Three-quarters of drivers don't want to own an autonomous car
bloomberg.com
bloomberg.com
But in the end, people went ahead and developed more and more infrastructure for the internet regardless, and now most people couldn't live without it.
I'd just read Kurzweil's book and I remember saying they'll probably have one by 2010-2011 because smart phones will be the free phones by then. I was right for one simple reason: Most people tend to dismiss new tech right up to when it stops becoming a luxury item. As soon as people find value in something and they can afford it, adoption is practically inevitable.
I hope my self driving car has a nice sleeper option so I can take long road trips and wake up at my destination.
People don't know what they want, so we'll make them want it.
Of course this does not confirm that tablets are falling out of favour - they could simply have reached their saturation point. Does anybody know if saturation sales are expected to look like this normalish (gaussianish) graph?
[0] http://www.statista.com/statistics/269915/global-apple-ipad-...
[1] http://www.statista.com/statistics/272070/global-tablet-ship...
things changed when a computer appeared that people really liked to carry. The same wait is here - for the iPhone of autonomous cars. What right now drives around in Mountain View seems to be more like HP Treo.
This does not need any shiny packaging to be a no brainer. It's a drop dead delicious invention.
That some even entertain the notion that they don't want one goes to show how clueless people are. They are just saying what they think everyone else thinks, without engaging thier brain.
The enjoyment of driving is not about human warmth, it's about control and adrenaline. I don't think the comparison makes sense. I'd be more excited about autonomous cars if I could turn it off in the country (as some cars today allow) for recreation.
I definitely agree with what you're saying—I know I'm probably in the minority for enjoying driving at all—but I frankly can't imagine anyone enjoying reading their credit card over the phone.
For how many of those 46 minutes do you think they are enjoying control and adrenaline? I would guess approximately zero, and once you tell them that they can spend that time instead surfing the internet, or eating breakfast, or videochatting with their family, they will.
[1] http://newsroom.aaa.com/2015/04/new-study-reveals-much-motor...
When autonomous cars are ubiquitous, the price of insuring your thrill ride will be roughly equivalent to the hourly price of an amusement park.
Long before "automonous cars are ubiquitous" something else is going to happen - automatic collision detection and prevention will become ubiquitous, long before all cars are fully automatic. So even if I drive the car myself, it will be very difficult to crash, therefore the insurance will have no reason to increase.
Either way, I am sure fully automatic cars won't become ubiquitous in our lifetimes no matter what manufacturers say, so I'm not too worried.
Any transit system I've ever used was:
Over crowded.
Purposefully designed so no-one could sleep. Also, I really wouldn't want to sleep on an artificially created publicly shared surface.
Frowned upon to talk on your cell phone, or eat a meal.
I definitely couldn't get any work done, even if I tried.
The bus is not a great place to do any of the things mentioned above.
But this is a great illustration of self-driving cars: it's public transit for the upper-middle class. (The middle class can't afford a reliable car for the whole household; the rich have chauffeurs.)
This is exactly what makes autonomous vehicles so great - they eliminate the public transportation problem.
(hi; I live in a city run by "fiscal conservatives")
The US laws are crazy: http://duiwise.com/dui-dwi-answers/can-you-get-a-dui-dwi-on-...
Is Waze for horses patentable?
That was before they turned calling up a company into an audio version of the 90's internet.
Why assume that you won't be able to? Even the most elaborately "uncarlike" self-driving car concepts have a steering wheel hidden somewhere, for the simple reason that there's always the chance of some extreme conditions that a self-driving car can't navigate.
1. For minor uncertainty, the car could ask what to do. For example, the user might confirm that an object blocking the road is harmless and can be run over or bumped out of the way. This is not real-time; the car only asks after it comes to a stop.
2. The above can be done by remote humans.
3. When the car is really confused, you should get power assistance while you stand outside and push. It should be possible to push the car up any San Francisco hill with just one person. Put your hands on the bumper, press hard, and the car responds to the pressure. Unlike traditional car pushing, the steering should respond. You should be able to parallel park, park on grass, force the car up onto the sidewalk a bit, go the wrong direction, and so on.
Because the attractive argument for driverless cars that makes sense (to me) is that they're safer than human drivers. Presumably at some point humans will not be allowed to drive anymore.
This stopped happening because humans are expensive, not because typing numbers into a 2 inch touchscreen keyboard is actually a better experience.
That's what we mean by "calling someone to book a flight" i.e. an agent who will take your request and do the legwork. Not simply talking to someone at a call center and dictating everything you'd otherwise be typing into an online form.
If I can do something myself, in 5 mins, inconveniently, for free.
Or spend 5 mins on a phone call, waiting for a human to do the same thing, for $10.
I would call the latter a poor user experience.
In my own opinion, being a passenger in a self-driving vehicle wouldn't be riskier than being a passenger in a normal vehicle. On the contrary, I think it's safer. But there's the issue of how human drivers might behave or be forced to behave once self-driving cars become widespread.
What if human drivers' become more trusting of other cars' behavior and start paying less attention? Human drivers might become involved in worse crashes between themselves.
What if there's an adverse selection effect? If the public is convinced that self-driving cars are safer for individual passengers, then the most risk-averse drivers might be more likely to join in, leaving less risk-averse (and prone to accident) drivers on the road.
How would human drivers interact with platoons? What if a human driver is stuck in a suddenly decelerating platoon? Would he panic and swerve? Would he crash?
My point is that there might be all kinds of unforeseen consequences. Maybe good, maybe bad. That's reason why I'm not particularly enthusiastic about self-driving cars (or anything else for that matter): we'll only know whether it's good or bad in hindsight.
It will actually be insurance companies that force the change at the consumer level. Charging crazy premiums for irrational holdouts that insist that their love of driving is better than safer transportation will force the hands of many from "no thanks" to "well, if I can save $500/month..."
However, regardless of consumer uptake it's the long-haul trucking companies that stand the most to gain from autonomous vehicles. I don't have a stat handy but robot long-haul would be 25% the price and 4x faster. Something like 20% of highway accidents involve big rigs.
One thing I'd love to see is a safe racing mode where the computer allows people to drive like maniacs, but intervenes before the situation actually becomes dangerous. I suspect at some point automatic drivers will become so precise they will truly be able to scare the shit out of passengers safely.
You can pretty much get that already on closed test tracks, with self-driving cars that do 'speed runs'.
An example (though without passengers): https://www.youtube.com/watch?v=Ol3g7i64RAI
As such, I see no reason why insurance companies would have an agenda against human drivers. Insurance companies will charge rational premiums that make them competitive against other insurance companies. It may very well be that insurance premiums for human drivers go down because AI drivers improve road safety.
> "The probability of having an accident is 50 per cent lower if you have Autopilot on," said Musk, speaking at an energy conference in Oslo, Norway. "Even with our first version, it's almost twice as good as a person."
The people who choose to do something because they are personally motivated to do it tend to become much better than the people who do something because they have to... a phenomenon which is clearly observable in most high-school math classes.
Insurance companies will force the change. Starting, as you pointed out, with trucking.
The trucking industry will be forced to automate. As soon as it is demonstrable that automated trucks are far safer - and faster, as they need no breaks - then trucking companies will be forced to automate or die.
Passenger cars will follow thereafter. People will maintain the right to drive their own cars, but only at prohibitively high insurance rates.
When humans are no longer responsible for piloting the vehicle, it no longer makes sense to sell the vehicle to them, as there is no need to assign the risk to one individual. Instead, rented / leased vehicles will be much cheaper, and timesharing / ridesharing will be much more feasible.
Monthly costs,
Capital $2000 $2500
Driver $3000 ---
Fuel $5000 $10000
Total $10000 $12500
Miles 10000 20000
$/mile $1.00 $0.63
The poor title notwithstanding, the trust issues cited in the article are good to be aware of. Nevertheless, I'll be shocked if self-driving vehicles aren't a technology that utterly reshapes transportation in the U.S.A. in very short order.
I'm amazed that so many people in technology think that these rolling physical software boxes are going to be reliable enough to be better than us. We're pretty good at processing new, ambiguous situations.
We can't even build a secure internet. Which, by the way, these cars will be part of.
I think what will happen is that some accident categories will go down, others involving autonomous cars will go up, and there will be a lot of instances of people who can't get where they need to go because the car can't figure out some ambiguous situation.
Or a car will try to follow a Google route that goes around a block three times before it's (hopefully) finally routed on its way. And there will be rare instances of cars trying to drive to Ireland through the Pacific Ocean.
It's silly to assume that a car is going to trust a mapping service over its own sensors saying "there's no road or path here".
That said, I absolutely do expect cases of "I wanted to visit a friend in the next state, fell asleep overnight, and woke up in a different town of the same name eight states away".
Even with MISRA, we had the Toyota unintended acceleration case (not to say the developers were following it though. Grab the court transcripts if you have the time, they're brilliant). Probably dozens of other corner cases we'll never know about either.
Trucks, buses, and taxis going to airports and hotels, don't necessarily have hard route-finding problems. You also say things like "Can't build a secure internet" as though that's ever been tried, or necessary in this case.
And OK, let's say it's never been tried. Those cars are going to be part of that internet.
Yes, it is.
> Road implementation is a contributing factor, for example.
"Road implementation" is done by humans, so, even if it is a contributing factor, that doesn't make it any less humans' fault.
The accident rate is around 74 per 100 million miles (and fatalities is 1.13).
It's unclear exactly how to turn that into a percentage, but no matter how you do it it's quite high.
Say an accident takes 5 minutes, and people drive 30 miles/hour. Then that works out to 99.999% for humans. If you use the numbers for fatalities then it's 99.99999%.
So the human track record for fatalities is a failure rate of 0.00001% - driverless cars are going to have a VERY hard time beating that.
Would you like to calculate what that works out to for the reliability of each individual computer, or order to have such good results in aggregate?
I suspect that at the end of the day, unless autonomous vehicles are much worse on the fatality rate, they only have to roughly match the human driver-based fatality rate, and commercial interests will push extremely hard for autonomous vehicles. The number of miles driven in the US is tracked by the Federal Reserve among others [2]. Current 12 month moving total is 3,147,579 millions of miles (3.14*10^12).
Latest figure for just crashes (not only those with fatalities) is a 2011 article [3] that asserted 185 per 100 million Vehicle Miles Traveled (VMT). The Insurance Information Institute has a report [4] that cites a $277 billion in 2010 cost of crashes, not including quality of life valuations. Autonomous cars don't have to move the needle that much over the current regime before insurance will put in very strong incentives to use them; there is a ton of room to improve, and the insurance companies stand to gain a huge windfall by not lowering rates commensurately with actual cost savings. The trucking companies must be going bonkers over the possibilities to run diesel trucks 24x7.
[1] http://www.iihs.org/iihs/topics/t/general-statistics/fatalit...
[2] https://research.stlouisfed.org/fred2/series/M12MTVUSM227NFW...
[3] http://www.caranddriver.com/features/safety-in-numbers-chart...
It's bad enough now with ambulance chasers, but when the "guilty" party is a corporation worth billions, it's going to get a whole lot worse.
Personally I'd love self driving cars - tomorrow, if not sooner. Most driving is boring and pointless and I'm totally happy to let a competent machine do it.
But politically too many people have been conditioned to see driving as an expression of personal freedom, and the whole industry is politically vulnerable in a way that the smartphones and the Internet never were.
Generally there's a whole extra layer of legal, political, social, and psychological confusion around any kind of computing that doesn't live in a box with a screen, but starts being present in and manipulating physical reality.
Cars are just the start, and I think they'll turn out to be one of the more straightforward problems.
Conditioned by some of the very same companies that are going to take away the expression, but the citizens will blame the government for this "restriction of freedom," while those same companies keep making profit regardless.
Those companies have been very good at various reshapings of public demand when it served their interests, I'd be very surprised if, when it suits them for automated cars to be the next thing, they aren't just as good as when they made minivans and then SUVs must-haves through marketing (when the real motivation for each was to exploit regulatory exceptions.)
Given the choice for £x a mile for a human driver or (£x * 0.2) for a robot car?
In the meantime, I agree that you can approximate a self-driving car experience pretty closely by ordering an Uber.
That group has already been obsoleted by Tesla. It's embarrassing to the gas car guys that Tesla's family sedan has a better 0-60 time than everything except a few supercars. Supercars that are not only insanely expensive, but unusable for ordinary driving. (Top Gear had a funny video in which they're trying to get a Veyron out of a driveway without scraping the undercarriage.)
If you're really into that stuff, you have to get into electric supercars.[1] A friend of mine who grew up in a rural town fixing cars now works for the company that makes those. Zoom is not dead, but it now sounds like "chunk-whine", instead of "vroom".
[1] http://www.theverge.com/2015/1/9/7517291/renovo-coupe-electr...
I think the term "embarrassing" may be a little excessive here since you are singling out only a single aspect of a car's performance (0-60 time in this case).
Sure, the Tesla beats many (family) cars of a similar price range at short-burst/short-term acceleration, but it falls really short when you need that kind of power over longer periods of time [1], such as around an actual race track with turns. For example, a GTR is known to be a few 10th's slower in terms of 0-60 times, but would destroy any Tesla in nearly all other forms of racing. The GTR is also a 4 seater Nissan, which can arguably be considered just as practical as an electric car of today.
When you compare overall performance to other sport oriented family cars (such as the M3 or M4), the Tesla isn't embarassing anyone.
That being said, I realize the Tesla isn't a purpose built track car, and for what it is, it's one hell of a performer, but I constantly see people bringing up the 0-60 time, saying how it destroys so many gas cars, but it's really more complicated than that.
EDIT: Just to add (in reference to your "you have to get into electric cars" comment), a CTS-V with under $10,000 of modifications will beat the Tesla P90D in a drag race, and still be ~$10,000 cheaper all said and done. I don't think it's possible to add performance modifications to the Tesla at all, so this is another area where gas cars have the advantage.
[1] http://insideevs.com/expected-tesla-model-s-fails-lap-nurbur...
"[If the Tesla didn't go into] reduced power output mode, a B-T-G lap under nine minutes is possible. According to the Bridge To Gantry site, that would put it in the company of some really quick hot hatches."
So if the Tesla didn't have insufficient cooling for going around a race track, it would be about as fast as a Honda Civic Type R. I.e. supercars are way out of its league.
Plus, the Type R is about $40 000. How much is the Model S again?
Fast in a line is not the same as fast overall. One of the best parts of Spec Miata is the frustrated point-bys by Ferrari drivers.
A properly equipped supercar will still eat a tesla in turns and handling etc. The Renovo is actually a supercar and has more going for it than just the power train. (I have been driven in one by the founders)
I am all for autonomous vehicles ruling the road. People shouldn't be driving like idiots there anyway. And I believe they are interesting on the track too: http://selfracingcars.com/
You are just redefining things without proving your point. I do not think you know much about this stuff, just have a bunch of opinions without experience.
Sorry to be blunt.
when autonomous cars dominate the roads, there will still be lots and lots of people driving around in their sports cars or classics or canyon carvers, for fun.
there's a difference between letting/forcing people to adopt new technology in new purchases, and forcing people to give up their expensive cars. pretty much all automotive legislation tries to avoid this scenario.
If the questions were "How safe a driver is your 75yo mother?" or "Do you think autonomous cars will increase or decrease deaths caused by drunk driving?" the survey might have shown different results.
I am TERRIFIED of the "semi autonomous" cars that require the driver's attention to take over if the computer gets confused, like current tesla or mercedes models... because the people driving these cars are NOT going to be able to take over when required. Many people are terrible drivers when they are pretending to pay attention because they are doing the driving, there is NO way they will have the situational awareness to take over in an emergency.
Basically there should be two modes. Full auto when I'm too tired to drive (or sleeping) and semi auto where it takes over in an emergency. And a third mode of full manual for off roading.
You'll need to come up with better names if they're ever going to be sold in Commiefornia \s
For example, as it has been explained to me, self-driving car software will drive defensively by default, and offensively/efficiently when they judge it safe.
Okay, so say self-driving car driving hits critical mass. In this scenario, most people in cars are free to amuse themselves with their phones or reading books, etc while riding their self-driving cars to work. Many are looking down, not at the road, perhaps they're not even in the drivers' seat.
What will happen if there's that one lunatic who opts to drive himself to work and drives in the most offensive, reckless manner? Won't all of the self-driving cars he encounters slam on their brakes to avoid collisions? There likely won't be any accidents, but there will be MANY frightened or startled commuters!
What will this phenomena be called? Road Trolling? And, will there be collective Twitter hashtag outrage that develops as a result?
I can imagine in the most extreme Los Angeles commuter scenario, a serial Road Troll being hunted by one hundred fellow commuters that he rudely cut off on his drive - their auto-drive cars set to "follow that car!", eventually becoming cornered in some residential area when he runs out of gas and then beat to death by the angry mob.
One can be an intensely rude driver and still fall within the law. Trolls might randomly hit their brakes hard just to see the auto-driver behind reflexively lurch and enjoy the passenger's reactions. Probably no one gets hurt, and there's no way to prove the troll wasn't legitimately trying to avoid hitting something.
And driving a manual car will be an expensive luxury, comparatively, because once autonomous cars are mature (which includes being significantly safer to operate), the total cost of riding an autonomous car vs. the total cost (including insurance) of driving will heavily favor the former.
I've done exactly what you've described but also had the passenger telling me spontaneity saying go left right forward.
There is nothing inherent that says you can't do that with autonomous cars. But also it's likely that there will always be cars you can drive for fun just like you can still ride horses for fun.
"pick a random route"
"pick the least-traveled route over the last three months within 50 miles of my location"
I think you'd very quickly wind up going down an unmaintained dirt road and getting stuck.
If you enjoy wandering around in your city: keep a car and do that.
I personally like to drive, especially on road trips, but I like the idea of a car that has much better reaction time that can correct if I do something stupid, or it sees something I can't (via vehicle to vehicle reporting).
Speaking of which, there will be many many opportunities to reduce impact severity. Example: a vehicle pulls out in front of you unexpectedly. If the car can react instantly where you'd take a fraction of a second, it can shed enough speed to reduce the severity of impact significantly (if not avoid it altogether). Reducing speed by 25%, will reduce the impact energy by almost half, so even small improvements will result in big differences in injuries and fatalities.
The actual article presents evidence that Americans are getting more comfortable with self-driving cars, and points out that peoples' concerns aren't very valid.
Honestly, I almost didn't read the article because of the title. Fortunately there was a tiny fit of rage in me that clicked to see what the author actually was saying.
Edit: for posterity, the original HN title was identical to the article's title: "Billions Are Being Invested in a Robot That Americans Don't Want".
"Despite the name 'automobile' these vehicles had no autocontrol circuits; control, such as it was, was exercised second by second for hours on end by a human being peering out through a small pane of dirty silica glass, and judging unassisted and often disastrously his own motion and those of other objects. In almost all cases the operator had no notion of the kinetic energy stored in his missile and could not have written the basic equation. Newton's Laws of Motion were to him mysteries as profound as the meaning of the universe.
Nevertheless millions of these mechanical jokes swarmed over our home planet, dodging each other by inches or failing to dodge. None of them ever worked right; by their nature they could not work right; and they were constantly getting out of order. Their operators were usually mightily pleased when they worked at all. When they did not, which was every few hundred miles (hundred, not hundred thousand) they hired a member of a social class of arcane specialists to make inadequate and always expensive temporary repairs.
"Despite their mad shortcomings, these 'automobiles' were the most characteristic form of wealth and the most cherished possessions of their time. Three whole generations were slaves to them."
[1] https://en.wikipedia.org/wiki/The_Rolling_Stones_(novel)
* First Generation: expensive, mostly working, in very few cars
* Second Generation: bug fixes, slightly less expensive, in more cars
* Third Generation: an add-on option for the vast majority of cars, work 99.9% of the time.
* Forth Generation: politicians stop complaining about all the people having sex in cars.
I don't fully agree - horses didn't disappear, but instead are now mostly kept by people that just love horses, love interacting with them and love riding them.
The same is likely to happen with cars. Most likely people that race them will continue to do so, provided they have the means. If at some point driving a non-autonomous car becomes illegal to drive on certain roadways, they will be trailered to where they are legal to drive. Horses are legal on regular road, but you cannot have them on a Freeway because it's no longer safe.
The history of people being scared and unprepared for technological change is pretty thorough:
http://www.techradar.com/us/news/world-of-tech/12-technologi...
My grandfather grew up in a rural part of Ireland that had no electricity until the late 1920s. One of his earliest memories was when a man came to town in a car with a battery powered lantern. It was the first most people had ever seen if either, and everyone around went to the pub to see the car and light.
The car was an obvious improvement over a horse, in terms of speed, cost and versatility. Very few people voyages more than 10-15 miles from their place of birth pre automobile.
Autonomous cars are not that. They will clearly be expensive, and being a service with complex liability problems will most likely be an Über-style, pay by the drink model. As amazing as Uber is, riding them is a lot more expensive than leasing a Honda Accord.
They also make occasional distance trips prohibitively expensive. I do a 1,000 mile road trip every year -- which costs about $500. What would a robot Uber cost?
IMO, they are a solution looking for a problem. Negative effects will include undermining public transit, and making suburban sprawl worse.
Or I drive for 20 hours and get another week in a 5 star hotel.
The promise of self-driving cars is that, for less money than you're spending to own a car today, you'll summon a self-driving taxi whenever you want, of whatever type you want right at that moment (convertible, economy car, four-door, pickup truck, SUV, limo...), and then after it drops you off, you never see it again.
But for commercial purposes, self-driving cars are inevitable, of course.
When I poke around Google flights at my common trips, it's hard to estimate tightly the cost of a first class ticket, but it seems around $1/mi for most of my routes (some Carribean routes are much cheaper than that). There's no way to operate a jet for that amount. Even if you take a family of four (where the personal airplane cost doesn't go up much, but the first class tickets obviously do), you can't quite cover even the variable costs of the Cirrus SF50 (the cheapest variable cost jet). That doesn't leave anything for maintenance, positioning flights (which require crew), capital costs, insurance, hangaring while not rented, training, or profit for the rental outfit.
I'd love to be proven wrong, because I'm absolutely in the target market for even a $2/mi all-in rental jet (and already have the pilot quals/hours to qualify, except for the type-specific type-rating). I doubt it can be done for under $8/mi to the end user.
But I would LOVE a world in which I can call an autonomous car to my door whenever I want one, just like I'd call Lyft today. Or book one in advance to drive me a greater distance and wait to bring me back.
Get the user experience right. The youtube videos of tesla drivers drinking coffee and reading the paper at high speeds looked like great fun to me.
First priority is not getting me hurt. Second priority is not getting me in trouble. Third priority is not causing damage that costs me. Fourth priority is to go fast.
I thus want a car that can figure out potential hiding spots for cops. When there are no cops, I expect the car to disobey the law. I want it to go 2x the speed limit, drift around corners, and take winding roads faster by sometimes going into the wrong lane.
Let them see it, as a reality and not an abstraction, and then lets see if their minds change.
I have a feeling it's like, in the early IBM PC era, people would have said, why would I want a portable handheld computer with a Star Trek touch screen and a global network, that sounds useful for engineers and nerds and maybe executives. And it doubles as a phone? How very Buck Rogers.
[1] may not contain actual statistics.
Means that there's a market of ~60 million people that would buy an autonomous car.
The last barrier to their legalization will be 'what happens in a crash'. To deal with that, they'll have to be permitted to develop moral opinions as well as the ability to testify about those in court. It's a core feature.
Or people who would "never" drive an automatic car or ... blah blah blah
It's going to happen at some point here ... just like HDTV finally did :)
...no. Turboprop and commercial jet aircraft complemented general aviation, they didn't supplant it.
It's part of why I find the autonomous car argument so absurd. People are investing billions upon billions of dollars to create a more convenient and energy-efficient transportation infrastructure using cars... the LEAST efficient method to do so, and the most expensive to implement.
It's for this reason why I think truly autonomous vehicles are much further away than people think, and when they DO arrive I'd predict that the vast majority of them will be 10-12 passenger on-demand autonomous vans that optimize per-person efficiency and reduce traffic.
I'm living in Europe, in a city, and looking forward to being driven to and from social engagements this weekend, because drinking. There is going to be a human behind the wheel of that taxi but that's incidental. I don't own a car because the costs greatly outweigh the benefits and there are always alternatives.
Interestingly, everyone I know in a city in the US has a car. All they do is complain about traffic, but they own one nonetheless.
For everyone owning a car today, I think tomorrow they'll want a car with self-drive feature.
Its not strange that a survey conducted by car interest groups find their readers uninterested in self-driving cars, as their readers are car enthusiasts. The kind of people who aspire to owning sports cars. The kind of customer who probably doesn't want an automatic either, nor cruise control etc? ;)