account for is that there's more to driving than simply moral hazards and safety. There's also the matter of liability and financial risk.
While it may be arguable that it's morally preferable and safer to adopt self-driving cars if they can be shown to reduce the incidence of accidents by any amount compared to human drivers, it isn't desirable for the corporate entities producing vehicles capable of fully autonomous operation to make them a mass market product until it is remarkably safer to ride in an autonomous vehicle compared to a human-operated vehicle.
One of the biggest boons for automakers in the status quo is that the liabilities associated with operating their products is mostly distributed to the users. From a financial risk perspective this is a massively preferable operating model (and not dissimilar from what you see with 'disruptors' in the consumer space like Uber, Air BnB, etc). When we remove the primary cause of faults while driving from the picture (the human agent operating the vehicle), the distribution of risk that has generally protected automakers in the past disappears. When an autonomous vehicle crashes, the automaker now must fend off claims about vehicle features (sensors, software, etc.) that will in some way be directly responsible for causing traffic accidents.
Cars dropping you off autonomously reduces the demand for parking, however it will also double traffic. This leads to worse traffic flow and more congestion, contrary to the claimed positive benefit.
A lot of things on the list are also almost completely orthogonal to autonomous driving, like electric cars and flexible ownership models
If autonomous cars are operated by centralized services (Uber/Lyft/...), then they can offer cheaper rides if you are willing to share a car with different people. Especially when congestion is at its worst, chances of matching people with the same destinations are also the best - the daily morning jam on highways is mostly caused by people driving alone in their cars. Why not share 90% of the way with X other people, if it lasts only 2 minutes more for changing from/into a small first/last-mile vehicle. And why not deploy lots of small buses which carry 10+ people through the congested areas? This could even be encouraged by lawmakers; but I think this group rides would be so much cheaper that many people would use them whenever possible. Just like you don’t use your car if you live next to a subway and your destination is also right next to a station.
My Volvo (XC90) is the same way. It's not as advanced as the Tesla but it has a traffic jam assist feature. If the traffic goes from almost full stop to moving along I often have to override it and accelerate manually to avoid someone cutting me off or honking at me.
Ah yes! Just the other week I turned on Pilot Assist on our XC90 during a long road trip and people playing on their phones in left lane stabbing their brakes make for a very rough ride as the car keeps braking and then re-accelerating to try to keep a constant distance.
Not to mention all of the people cutting you off because you've left enough space for them to jam their car in front of you. I had to intervene a number of times and reposition myself in traffic to get away from "non-smooth" drivers,
My issue with mass transit is destination granularity. I can imagine having individual cars that can "clump" together to form a mass transit system, and then break off as and when required to arrive at specific destinations.
Isn't Volvo planning on using Chinese self-driving software? As with the vast majority of Chinese software, I wonder if it will also contain a backdoor for the government.
If all cars were autonomous you could change how roads are built. They could become more like sewer pipes, sunk at least partly into the ground. Car designs would change and become more cylindrical and without windows. The car interior would be covered with displays. Travel videos will become a thing.
You can see why you might want to start practicing with boring machines and hyperloops. These things aren't just for Mars.
Much of Houston's freeways are built far below grade and surface roads slightly below grade. It allows the roadway to function as drainage during heavy rains.
I don't think that this is obviously true, especially in cities. When the cost of taking an autonomous car from door to door is comparable to the cost of public transport, there would likely be more traffic on the road, not less. We've already started to see this to some extend with Uber in big cities
If all cars were autonomous, society, the economy and the environment would all suffer. Here is how…
* You will be forced to watch non-stop ads in these automotive cars, similar to how some taxi services work today. You can turn these off for a fee of course.
* Hackers or bugs in softawe may easily cause global traffic jams
* Congress will pass laws controlling when, where and how you may travel
* You may be blacklisted from traveling in any autonomous by corporations, governments, police etc.
* The aggregate amount of automotive travel will increase exponentially, causing more carbon emissions and even worse traffic since more people will have access to travel
* Sprawl will get worse, as people move further away from urban areas since they can do other activities in cars.
* Investment in mass transit will diminish, since funds will move towards automatic cars.
It is said that driving is one of the most dangerous activities we do, but I think it's no coincidence that it's also one of the most liberating. It's a very pleasurable feeling to be able to get into a car and drive, controlling it as if it were an extension of your body.
The phrase "If all cars were autonomous" has a similar feeling to others like "if there was no crime" --- sounds good on the surface, but also has deeply authoritarian and dystopian implications.
Edit: there's also this, which is basically like Stallman's "Right to Read" but for self-driving cars: http://this.deakin.edu.au/lifestyle/car-wars
I experienced this to an amplified degree the first time I rode a motorcycle. I now commute almost everyday on a motorcycle and my commute has transformed for a soul sucking hell to one of my favorite parts of the day.
[1]https://youtu.be/Vrxyr1CjiSM
[2]https://youtu.be/7bXJ_obaiYQ
Not bad for a 2002 film adaptation of a 1956 short story.
On an open highway, sure. In rush hour gridlock, it's awful.
In the USA cars kill a small city every single year: https://upload.wikimedia.org/wikipedia/commons/d/d1/Motor_ve... This is roughly on par with gun deaths (which are ~2/3 suicides). Cars, not designed to kill people, kill roughly as many people per year in the USA as things designed to kill people.
And if you want that feeling of freedom, ride a bicycle. Not as "powerful", but it surpasses the "extension of body" feeling in every way possible.
That said, granted, the very layout of how people live in the usa is much different than in other countries. You guys have, compared to most of europe, vastly inflated distances between everything outside of city cores, making cars an evil necessity for society to keep functioning.
Well, I mean, you can already be blacklisted from travelling in regular cars: your driver’s license can be revoked.
Or are you referring more to capability than legality? Because even in a world with entirely autonomous cars, I would find it hard to believe that all such cars would be “non-rooted”, for the same reason we don’t live in a world where 100% of computers run iOS. Large companies (e.g. logistics companies) need total control and predictability over their vehicle fleet, just like large enterprises need total control and predictability over their servers. So there would likely always be a category of “dumb autonomous” vehicles like all-terrain pick-up trucks, sold to farmers and loggers and quarry workers et al, that you could buy; just like you can buy a regular PC today. You, as its “OEM” owner, would have complete control over what cloud traffic-control systems it submits to, if any. It might be illegal to configure it in certain ways and then drive it on the road, but that’s not the point here.
That's a nice dream!
Car companies will fight this tooth and nail. For an example, see John Deere who use patent and cyber crime law to prevent their product owners from fixing/modifying their equipment.
Or, look at Tesla and how they have fought releasing service manuals or allowing owners to make any kind of modifications/repairs to their cars. Don't be surprised if you have someone coming into this thread to also tell us how these are "good things".
And you can forget about ever visiting your favorite fishing hole again. You know the one, where you have to drive a mile or two down that tricky dirt road that's not on the GPS. Not allowed.
Yeah, that's why washing machines have never become popular; after all, we could already hire maids to wash clothes for us, yet most people washed their own clothes.
Such a person might favour the law over you, but they won’t like doing it, because they try to hold your beliefs as their own.
Or, for a simpler comparison: a horse. Your horse, from your personal stable.
And today a lot of cars waste a lot of space on cargo space in situations where you're not carrying any cargo. With hailable fleets they could have a mix of vehicles with and without trunks, the trunkless ones offering more interior space for the passengers.
I don't see those general feature changes to be specific to manufacturer brands, they are more a feature set optimized for sale towards ride hailing services because they might not work as well for individually-owned cars.
Congress already passes laws controlling when, where and how you may travel and you may be effectively blacklisted today by insurance companies, courts or police officers for a range of reasons. Do you honestly think that you are entitled to roads wherever you want and your license can never be revoked and insurers can't charge what they want?
Half of the total lifetime emissions of a car are in the driving of it. Autonomous vehicles open up the opportunity for an Uber-esque experience where you're less likely to buy a car and just rent it for when you need it: most people actively need a car for less than 10% of the total time that they own one for. That means significant environmental gains will come from lower demand for car manufacturing. How is that a problem?
Sprawl "getting worse" is another way of saying housing becomes more affordable as it does not need to be so concentrated. City centres will become more comfortable, communities will become more connected.
And finally, if there is one area that will cause a boon for investment by autonomous cars, it's mass transit: driverless buses, trams and trains, will all allow for more money to be spent on more vehicles and the net price of using these services will drop as the cost of employing, training and managing driving staff is reduced.
We can go around and around on this: it's happening, come to peace with it, and try and make sure we maximise the positives and reduce the negatives.
Note though, I did not counter the advertising thing. Tesla's T&Cs on their cars already suggest the owner does not really own the car. I think that's a problem. RMS in one of his more popular talks says he doesn't want software inside his microwave to Free because he would never want to modify it: I think a good argument can be made that if I buy a car that drives itself, I should be able to see and modify the source code.
To be fair, I won’t even blame the auto industry for having vulnerabilities in their software. At this point I’ve come to accept that any bit of technology has flaws and can be exploited. No matter how great their dev team is. Comical as it was and unrealistic for the time period, the latest fast and furious movie showed a potential scenario of a fully autonomous traffic. It’s going to be the same reason there’s no flying cars. The tech has been around for easily 20 years. But if you watch enough Air Disasters, autopilot and other computer controlled airplane functions still fail catastrophically. To this day. The average person is not capable of flying an airplane. Shit, the average person still isn’t great at driving a car. Sure that brings to argument that then all cars should be self driving. But the added risks, I don’t think so. The current sample data of self driving cars is still small and there aren’t active hackers looking to truly exploit yet. It’s like when people said that Macs were immune from viruses back in the day. No, people made viruses for windows because they can cause more wide spread damage since they were more widespread. Few had macs thus, why bother? But it doesn’t matter. Self driving cars will happen and something bad will happen in that time. There’s too much money to be made. This isn’t being done to save the world. It’s a control scheme as you pointed out. It’s funny how the older you get the more you realize that when someone starts to spout they’re trying to saving the world, they’re covering up some type of scheme to make money and make the world a shittier place.
I agree with all the points the OP made for self-driving car downfalls, but I still think that the people working on the tech have the best intentions.
It's just that we as humans do a bad job of having foresight and the critical analysis skills necessary to truly assess the whole of a technology, and not just the good parts. Maybe that reflects well on humans as it hints at an optimistic nature?
I see the potential benefits of self-driving cars, but am worried that the medium-term to long-term negative points will outweigh the positive. There should really be a council of techies, representatives, legal experts, and most importantly average citizens that sit down and heavily critique this technology to address all the negative points before it becomes widespread.
37 000 people in the US died in vehicle accidents last year and around 2 million were injured. Obviously not all accidents would be preventable by autonomous cars, but I wonder how many of those people would be alive or uninjured if the cars could have been autonomous.
Can school children take a self-driving car to school every day? Will this increase traffic significantly? How long until a hacker takes control of a car and abducts the child inside? How safe do parents really feel letting their kids be in this machine on their own? etc.
Or the other way around, you can move "transit" underground and contract cities as there's no need for non-pedestrian surface streets.
[Digress: Volvo's pitch to legitimize joining forces with Uber to fully automate driving.]
I don't think car manufacturers are rubbing their hands at the thought of becoming suppliers to be squeezed by Uber et all, rather than sellers of aspirational items to billions of individuals. I see them as resigning themselves to that future.
It has nothing to do with the number of cars being sold.
Maybe we'll get there one day.
No, you start with a short delay because otherwise there is no way to reestablish the safety distance to the car in front of you. Which you still want to have with automated driving because self-driving cars won't solve sudden mechanical failures.
I find it really disturbing that some people think that non-walkable cities full of speeding robot cars without any regard for pedestrians sound like a bright future.
But even if you allow crosswalks and stoplights, cars can use buffer space in traffic (assuming the road is not saturated) to minimize the stops/starts.
It really doesn't. Here in Germany most underpasses built in the car-friendly 60s and 70s are being abolished because pedestrians don't like them.
The safety message will be detected by all nearby autonomous vehicles, giving them a chance to slow down or stop while you cross.
That's the theory, anyway.
That's more or less what ants and bees do -- they exchange information with each other, but each unit is still quasi-independent.
While I take your point, many of the system behaviours of IP networking rely on inferences from failure. Information gathering needs to be a bit more proactive and connected without it. Eg determining a jam shouldn't be a matter of noticing that cars sent through a particular area are getting stuck ("dropped") but something that can estimate viable throughput before triggering phase changes like jams.
Why do you even need traffic lights anymore? Autonomous cars can brake and stop for people who need to go through, and synchronise with one another to organise crossing.
We already have traffic controller systems, but I'd imagine that vehicle communication would allow them to be even more sophisticated.
If you go to Germany, or Belgium, cars generally stop whenever somebody gets close to the border between the sidewalk and the road, at which point the pedestrian just goes across (then wait 5 minutes and crosses back because they weren't actually trying to cross the road, they were just a foreigner not expecting this). And these are human-driven cars.
This can just be universal policy for self-driving cars. It will necessarily be anyway since they will have to deal with people just running across without warning.
> And while there are options (elevated pedestrian walkways, etc.)
That would be very dumb, one of the benefits of self-driving cars is that you can significantly scale back the car-centric infrastructure and especially affordances towards drivers, there is no risk of drivers being bothered or annoyed or taking rash decisions, the cars just follow policy which is whatever you want to be, every merge is now a proper zipper at the very end of the merge lane & all that.
And wherever that's an option you can just bury the road entirely and keep pedestrians in the sun up top as they did for the centre of the planned city of Louvain la Neuve in Belgium:
> one of the main points of the urban design of Louvain-la-Neuve was to make it people rather than automobile centred. As a consequence, the city center is built on a gigantic concrete slab, with all motorized traffic travelling underground. This allows most of the ground level of the city center to be car free.
(when they need new accommodations, they expand the slab and build the buildings at the same time)
> I'd imagine that vehicle communication would allow them to be even more sophisticated.
That's a really sad lack of imagination.
Which of course bring sup the idea of hacking the negotiation so that you always have the favorable outcome. The same way some people try to hack the sensors that turn lights green for emergency vehicles.
Even when car-to-x and vehicle-to-vehicle communications are standardized with widespread use (and they'll both be usable even with non-autonomous vehicles), I doubt we'll see the sort of centralization with hyper-optimized traffic patterning in the animation linked by philrw. Cars will still be able to share information, and make use of it to change their behavior for minimal disruption of traffic flows. They'll talk to infrastructure, and probably even make use of relays well ahead of key infrastructure elements (ramps, bridges, construction zones, etc.). But I wouldn't consider it "centralized" in that situation.
For example, a car is entering an on-ramp. Local infrastructure could communicate that fact, as well as other information about the car to traffic approaching the on-ramp. Is it a heavily-loaded truck, with slower acceleration? Or is the vehicle 'dumb' (maybe it's an older car with no communication abilities)? Approaching cars would leave a greater gap ahead of them, switch lanes, etc. Even without giving control over to a central traffic system, we'd see significant improvements in traffic flow.
The other challenge for highly-centralized control is the sheer amount of data that's collected by autonomous cars. They apparently generate upwards of 25Gb/data an hour[1] from their various sensors. Figuring out what part of that data needs to be kept, normalized, and then sent to a centralized system is going to be tough. And it'll be hell on the wireless spectrum. Of course, that's assuming V2V spectrum isn't carved up and used for internet access.[2] In which case, things will get...trickier.
0. https://www.mercedes-benz.com/en/mercedes-benz/innovation/ca...
1. https://hackernoon.com/automated-cars-and-data-786dfb1e3eb4
2. http://www.latimes.com/business/autos/la-fi-hy-talking-cars-...
If, instead, my brake pads were worn and the car were fully autonomous, I could have scheduled it for service while I am at work and it could have driven itself in, got new brakes and driven itself back to my office. No loaner car, no time off from work, no hassle waiting in line for a service adviser, ....
File under the 'productivity' column, and I expect cars will be better maintained than now, since we procrastinators will be able to simply click a link and forget about it.
If you pay by the drive for a car service operated by a large company that borrows at the prime rate, your cost per mile is liable to be much lower. Given the huge costs of car ownership, it's hard to believe any residual emotional bond of owning a car could trump the economic benefits.
And as for electric: knowing aggregate demand means you never have "range anxiety", and the charging infrastructure can be much less obtrusive.
Social consequences? Well, when you take that Google-operated car to Manresa it will likely detour past MacDonalds and offer you a 2-for-one discount on Big Macs if you go there instead. And there will be no ad blockers...
Oh, and you want to take that car to where there will be an alt-right rally? Oops, no cars allowed to drive there (either by government edict or because the car owners don't want to risk their cars being destroyed).
The self-driving fleets can't arrive soon enough for me, but I am aware of some of the malign risks as well...
No, they're assuming you still own your car - it's just that your car would be autonomous. Only the last section ("More flexible ownership") acknowledges the possibility that you might not own the autonomous car.
Self-driving cars and subscription-ownership cars are two circles on a Venn diagram that don't have to exactly overlap. I'm a great example - I want a self-driving car of my own, but I don't want to share it with anybody else. (I've played the Zipcar game in Chicago and I'm over that.)
That's transitional and based on the hysteresis of looking at the current system. I remember when PCs started to be introduced; they replaced dedicated word processing systems (and, via Visicalc, some central data processing) but it was pretty well accepted that they would never make it into the corner offices because "executives don't type". In 1982 I worked at a lab where the director had his secretary retype his email onto paper; he would read it, mark it, and give it back for filing, and would dictate responses (she was overjoyed when I showed her how to print the mail out).
I have a longer reply to a different reply to my message which emphasizes the primacy of the economic drivers (and mentions some corner cases where it might be worth buying one)
That's assuming that you will even be able to buy a self-driving car as an individual. (Or at all.) I find it highly plausible that car companies will expand to fill the entire vertical of car design, manufacturing and as-a-service operation.
The service-oriented model was not really viable until now because the model of individuals buying cars was pretty much the only accepted option. But car companies have already started the vertical integration (many of them all have their own carsharing subsidiaries).
But although I don't blindly believe in homo economicus, I wasn't kidding when I said that the car is an (or for many the most) enormous expense for most people, and the size of that expense will aggressively drive the adoption rate. The ability to have a box of Kleenex or your gym bag will become a nice to have when the price of that convenience doubles or triples your transportation costs. The difference in price might allow you to afford gym membership in the first place.
It's also a generational thing. My 19 yo kid and his friends are pretty uninterested in cars. A bunch of them don't even have licenses (which is common in some big cities like NYC but uncommon out here in the sub urbs). They care about phones though.
In the case of the toilet and bed, those are typically bundled, but even then some people do make the economic choice to have only a shower rather than a bathtub just like they make such a choice for a normal car vs paying extra for automatic transmission.
(there are also transitional and special cases; it's hard to hail a car where there is no phone service!)
[1] - https://i.imgur.com/IH21TYj.png
[2] - https://i.imgur.com/h0hrXqx.png
Much lighter. All the people driving alone could be in cars one-quarter the size if they didn't have to worry about being crushed by Canyoneros. Aside from the cost savings, half-width cars can offer a traffic volume / speed improvement by sharing a lane. (Ultimately though lane markings can go away along with traffic lights and signs.)
The car is a relatively reasonable standard for covering an individual's regular use cases for transportation.
> The space that is currently used for car parks could be used for something else.
We should be careful that this space does not uniformly end up being used for yet more skyscrapers and office buildings, but open space and parks instead. I feel like this is a major decision moment for ending up with beautiful cities or dark cyberpunk monstrosities.
2. I don't want to have to travel 30+ minutes to see more than a single tree in one place. Frequent access to nature (or a sufficiently advanced facsimile) is life quality.
3. If a city is sufficiently large, traveling to its outside may be prohibitively expensive for poor people who are thus barred from experiencing nature. (This concern affects developing countries and sci-fi dystopias more than current developed countries.)
The thing is, let's say that autonomous cars are only 0.25% safer than the average human driver. That means switching to them will reduce accidents by 0.25%. That's a significant amount considering how many traffic accidents there are.
Considering that traffic accidents result in the actual death of human beings, it is morally imperative that we require as much autonomous driving and reduce as much human driving as possible. Even the far-from-perfect autonomous cars we have now are safer than continuing to let humans drive.
While it may be arguable that it's morally preferable and safer to adopt self-driving cars if they can be shown to reduce the incidence of accidents by any amount compared to human drivers, it isn't desirable for the corporate entities producing vehicles capable of fully autonomous operation to make them a mass market product until it is remarkably safer to ride in an autonomous vehicle compared to a human-operated vehicle.
One of the biggest boons for automakers in the status quo is that the liabilities associated with operating their products is mostly distributed to the users. From a financial risk perspective this is a massively preferable operating model (and not dissimilar from what you see with 'disruptors' in the consumer space like Uber, Air BnB, etc). When we remove the primary cause of faults while driving from the picture (the human agent operating the vehicle), the distribution of risk that has generally protected automakers in the past disappears. When an autonomous vehicle crashes, the automaker now must fend off claims about vehicle features (sensors, software, etc.) that will in some way be directly responsible for causing traffic accidents.
A lot of things on the list are also almost completely orthogonal to autonomous driving, like electric cars and flexible ownership models
Not to mention all of the people cutting you off because you've left enough space for them to jam their car in front of you. I had to intervene a number of times and reposition myself in traffic to get away from "non-smooth" drivers,
https://www.ted.com/talks/wanis_kabbaj_what_a_driverless_wor...
Wouldn't it be far easier for companies to embrace telecommuting than to have a worldwide network of autonomous vehicles?
https://www.theverge.com/2017/6/27/15878900/volvo-nvidia-sel...
You can see why you might want to start practicing with boring machines and hyperloops. These things aren't just for Mars.
The Big Dig in Boston cost what, $20Bn? For less than 4 miles of road tunnel? The maths just doesn't work. What does average road cost per mile?
I don't think that this is obviously true, especially in cities. When the cost of taking an autonomous car from door to door is comparable to the cost of public transport, there would likely be more traffic on the road, not less. We've already started to see this to some extend with Uber in big cities