Self-driving car advocates often note that more than 90 percent of serious accidents result from driver error, and thus estimate that autonomous cars will reduce fatalities by 90 percent. Indeed, in 2008 a National Highway Traffic Safety Administration (NHTSA) study of crashes between July 2005 and December 2007 found that 5,096 were caused by driver error, while just 130 were caused by vehicle failure and 135 were caused by the weather.
After some adjustments, NHTSA concluded that 93 percent of accidents can be attributed to driver error. So it seems reasonable to conclude that self-driving cars will save more than 31,000 lives per year (i.e., about 93 percent of the 33,560 fatalities suffered in 2012).
Not so fast, says a group called the Casualty Actuarial Society. It took a close look at NHTSA’s 2008 study and found that “49% of accidents contain at least one limiting factor that could disable [autonomous vehicle] technology or reduce its effectiveness.” That means self-driving cars will only reduce fatalities by about half, not 90 percent. While 16,000 lives saved per year is nothing to complain about, there’s a big difference between 16,000 and 31,000.
A close look at the Actuarial Society’s data (summarized here), however, provides more grounds for optimism. First, of the 49 percent of accidents that self-driving cars supposedly won’t prevent, 32 percent of them (about two-thirds of the 49 percent) would supposedly result because the drivers were too stupid, too inebriated, or too “aggressive” to actually use the self-driving option and insisted on taking control themselves. “Drivers who state they ‘always drive aggressively’ may choose to drive themselves,” says the summary.
The Antiplanner doesn’t buy this reasoning. While I understand that “100 percent adoption does not equate to 100 percent use” (at least initially), no one says that seat belts don’t save lives because not everyone uses them. Seat belts do save lives if people use them, and self-driving cars will save lives if people use them.
The study blamed another 12 percent of “driver-error” accidents on inclement weather. “Currently, automated vehicle technology cannot operate in inclement weather,” so the study assumes the driver would take over and suffer the same accident as a non-self-driving car. However, there are only some kinds of inclement weather than self-driving cars currently can’t handle (mainly snow and ice that obscures lane stripes), and there’s every reason to believe they will be able to handle such problems by they time they are introduced to the market.
Overuse can lead you http://deeprootsmag.org/2014/12/02/christmas-close-home-hearth-heart/ order viagra online to ulcer and inflammation of the stomach and food pipelining. Impotence is all about bad erections or the erections free generic cialis which are not so firm. We realize the importance of a locksmith, only when we move back into the body and in particular the info about side effects, which comes with the bad, and there are times discount viagra from canada when every couple experiencestrying times in the relationship. Remember to take the medicine with full large glass of water to ensure the sufficient canadian viagra blood supply to the penis muscle and helps sustain the erection of a man until he achieves orgasm, thereafter the operate regresses.
Another 12 percent of accidents were found to be associated with vehicle or infrastructure problems such as braking deficiencies, traffic signal failure, or potholes. While these weren’t usually the main cause of accidents, they could contribute to accidents even of self-driving vehicles. However, if the accident is caused by vehicle failure, it isn’t due to driver error.
In short, nearly all of the driver-error accidents that the report says self-driving cars won’t prevent are due to drivers taking over from the self-driving car. But the long-run vision for autonomous vehicles is vehicles than have no option for drivers to take over. So, in the long run, self-driving cars really will reduce accidents and fatalities by around 90 percent.
In related news, the Antiplanner’s argument that self-driving cars would eliminate the need for mass transit in most cities was reported by Congressional Quarterly’s Roll Call news service. But another report by the same writer focused on an answer I gave during the Q&A session of last week’s policy forum.
Asked what effect self-driving cars would have on local governments that get a high percentage of their receipts from traffic tickets, I noted that not only would those governments have to find another revenue source, but hospitals will need to find another source of organ transplants as some 70 percent of organs come from victims of traffic accidents. While that’s not necessarily a bad thing, an advertising firm called Sparks & Honey put a more positive spin on both effects.
Page 43 of the firm’s 62-MB presentation on self-driving cars notes that such cars will allow us to “enjoy more efficient government” because local government will have to cut budgets in response to declining ticket revenue. Page 38 says “we will see a surge in 3D printed organs” because organs available for transplant will decline. The 52-page presentation makes a number of other interesting predictions, most of them positive, though it predicts that self-driving cars will lead to more substance abuse because people won’t need to stay sober to drive.
Economically, a change that forces people to adopt new technologies (such as 3D printed organs) isn’t always a good thing if that new technology is more expensive than the old. But when you combine the high cost of organ transplants (including the high risk of rejection and the lifestyle changes imposed on recipients in order to minimize rejection) with the fact that 70 percent of organ transplants require someone’s death in a vehicle accident, then an alternative, if it can be found, is probably worthwhile.
I love that Mr. O’Toole advocates for self-driving cars. I too like the idea and hope to see it in full effect. I think that a partially or fully automated self-driving car system will surely help bolster many of Mr. O’Toole’s anti-transit arguments and help erode many arguments of transit advocates. I also believe that automated vehicles will reduce the number of accidents over time (you can split hairs as to how many).
However today’s post is rife with technological determinism and I don’t see the autonomous car as some grand panacea (which seems to where Mr. O’Toole’s posts on the subject are heading). I can tell he’s getting too excited about the idea with “such cars will allow us to ‘enjoy more efficient government’ because local government will have to cut budgets in response to declining ticket revenue.” Let’s pump our breaks here đ Yes we may see a decrease in ticketing and moving violations, but how many new bureaucracies will a driverless car system create? How many existing agencies will need errr request additional funding once these systems start to be implemented? Starving one beast and feeding it to another isn’t really solving a problem.
While I agree with Mr. O’Toole on the virtues of driverless cars, technological determinism is dangerous. Driverless cars will solve more problems than they create, but they will create problems. Furthermore, while advances in technology have made our lives more productive and comfortable, the pending technologies do not absolve us from our personal responsibilities or our responsibilities to eachother. I’m not saying this is the argument Mr. O’Toole is making, I’m saying that this is the road technological determinism inevitably leads you down. Tis better to sip the Kool-Aid than guzzle it (Hezekiah 2:15).
The AP, in writing about self driving cars does not take into account the inherent complexity of navigating through 4 dimensions, 3 of space, 1 of time, and human frailty. Such navigation obviously requires computers with extreme capabilities, and extreme fail-safe protection, even though the 4-space involved is mostly an oblate spheroid of a few hundred yards around the vehicle.
The real problem is the software. I write the stuff and have to deal with it every day. I’m typing this on a Windows-8 machine that needs periodic re-starts because of some gigantic fundamental flaw in Windows that the systems people still cannot find. My Firefox browser often dies because of the Flash plug-in, or runs so slowly I have to re-start it. My QuickBooks periodically tells me my data files are corrupted. How’d that happen, buddy – you’re the only one what accesses my data files? Why did you corrupt them?
The software for automated vehicles will be orders of magnitude more complex. What am I do do at 80 mph if the system freezes or needs a re-start, a major software component simply stops or the data files get “corrupted”? Or the data files claim the road is in A-1 shape, but a herd of construction workers have torn it up and forgotten to upload the coordinates and details of the damaged road? Or the master system, even more complex, failed and the data did not get downloaded to my car? What happens when poorly designed or tested system upgrades are distributed and fail?
I’m gonna die, that’s what I’m gonna do.
Without question any such system would be government run. How often do DMV systems go down? Does the Obamacare web system work properly yet? Can’t even sue the bastards – most of them have immunity.
And please, don’t tell me that Google writes better software, or anyone else for that matter. It’s shite too, just has a different smell. When you leave street view in Google Earth you do not get back the same view that you had before you went into street view, just a seriously distorted view from altitude zero, which you have to manually fix. Some of their data is getting ancient, outdated by new construction. etc. etc. I just returned from driving Frankfurt to Prague to Munich to a village in south Moravia to Karlovy Vary to Frankfurt. Google maps was good, not great, with many inaccuracies.
If self driving cars work as well as AP expects one way to get people to let the cars drive is to aggressively prosecute bad drivers. Today if you kill someone with your car in a fit of rage or because you were playing with your radio you’re likely to get off without so much as a traffic ticket. The politics of requiring a level of proficiency in vehicle operation that for many is likely unattainable is messy when the alternative is an infrequent bus that takes twice as long if they’re lucky. Even when laws are passed police refuse to enforce and DAs refuse to prosecute. Once the driver has an option to let the car drive itself then bad drivers can be held accountable, including the aggressive ones who will make better time in a city than the self driving car could.
“Today if you kill someone with your car in a fit of rage or because you were playing with your radio youâre likely to get off without so much as a traffic ticket.”
It’s hyperbolic nonsense like this that make the rest of the comment unreadable.
“And please, donât tell me that Google writes better software, or anyone else for that matter.”
You don’t want to hear the truth that Chrome OS is in many ways superior to Windows and OS X?
http://cironline.org/reports/bay-area-drivers-who-kill-pedestrians-rarely-face-punishment-analysis-finds-4420
Sixty percent of the 238 motorists found to be at fault or suspected of a crime faced no criminal charges during the five-year period, CIR found in its analysis of thousands of pages of police and court records from Alameda, Contra Costa, Santa Clara, San Mateo and San Francisco counties.
When drivers did face criminal charges, punishment often was light. Licenses rarely were taken away. Of those charged, less than 60 percent had their driving privileges suspended or revoked for even one day, an automatic penalty in drunk driving arrests.
Forty percent of those convicted faced no more than a day in jail; 13 drivers were jailed for more than a year. By contrast, those charged in accidental shootings often serve lengthy jail terms, according to media reports.
District attorneys fail to charge the drivers, saying juries sympathize with the motorists. Police occupied with other crimes donât track down hit-and-run drivers. The state Department of Motor Vehicles doesnât always pull licenses, at times hindered by the morass of paperwork required to get information from courts and police departments.
Echoing a common sentiment among prosecutors, San Mateo County District Attorney Steve Wagstaffe said jurors see the cases as tragic accidents with no guilty party: âTheyâre all thinking, âThere but for the grace of God go I.â â
Give them the option of letting the car drive them safely and they won’t think like that. They’ll vote to convict the reckless maniac who wanted to save a few seconds. Driverless cars can be a boon to the liveable streets movement.
“Google maps was good, not great, with many inaccuracies.”
I work in rural America quite often. The Google data in NM for example is laughable. Street names are inaccurate, Town or destination locations will be in open fields 2 miles away from any road. Apple maps is no better. I asked Siri where the New Mexico State Corrections Facility in Santa Fe was and she responded “I have no record of New Mexico.” No joke.
ahwr has gone from massive generalizations about ALL drivers to a specific town in California. Generalization followed by cherry picking. Classic.
But I do empathize with the sentiment; I’ve complained here about a-hole drivers who blow stop signs, nearly running me (and my dog) down. The tailgaters. The speeders. Jerks. Every last one. More than jerks. Criminal jerks.
More laws aren’t needed. Current laws need to be enforced.
I also agree that driverless cars may take the a-hole drivers out of the equation, thereby increasing safety. Yes, there are technical hurdles to overcome. The world was once skeptical of the Wright Brothers and flight, but look at our aviation system today.
And to return to the software red herring, look at modern autopilot software for aircraft. How often do you hear about it failing? There are redundant systems and computers in place.
Frank is there any large metro are where this isn’t the case? Everywhere I’ve been has been the same. Even in allegedly walkable cities like NYC, Portland, seattle, San Francisco etc…it’s so common for people to drive that they sympathize with the drivers like them who don’t pay enough attention to the road.
” The world was once skeptical of the Wright Brothers and flight…”
And we trust our lives to automated flight all the time. To me driverless cars have the feeling of inevitability. Seems like it’s not if but when.
“is there any large metro are where this isnât the case?”
IDK. Skimming news articles, it seems like a lot of vehicular manslaughter is attributable to drugs/alcohol. Found the first charges against a driver in CA from 2011 who was distracted by a phone and killed people in a crosswalk. There should be a multi-city study, but I suspect you’re right. The problem may go beyond sympathizing with other drivers to lawyers’ ability to create reasonable doubt. Nevertheless, it’s a real problem.
I think you are putting the cart before the horse. Sure some will choose to upgrade, but probably , lets say 30% will not because they can’t go out an by a new car, or choose not to. Many will choose to drive just like today – Many will choose to not let “Big Brother” control their destiny, and I’m sure many, if not all, will choose to drive more than 30% of the time, despite the capability of their automobile. Gas prices are declining, probably for the foreseeable future, and people will choose bigger trucks and SUV’s over fancy gadget filled automobiles.
Especially when those cars can’t tell the difference between a rock and a crumpled up newspaper or understand that a cop is directing traffic or know to stop for a traffic light not in its map.
Especially when those cars canât tell the difference between a rock and a crumpled up newspaper
If radar is used, the density of an object can be determined. This is also only a factor when an object is directly in the tires’ path and there is no room to avoid it. At any rate, software can identify objects.
or understand that a cop is directing traffic or know to stop for a traffic light not in its map.
“The software can recognize objects, people, cars, road marking, signs, and traffic lights, obeying the rules of the road and allowing for multiple unpredictable hazards, including cyclists. It can even detect road works and safely navigate around them.” Source.
Tonight, there was a police officer barricading a street in my neighborhood because 60 mph winds toppled a tree, which fell across the road. The current autonomous car would recognize the barricade and re-route. Certainly, the same would be true for recognizing a person. Certainly, it would be possible to communicate with autonomous vehicles electronically (think of the Waze app) to alert vehicles to detour.
The same source shows that in 2012, the autonomous car completed 300k of test miles without an accident. The average US driver has an accident every 165k miles.
So…
Well, no, radar cannot tell the density of an object.
Put a sheet of tin foil out in front of a radar and it looks as dense as a mountain.
“Well, no, radar cannot tell the density of an object.”
Of course it can, or perhaps you can provide evidence to support your assertion. Radar used in meteorology certainly can detect particle density.
“Put a sheet of tin foil out in front of a radar and it looks as dense as a mountain.”
Do you have evidence of this? This seems to contradict your assertion.
And somehow, during 300k of tests, no sheets of aluminum foil foiled the autonomous test vehicle, and its radar, lidar, and other sensors manged to avoid objects, including people and other cars.
All of those miles were driven on a small set of road miles meticulously mapped before hand with a human driver ready to take over when the car required it.
I didn’t mean the car would hit the traffic cop. It would see him as an object to avoid. It just couldn’t tell if it was being waved through, around some obstacle.
http://www.technologyreview.com/news/530276/hidden-obstacles-for-googles-self-driving-cars/
Kinks to be ironed out.
Chrome OS may very well be “in many ways superior” to windows. Even so, I’m not prepared to risk my life with software “in many ways superior”. I also have a fundamental distrust of Google because of their political dishonesty, but most importantly because their human interface is horrible, vile, beyond disgusting. Unclear, opaque even, unchangeable options, an assumption of superiority that means only, to me anyway, trouble and more trouble.
Bennett: “And we trust our lives to automated flight all the time.” No, we do not. Aircraft have pilots, always, because we emphatically do not trust our lives to the autopilot. Not to mention air traffic controllers.
Hey there AP, any chance you could change your web-site options so that replies to comments appear with the original comment in thread format so the replies make more sense?
The technophobia is strong on this board.
Libertarians should embrace tech because technology is liberty.
It’s understandable that statists and totalitarians would oppose tech, but come on crotchety old dudes. Tech will undermine big government.