Newest Tesla Autopilot Data Shows A 40% Drop in Crashes (bloomberg.com)

There's a surprise in the data from an investigation into Tesla safety by the U.S. National Highway Traffic Safety Administration. An anonymous reader quotes Bloomberg: [W]hile all Tesla vehicles come with the hardware necessary for Autopilot, you need a software upgrade that costs thousands of dollars to make it work. Since buyers can add Autopilot features after purchase, this provides a perfect before-and-after comparison. It turns out that, according to the data Tesla gave investigators, installing Autopilot prevents crashes -- by an astonishing 40 percent...

Now -- thanks to an investigation that initially hurt the company -- there is finally some real data, and it's good news for Tesla... As the software matures to match the new hardware, Musk said on Thursday via a Tweet, Tesla is targeting a 90 percent reduction in car crashes.
UPDATE (5/4/18): The NHTSA has now clarified that their study "did not assess the effectiveness of this technology.

UPDATE (2/16/19): The study's underlying data reveals serious flaws in the methodology that undermine its credibility, according to new analysis from a research and consulting firm.

Comments

Good idea, bad name

[alvinrod]

I think the technology is a good idea, but they've picked a terrible name for it. To someone who is uninformed, it makes it sound as though the feature enables automated driving for the vehicle, and while that may be the end goal, it's currently not at that level and may give a false sense of capability. They should refer to it as "Driver Assist" or something that doesn't leave anyone with a false impression of the capabilities of what it does.

Re:Good idea, bad name

[mrclevesque]

Absolutely. And they should make it much more clear that drivers should keep their hands on the wheel at all times, which autopilot doesn't imply.

Re:Good idea, bad name

[mykepredko]

Agreed. If I bought a car with a feature called "autopilot", I would think I would be comfortable taking a snooze (or watching a Harry Potter movie) while the car did the driving for me. I am really surprised they haven't dropped the name "autopilot" as it is totally misleading and something that has been pointed out repeatedly. I suspect that there are egos involved in the decision not to change the name.

Or, maybe, they think their software is close enough to achieve certification for totally taking over responsibility for driving the car that they think that can weather this storm and keep the name for the big roll out.

Re:Good idea, bad name

[ShanghaiBill]

To someone who is uninformed ...

Since these "uninformed" people don't actually own a Tesla, it doesn't matter one iota what they think about the system. If you actually own a Tesla, the capabilities and limitations of the system are very very clear.

Also, an autopilot on an aircraft doesn't completely fly the plane all by itself either. Pilots understand that. Do you think they should rename it so the passenger in seat 22C also understands?

Re:Good idea, bad name

[Calydor]

So you are fine with only being informed about the capabilities of a product with a price tag of a car ... AFTER you buy the product?

Re:Good idea, bad name

[AthanasiusKircher]

Since these "uninformed" people don't actually own a Tesla, it doesn't matter one iota what they think about the system. If you actually own a Tesla, the capabilities and limitations of the system are very very clear.

And yet from almost the day that the "autopilot" feature became available, videos started circulating online of actual Tesla drivers doing stupid stuff like jumping into the passenger seat or back seat while letting the car "drive."

Whether those idiots are representative of Tesla owners is beside the point. Clearly SOME idiots who actually have access to Teslas have done stupid stuff, and I don't think it's coincidence that this started when the feature named "autopilot" was released.

Also, an autopilot on an aircraft doesn't completely fly the plane all by itself either. Pilots understand that. Do you think they should rename it so the passenger in seat 22C also understands?

Nope. But TERRIBLE analogy. When Tesla drivers have to pass a special exam with the rigor of a pilot's test before operating "autopilot" on their Tesla, then you might have a point. For now, any idiot can drive one of these things... including licensed drivers who aren't the primary driver of the Tesla and may "experiment" without understanding what's really going on.

Re:Good idea, bad name

[Anonymous Coward]

Hmmm.... a pilot who gets (at least) a couple hundred hours of classroom, simulator and instructor led fly time before he's allowed to take the stick himself and even more time and testing before being able to be certified in a plane* so that he can fly solo (and these are the minimums to get even a small plane license) verses people who have a twenty page guide of driving instructions, a 50 question multiple-guess quiz and 20 minutes to prove they can drive around a parking lot and pull the car up beside a curb....

Hmmm.... Such a fair comparison.

*Certified as in on the actual instrumentation of the plane. No where do you have to prove extensive knowledge of the workings of your car's instrumentation. No where.

Re:Good idea, bad name

[Anonymous Coward]

You know, from story after story after story of Tesla Autopilot-related crashes all caused by drivers expecting the car to handle itself without their intervention or attention, I have come to the conclusion that most people who own Tesla vehicles aren't actually aircraft pilots after all!

Re:Good idea, bad name

[ColdWetDog]

So, these Darwin Award candidates are going to go into smart mode just by giving the system a different name?

Stupid always wins.

Re: Good idea, bad name

[Anonymous Coward]

Thanks for trying to keep us in the 20th century luddites, the department for avoiding progress appreciates it

Re:Good idea, bad name

[murdocj]

So, when you tell someone that their car can drive itself, BUT IT CAN'T, are you going to laugh at them?

Re:Good idea, bad name

[HiThere]

There haven't been that many stories, but a few stories have gotten a large amount of publicity, often repeatedly. And there's selection bias, too.

This doesn't prove that it's an incorrect assumption that Tesla drivers try to get the car to drive itself. Just a week ago I saw an electric wheelchair jay walking diagonally across a four lane boulevard (official speed limit 35 mph) while the driver of the wheelchair was busily engaged in texting on their phone. But notice that nobody EVER claimed THAT was safe behavior. So if anything that's evidence that renaming the "autopilot" wouldn't help;

Re:Good idea, bad name

[thegarbz]

I have an idea. Why not inform the customers that their view of what an autopilot does is completely wrong. People will liken it to a plane, ask them if they would be happy flying in a plane without a pilot, and then point them at their car.

People are applying unrealistic expectations based on incorrect preconceptions because they don't understand a technology with the same name. Help them understand, don't just change the name.

Re:Good idea, bad name

[modrzej]

The name "autopilot" isn't confusing at all if you think of it as an analogue of the autopilot in a commercial airplane. The airplane autopilot isn't fully autonomic either; for example, it can't take off on its own. Even at the cruising altitude it requires full attention of the pilots: http://www.cnbc.com/2015/03/26...

Re:Good idea, bad name

[zr]

name is fine. wont take too long before autopilot does exactly what name implies.

however. we can't count on marketing to deliver information.

time to start including basic autopilot skills in the driver ed & license exam.

Re:Good idea, bad name

[zhiwenchong]

Names...

This reminds me of Spanish-French AI character in Space Quest 6 that would pop up when Roger Wilco pressed the autopilot button.

His name was "Manuel Auxveride". :)

Re:Good idea, bad name

[Wycliffe]

Agreed. If I bought a car with a feature called "autopilot", I would think I would be comfortable taking a snooze (or watching a Harry Potter movie) while the car did the driving for me. I am really surprised they haven't dropped the name "autopilot" as it is totally misleading and something that has been pointed out repeatedly. I suspect that there are egos involved in the decision not to change the name.

Or, maybe, they think their software is close enough to achieve certification for totally taking over responsibility for driving the car that they think that can weather this storm and keep the name for the big roll out.

I agree completely. I'm actually surprised that they haven't been sued for false advertising instead just like the cell phone companies have for "unlimited". It should be called "Driver Assist" until the driver is allowed to nap. Yes, I know that a pilot isn't allowed to nap but that's not really the point. A pilot has had a lot more training AND is less likely to hit something if it does dose off AND usually has a second pilot as well.

Re: Good idea, bad name

[Anonymous Coward]

Bullshit.

I test drove a Tesla and it was inoperable without my hands on the wheel except for short durations, and the salesman made it extremely clear exactly what it does. There is no excuse, and nobody is confused. They are just deciding not to heed the warnings and a different name doesn't change that.

Re:Good idea, bad name

[PoopJuggler]

Right, like how the name "automobile" makes people think the car does everything automatically. Oh, wait.

Re:Good idea, bad name

[phantomfive]

Because if they come up with a better name, all problems are solved. They don't have to go around with complicated explanations telling people about their misconceptions. Or maybe you think it's easy to go around telling people they have misconceptions about what words mean?

Rename it, it's just easier.

Re:Good idea, bad name

[phantomfive]

The name "autopilot" isn't confusing at all if you think of it as an analogue of the autopilot in a commercial airplane

Yeah, it's not confusing, all I have to do is think about it for a bit, then read the article you linked to, and everything will be clarified.

As a software developer, I can tell you that all my users are totally willing to put that kind of effort into understanding. They really try to understand I'm not being sarcastic here at all.

Re:Good idea, bad name

[Anonymous Coward]

How can it be false advertising if it works the same way as an airplane autopilot? It doesn't become false simply because people are uninformed.

Re: Good idea, bad name

[modrzej]

Well, every word has its origin, and the origin of the word "autopilot" is in the aviation industry. So, really, I can't understand why people began to think "autopilot" is equivalent to something like "robo-pilot" or whatever else they imagine.

Re: Good idea, bad name

[phantomfive]

Uh.......how much about the aviation industry do you really expect people to understand before they drive a car? Are you trolling or do you really, actually believe that the average person knows that much about avionics?

Re:Good idea, bad name

[Anonymous Coward]

If I bought a car with a feature called "autopilot", I would think I would be comfortable taking a snooze (or watching a Harry Potter movie) while the car did the driving for me.

Then you are fucking stupid, because it does not work like that.

Re:Good idea, bad name

[ShanghaiBill]

So you are fine with only being informed about the capabilities of a product with a price tag of a car ... AFTER you buy the product?

I didn't buy the car, my wife did. We both fully understood that she was not buying a full SDC for the following reasons:
1. We can read.
2. We can see.
3. We can listen.
4. We can think.
People that can do none of these things are going to have a lot of problems in life beyond believing that their car can drive by itself, and it is unlikely many of them are going to be buying a Tesla.

Re:Good idea, bad name

[Anonymous Coward]

I like Co-Pilot personally, but I agree.

Re:Good idea, bad name

[mysidia]

I'm actually surprised that they haven't been sued for false advertising instead just like the cell phone companies have for "unlimited". It should be called "Driver Assist" until the driver is allowed to nap.

The brand name of a product such as AutoPilot is not a legal representation of what its capabilities are.

Even if that were the case, the name is not inaccurate. Some members of the public have a misconception of what Auto Pilot means, and are of the false belief that an Auto Pilot refers to a complete hands-off system that replaces the need for a human operator: That is not the case, and Tesla is not responsible for public misconceptions by people purchasing their product ---- Instead, Tesla's only required to provide truthful information regarding the description of their product, And they are truthful..... Tesla is not lying in their advertising and not announcing that they have a fully self-driving car with no requirement for human monitoring and management.

Re:Good idea, bad name

[mysidia]

So you are fine with only being informed about the capabilities of a product with a price tag of a car .

The brand name Autopilot is Not automatically deceptive, just because some people have a false idea in their brain about what an autopilot refers to. Besides, it's not even proven that a substantial proportion of the population have this false idea. You haven't presented any scientific survey to prove it, so I would say that a few people on Slashdot are probably the only people who hold this misconception as a complete idea.

The Tesla vehicles are not a consumer commodity, or an item presented for sale expecting consumers to not appropriately research their purchase --- You don't spend 5 minutes looking at the sales pitch and then buy a Car or House --- you look it over, ask questions, and get information disclosed from the seller to weigh on your complex decision. Plenty of information has been published by Tesla and 3rd parties about the Autopilot, and consumers can discuss the capabilities of the AutoPilot and other Tesla features with the salespeople before committing to a purchase.

As with all major purchases: the burden of the appropriate Due Dilligence in regards to your purchase lies with the buyer, not the seller. The seller, in this case, has more detailed disclosures available of the capabilities of their product.

Re:Good idea, bad name

[mysidia]

And yet from almost the day that the "autopilot" feature became available, videos started circulating online of actual Tesla drivers doing stupid stuff like jumping into the passenger seat or back seat while letting the car "drive."

That's not because of the name.... People do stupid stuff period. E.g. People do stupid stuff with LaneAssist too, and AutoPilot is not in the name.

What makes the Tesla a little bit different is not the name, so much as the fact that the Car doesn't instantly disengage the assist features if the driver removes hands from the wheel.

Re: Good idea, bad name

[Stewie241]

I don't think the name really has anything to do with it. The big worry about this technology is liability, and there is the idea out there that manufacturers are trying to be care ful about advertising what the car can and cannot do.

Tesla I'm sure makes it abundantly clear that the car needs driver attendance. But if you sit there long enough and the car continually makes good decisions, you are gradually going to become complacent and maybe start to think that it really can do more than you thought. This is when you get into trouble because the technology has made enough progress to be convincing but isn't quite all the way there.

Re:Good idea, bad name

[Anonymous Coward]

Absolutely. And they should make it much more clear that drivers should keep their hands on the wheel at all times, which autopilot doesn't imply.

Or, perhaps they should make sure that Tesla owners can read at a 4th grade level before becoming owners. I'm fairly certain that Tesla lawyers ensured that the EULA describes EXACTLY what "autopilot" does and does not do.

And I grow tired of trying to dumb down consumers with this. Drop the "autopilot name will cause crashes!" argument already. There's not a single fucking commercial aircraft that takes off without a licensed pilot in the cockpit.

Re:Good idea, bad name

[Anonymous Coward]

Because, the popular conception of an autopilot is that it's an automatic pilot, auto and pilot both have meanings and trading on the normal meaning of those parts of the word to say something that isn't true is false advertising.

Bottom line here is that when determining whether or not something is false advertising, they don't go by experts, they go by what the typical person is likely to understand. This isn't that much different from that asinine suit against the HDD manufacturers over their "overstated" capacity claims. Even though people who were tech savvy would have known the difference between base 10 and base 2 measuring conventions.

Re:Good idea, bad name

[Anonymous Coward]

Some? If by some you mean most, then solid point. But, I think most people don't know what an autopilot is specifically, and just go based upon what auto and pilot mean. As in an automatic pilot. If it's automatic, then it shouldn't require your hands to operate. That would be like an assist.

Re:Good idea, bad name

[BasilBrush]

There is some mild rebuking if you don't hold the steering wheel.

If you don't hold the steering wheel then after a time between 1 and 5 minutes (depending on situation) you get an audible and visual warning. Ignore the warning for 15 seconds and you get a "strike". 3 strikes within an hour and you're out: you won't be able to use autopilot until your next journey.

Re:Good idea, bad name

[BasilBrush]

If you were buying a Tesla you'd be so excited about the AutoPilot, you'd find out all about it before you got delivery.

Heck you'd have to read or be taught at least something about it, because you wouldn't know how to activate it otherwise. There isn't a button marked autopilot.

Re:Good idea, bad name

[Dog-Cow]

So, if you're a rich fuck who buys a plane with auto-pilot and takes it for a spin and crashes, you're going to sue Boeing for false advertising?

You're an idiot and you deserve to be the victim in such a crash.

Re:Good idea, bad name

[thegarbz]

Indeed, because new words we discover in the english language instantly make sense and there's no point in a dictionary or wikipedia even existing.

You touched on something very interesting. Knowledge, intelligence, and general inquisitiveness is dead. You just proved it by saying we need to dumb down something so people can understand it rather than explaining to people what they are getting wrong.

Re: Good idea, bad name

[Golden_Rider]

Yes, every word has its origin:

auto = word-forming element meaning "self, one's own, by oneself," from Greek auto- "self, one's own,"

pilot = "one who steers a ship," from Middle French pillote (16c.)

so, "autopilot" = "a device which steers a ship/car on its own". So I'd guess that anybody who knows what "auto/automatic" and "pilot" means but who is NOT a trained pilot and does not know precisely what the autopilot on a plane can and cannot do would immediately guess that a Tesla autopilot CAN steer the car fully on its own.

(btw, as for the plane autopilot - there are enough news stories and videos out there about how modern autopilots can even start and land a plane fully automatically, so why would people NOT think that it can do everything in between fully automatically, too? https://www.youtube.com/watch?... )

Re:Good idea, bad name

[phantomfive]

Knowledge, intelligence, and general inquisitiveness is dead. You just proved it by saying we need to dumb down something so people can understand it rather than explaining to people what they are getting wrong.

Welcome to the consumer economy.

Re: Good idea, bad name

[pezpunk]

Not really. Everyone knows autopilot in aircraft basically just flies the plane in a straight line. Nobody expects to take off or land for the pilot. In that sense, Tesla's autopilot actually does more than it's name implies.

Re:Good idea, bad name

[Anonymous Coward]

The name "autopilot" isn't confusing at all if you think of it as an analogue of the autopilot in a commercial airplane.

 
May be so, but I expect for most people an airplane autopilot is the inflatable one from the documentary "Airplane!"; people who haven't seen it probably understand even less about how it works .

Re:Good idea, bad name

[temcat]

Best comment.

Re:Good idea, bad name

[mrclevesque]

"Or, perhaps they should make sure that Tesla owners can read at a 4th grade level before becoming owners. I'm fairly certain that Tesla lawyers ensured that the EULA describes EXACTLY what "autopilot" does and does not do"

1) Are you saying the EULA is written so a 4th grader would completely understand it?

Re:Good idea, bad name

[mrclevesque]

Interesting, thanks

Re: Good idea, bad name

[Anonymous Coward]

Totally nullifief all the bullshit up this thread. That comment should be first.

Re:Good idea, bad name

[mrclevesque]

"And I grow tired of trying to dumb down consumers with this. Drop the "autopilot name will cause crashes!" argument already"

2) You're allowed to drive with an IQ of 70

Maine Adds Minimum IQ Requirement For All New Drivers https://en.wikipedia.org/wiki/...

100% drop

[Anonymous Coward]

If you just dont ever leave your house.

Re: 100% drop

[Anonymous Coward]

Unfortunately not always the case: https://s-media-cache-ak0.pinimg.com/originals/00/6b/00/006b00668c4ded7f875483cf33a9612c.jpg

Correlation vs Causation

[Anonymous Coward]

Correlation vs Causation anyone?

Re:Correlation vs Causation

[Lord_Byron]

Well, you say that but....

http://peopleoflancaster.com/wp-content/uploads/2015/04/car-crash-house1_6_1438534a.jpg

Re:Correlation vs Causation

[Lord_Byron]

Aaaaand I posted on the wrong comment b/c I'm a moron, apparently. /blush

Re: Correlation vs Causation

[Anonymous Coward]

*crash*

Re:Correlation vs Causation

[HiThere]

I think it's more selection bias on the part of the news media/sources.

Re: Correlation vs Causation

[WindBourne]

Yeah, I tend to suspect that the engineers, staticians at NHSTA have already made sure of this since they are the ones that did the study.

What other factors could be in play?

[Chas]

Could it be, with the crash history of "autopilot", people are now using it more as it was intended? As DRIVER ASSIST, rather than turning it on and dozing off behind the wheel?

I mean, THAT couldn't affect numbers at ALL, right?

Re:What other factors could be in play?

[BasilBrush]

The 40% reduction is of all Teslas that have Autopilot enabled. Right from the first day. And it covers all miles, whether autopilot was enabled or not. (Several of the safety features of autopilot are always on.)

So to answer your question, change of driver behaviour having heard of autopilot crashes could not possibly have affected the statistic, no.

Re:What other factors could be in play?

[Mr D from 63]

The 40% reduction is of all Teslas that have Autopilot enabled. Right from the first day. And it covers all miles, whether autopilot was enabled or not. (Several of the safety features of autopilot are always on.)

So to answer your question, change of driver behaviour having heard of autopilot crashes could not possibly have affected the statistic, no.

Yes, it absolutely could have. Driver's are now more aware of the limits of autopilot and are using it more responsibly. The sample set 'before' installation includes all driving before the rash of media attention. We may have seen a similar drop in the rate with no installation. The only way to truly compare would be large numbers of each for the same timeframe.

And we don't have the raw numbers so we don't even know if they are statistically significant.

Re:What other factors could be in play?

[BasilBrush]

Your argument makes no sense.

Anyway, given that you're unlikely to admit to being wrong. Let's just say that the NHTSA who created this study know a hell of a lot more about how to do statistics of crash rates than you do. Your approval is irrelevant.

Re:What other factors could be in play?

[Mr D from 63]

NHTSA did not claim the drop is due to Autopilot installation. That is where you are confused. They said that there are less crashes in the time period measured after when the upgrades were installed vs. before. Others take that a run with it. But 'before" is not broken down into the earlier time period where more people were abusing the system vs the later period where customers were better informed.

I bet if you would see a significant drop in rates if you started measuring from a time after all the publicity on crashes and abuse and up will the installation of the upgraded system. Unfortunately, we are not given that important data to be able to truly tell the impact of the upgrade.

Re:What other factors could be in play?

[BasilBrush]

You don't win an argument by claiming the other side is confused. I have actually consulted the original document so I know EXACTLY what they measured.

You claim "the earlier time period where more people were abusing the system". Yet they can't have been abusing a system that they didn't have because it was not installed.

Again, the NHTSA know what they are doing with automobile safety statistics. You, have less information and less ability.

Re:What other factors could be in play?

[Mr D from 63]

We all have less information. You assume they are telling us something that they didn't say.

I don't see any comprehension of data normalization for purposes of drawing conclusions on your behalf. The NHTSA didn't draw any conclusions, they presented a specific piece of information and others drew conclusions, including yourself. You did it without the data required.

I have drawn no conclusions, as the information presented is not enough to do so. Which is what I was pointing out.

Re:What other factors could be in play?

[BasilBrush]

I'm not assuming a damn thing. I read the report. Try it.

Re:What other factors could be in play?

[Mr D from 63]

Then tell me the difference between the accident rate before there was publicity about crashes, and afterword but before the new upgrade. If you can't distinguish that impact on accident rates, you should not argue with my point, because there is no attempt to normalize for that factor.

Still not enough data.

[mykepredko]

A crash rate of 1.3/ million miles and having 130 million miles of data means that there has been about 170 crashes.

I'm guessing accidents range from minor fender-benders (although with cars of today, a "minor" fender bender costs $2k+) to the fatal accident.

I would like to know where this 40% reduction takes place in the accident spectrum ? Does this mean that there are much fewer fender benders or fewer accidents which resulted in personal injuries?

If it's at the lower end of the range then big whoopdie fucking do - if it results in fewer injuries (and I would guess this would be significantly more than 40%) then it's something to look into and see if this improvement can be brought to other manufacturer's cars (and trucks).

Re:Still not enough data.

[BasilBrush]

They count as a crash for that statistic if airbags were deployed.

Short-term numbers versus long-term

[dgatwood]

One would expect that. Even a bad computer program with a dozen eyes is likely to be better than a bag of meat with only two.

I'm more concerned about the long-term secondary effects. Do drivers who get used to this technology become dependent on it, and thus have higher accident rates when driving rental cars that lack this technology?

Additionally, I'm less than convinced by the use of a single number here. To be meaningful, you need at least two numbers: the number of crashes avoided because of software intervention and the number of crashes caused by driver inattention. After all, if the system saves a bunch of lives because of things that a human driver couldn't have predicted, but costs a small number of lives because some humans depended too much on the vehicle to drive for them, then it is great from a statistical perspective, but that's little comfort for the families of people who died because the autopilot lulled them into a false sense of security.

Re:Short-term numbers versus long-term

[mykepredko]

Why would you assume "a bad computer program with a dozen eyes is likely to be better than a bag of meat with only two"?

I'm not up on state of the art on computer image/object recognition but the experience I have from about 10 years ago leads me to believe that there are still challenges to be solved, especially when it comes to recognizing movements and intentions. As a driver, some of the cues I rely on include turning indicators, wheel positions, other driver/pedestrian/cyclist eye contact as well as sounds.

I'm not saying that the challenge of coming up with software that allows a car to autonomously drive itself better than a human isn't possible. I just challenge the assertion that a computer with multiple cameras is likely superior to a human.

Re:Short-term numbers versus long-term

[west]

To be meaningful, you need at least two numbers: the number of crashes avoided because of software intervention and the number of crashes caused by driver inattention.

I think that two numbers would be deceptive because almost no-one is capable of acknowledging their inattention. If you found at that that 50% of accidents are caused by inattention, but the autopilot is a 20% *worse* driver than someone paying attention, you *know* that everyone would flee from AutoPilot it on the assumption they won't be part of the 50% failing to pay attention.

One of the primary problems is that humans (in general) are incapable of acknowledging the weaknesses that cause accidents, thus making it very hard to take measures that reduce the number of accidents. Splitting the numbers apart would contribute to this problem.

Re:Short-term numbers versus long-term

[Namarrgon]

Probably because the bag of meat's eyes are too often turned elsewhere.

Re:Short-term numbers versus long-term

[dgatwood]

I think that two numbers would be deceptive because almost no-one is capable of acknowledging their inattention.

They don't have to. With as much data collection as the Tesla systems do, assuming they collect the same data with autopilot disabled, too, it should be possible to do a post-mortem (so to speak) on a random sampling of accidents and determine whether a reasonable person should have noticed the stopped car in front of them (for example) or not and whether the driver failed to react in a timely fashion or not.

I think that two numbers would be deceptive because almost no-one is capable of acknowledging their inattention. If you found at that that 50% of accidents are caused by inattention, but the autopilot is a 20% *worse* driver than someone paying attention, you *know* that everyone would flee from AutoPilot it on the assumption they won't be part of the 50% failing to pay attention.

On the contrary. If the autopilot is 20% worse than a driver who is paying attention, then having those concrete statistics would provide the motivation to change the behavior of the autopilot feature to be more sensible, such as looking for signs that the driver isn't paying attention, and then automatically engaging when the driver's hands leave the wheel, when the driver's eyes leave the road, when the driver's grip on the wheel relaxes too much, etc., rather than making things worse by engaging when the driver would have done a better job. And as the statistics become more complete, you'd probably decide to add other weighting factors, such as time of day, whether they're driving away from home at night (e.g. to work the night shift) or towards it, etc.

Re:Short-term numbers versus long-term

[tempo36]

I'm anti-antibiotic and modern medical intervention because I think knowing that they're available just makes people careless and sloppy when they travel in areas where those interventions aren't available. I would much rather a few more people die because we don't use antibiotics at all than for people to become reliant on them and just become careless and unfit.

I agree that there could be secondary effects, but the more logical conclusion from my standpoint is that if the technology definitively improves safety then eventually it should become standard across the board in all cars and then the issue of "dependence" will eventually become moot. Pointing out that people who don't use the technology have worse outcomes than those who do is an endorsement of the technology, not a reason to block the use of it.

Re:Short-term numbers versus long-term

[fluffernutter]

Someone else who sees one bad driver out of a thousand and lets them ruin their whole day. Suddenly human drivers can't do anything right. The fact is that I have been on the road with many human drivers that drive just fine.

Re:Short-term numbers versus long-term

[thegarbz]

Why would you assume "a bad computer program with a dozen eyes is likely to be better than a bag of meat with only two"?

Because it's deterministic. Because it's mass upgradable. It may not be physically better right this moment, but it is conceptually and philosophically far better than the situation we have now.

Re:Short-term numbers versus long-term

[swillden]

I'm not up on state of the art on computer image/object recognition but the experience I have from about 10 years ago leads me to believe that...

Others have already responded to your other points, I just want to point out that experience from 10 years ago tells you basically nothing about the state of the art today. Deep learning methods have enabled dramatic progress on exactly the class of pattern matching problems that includes computer vision.

Personally, I still think that LIDAR is inherently superior to video cameras for this task, but Tesla's numbers are impressive, and prove that while their system may not be all that it should be, it's already better than a typical human driver -- at least than the typical Tesla buyer (note that I have no reason to believe that Tesla buyers would be worse than average drivers, but the possibility shouldn't be ignored).

Re:Short-term numbers versus long-term

[dgatwood]

I'm not saying that the challenge of coming up with software that allows a car to autonomously drive itself better than a human isn't possible. I just challenge the assertion that a computer with multiple cameras is likely superior to a human.

I say that for several reasons:

• Human vision is inherently focused on a single thing at a time. They teach you to move your eyes around and scan for things that might be problems, but the reality is that we're very limited in our ability to do so. Computers don't have that problem. They can see that kid on the side of the road who might fall out into traffic long before a human driver would happen to randomly glance in that direction, which means that on average, they can take corrective action much sooner even if their actual reaction time is much slower.
• Computers can also look behind and beside them constantly. You might look in your side or rearview mirror when you're about to turn or change lanes or back up. However, the odds of seeing someone cutting into your side or flying up behind you in time to avoid a collision is remarkably small. A computer, however, would see those vehicles every time, and would often be able to prevent the resulting accidents.
• Statistically, one in five collisions happens in parking lots, where human vision is hopelessly obstructed by other vehicles. Computers should be able to trivially avoid essentially all of those collisions. So right off the bat, even if computers were no better than human drivers while on the road, you'd expect a 20% drop in accidents just from having complete 360-degree vision while pulling out of parking spaces. And nearly half of all pedestrian accidents occur in parking lots, so the seemingly excessive caution that computer-controlled cars use should dramatically decrease pedestrian injuries and deaths as well.

Besides, Tesla's autopilot feature is designed exclusively for highway driving. (AFAIK, it still ignores stop signs and traffic lights entirely.) Highway driving is, on the whole, some of the safest driving possible, with nearly every accident caused by some combination of fatigue, distraction (particularly involving food/drinks), and/or drunkenness on the part of one of the drivers involved. To beat a human driver under those conditions, all Tesla's autopilot really has to do is keep the car in the current lane, reliably detect cars that have stopped in front of it without nodding off after half an hour or chugging one for the road, and avoid other people who have fallen asleep or are drunk. Of those, only the last one is particularly challenging, which is almost certainly why the crash numbers are only down by 40% instead of 80% or more. :-)

Re:Short-term numbers versus long-term

[Anonymous Coward]

Before the age of antibiotics, people would routinely die from paper-cuts. Paper-cuts were fatal because typical human skin carries organisms (Staphylococcus aureus & S. epidermidis) that become infectious and fatal when they get into the skin instead of just on the surface.

People who don't use the technology (antibiotics) have worse outcomes than those who do is an endorsement of the technology (antibiotics), not a reason to block the use of it.

You really should reconsider the logic of your arguments.

Re:Short-term numbers versus long-term

[Anonymous Coward]

> I'm anti-antibiotic and modern medical intervention because I think knowing that they're available just makes people careless and sloppy when they travel in areas where those interventions aren't available. I would much rather a few more people die because we don't use antibiotics at all than for people to become reliant on them and just become careless and unfit.

*cough*

_I_ would much rather a few people die because they got careless and sloppy when travelling in an area where antibiotics were not available than having many, many, _many_ orders of magnitude more people die of things that are (often trivially) TREATABLE WITH ANTIBIOTICS!

When's the last time you had first-hand knowledge of someone who wasn't immunocompromised or immunosuppressed who got a minor cut and died from the resulting infection? Never? Maybe once in your entire life? You know why? Because of antibiotics. That shit used to happen ALL THE TIME. Now it doesn't, _because of antibiotics_.

Re:Short-term numbers versus long-term

[dgatwood]

I'm anti-antibiotic and modern medical intervention because I think knowing that they're available just makes people careless and sloppy when they travel in areas where those interventions aren't available. I would much rather a few more people die because we don't use antibiotics at all than for people to become reliant on them and just become careless and unfit.

First, antibiotics are available nearly anywhere in the world you might go. By contrast, these sorts of autopilot features are available on a tiny fraction of a percent of vehicles, and probably will be for some time to come.

Second, the technology is highly limited, basically useful only on the highway, which means that if people get used to having that extra support during highway driving, it could easily result in an increase in accidents in cities, where accidents are much more likely to cause pedestrian fatalities. So there's a good possibility that this could actually make traffic deaths worse on the whole over the long term.

Early studies strongly suggested that partial self-driving solutions did more harm than good, which is why I think we should wait to make self-driving technology available until it can truly take the place of the human driver, rather than introducing a solution that only works part of the time and can lead to false confidence the rest of the time. I could be wrong, and I'd like to be wrong, but my gut says we'd be better off waiting a few more years for a more complete solution, rather than deploying a partial solution more broadly.

Re:Short-term numbers versus long-term

[dgatwood]

My opinion has nothing to do with bad drivers. Everybody gets tired. Everybody gets distracted. Anyone who says otherwise is kidding him/herself.

Besides, more than 70% of all drivers eat while driving, and that's responsible (according to one study) for about 80% of all crashes. When I say humans suck as drivers, I mean that the overwhelming majority of human drivers (if not all) suck at driving at least some of the time. The only reason we don't have orders of magnitude more wrecks than we do is that split-second reaction time is only important on very rare occasions (perhaps five-seconds in a typical hour of driving), so being distracted usually doesn't result in an accident.

Re:Short-term numbers versus long-term

[fluffernutter]

Who cares how many people eat while driving? What matters is how many of those that eat get into accidents they otherwise wouldn't have. Split second decision making is not often required because humans anticipate danger fairly well, even if they are eating at the wheel. By relying on AI we are relying on a situation that will rely entirely on split second judgement because that is all AI will be able to do. Right now I am not confident sensors are even robust enough to sense a situation which requires a split second decision.

Re:Short-term numbers versus long-term

[quantaman]

One would expect that. Even a bad computer program with a dozen eyes is likely to be better than a bag of meat with only two.

I'm more concerned about the long-term secondary effects. Do drivers who get used to this technology become dependent on it, and thus have higher accident rates when driving rental cars that lack this technology?

That's a concern, but my bigger worry is the seat belt effect. That in response to the perception of better safety people start to take more risks.

That's also the major reservation with this data set. These are all users relatively new to the auto-pilot, I know if you installed an auto-pilot in my car I'd be pretty damn paranoid for the first few months and my accident rate would plummet too. I'm not certain they're measuring the safety benefit of the auto-pilot or just their own drivers being extra careful while using it. Similarly drivers might start getting really careless as they start to trust the software, ie changing lanes without a shoulder check because they expect the Tesla to catch it.

Additionally, I'm less than convinced by the use of a single number here. To be meaningful, you need at least two numbers: the number of crashes avoided because of software intervention and the number of crashes caused by driver inattention. After all, if the system saves a bunch of lives because of things that a human driver couldn't have predicted, but costs a small number of lives because some humans depended too much on the vehicle to drive for them, then it is great from a statistical perspective, but that's little comfort for the families of people who died because the autopilot lulled them into a false sense of security.

I've been really critical of Tesla so far and their previous "statistical" safety claims have been ridiculous.

But other than sample size and the reservations I mentioned I think this is the proper test. Comparing "crashes avoided because of X" is a really arbitrary and hard to judge standard, plus it avoids secondary effects such as driving conditions and changes in behaviour.

The proper test is the total accident rate for Tesla drivers with the auto-pilot vs without, with proper controls to handle confounders. But this is a data set that needs to be continually monitored since both the auto-pilot software and drivers are moving targets.

Re:Short-term numbers versus long-term

[Anonymous Coward]

"the experience I have from about 10 years ago"

Re:Short-term numbers versus long-term

[Anonymous Coward]

I'm not up to the state of the art on mobile network technologies, but the experience I have from about 10 years ago leads me to believe there's still challenges to be solved, especially when it comes to delivering video or web content. I don't think we'll ever get to "web on your phone" at this rate. I remember that Iphone thing coming out, so maybe that helped. Its about time to check back in and find out!

Re:Short-term numbers versus long-term

[Anonymous Coward]

I just wonder if it knows that 27 states require you to pass cyclists with 3-5 feet clearance. I wonder how it handles that case, and does it do the right thing and slow down if a car is coming from the opposite direction and the Tesla can't move over on a two lane road?

Re:Short-term numbers versus long-term

[HussDelRio]

I just challenge the assertion that a computer with multiple cameras is likely superior to a human.

Computers don't get distracted, sleepy, angry, or become ragemonsters. Meatbags are inherently bad at motor vehicle operation.

Re:Short-term numbers versus long-term

[geekmux]

I just challenge the assertion that a computer with multiple cameras is likely superior to a human.

Computers don't get distracted, sleepy, angry, or become ragemonsters.

Oh really? Computers essentially can inherit all of these traits when they have hardware failures, software glitches, or become hacked.

Meatbags are inherently bad at motor vehicle operation.

And computers (or more to the point the meatbags programming them) are inherently bad at security, which causes them to become as vulnerable as meatbags, if not worse.

Re:Short-term numbers versus long-term

[Anonymous Coward]

I'm not up on state of the art on computer image/object recognition but the experience I have from about 10 years ago leads me to believe that there are still challenges to be solved, especially when it comes to recognizing movements and intentions..

There have been huge advances in computer vision in the last 10 years.

Re:Short-term numbers versus long-term

[bgarcia]

Personally, I still think that LIDAR is inherently superior to video cameras for this task

I think that Tesla's approach of combining RADAR and cameras is both technically superior and less expensive than LIDAR. RADAR can work in many weather situations that render LIDAR useless. And RADAR's ability to bounce off the road surface allows the vehicle to see past the car directly in front of it.

Re:Short-term numbers versus long-term

[bgarcia]

It boggles my mind that random nutjobs on slashdot keep thinking that they've come up with some corner case that the people who develop autonomous driving technology - who work on this very subject as their career - have never bothered to think about.

Re:Short-term numbers versus long-term

[wvmarle]

Early studies strongly suggested that partial self-driving solutions did more harm than good, which is why I think we should wait to make self-driving technology available until it can truly take the place of the human driver, rather than introducing a solution that only works part of the time and can lead to false confidence the rest of the time.

I have the same feeling about it. Make the human do more, keep them attentive, and let the car correct when an accident is about to happen (e.g. slam on the breaks when a sudden obstacle appears, or help keep traction when a drastic turn is made).

I could be wrong, and I'd like to be wrong, but my gut says we'd be better off waiting a few more years for a more complete solution, rather than deploying a partial solution more broadly.

Simple counter argument: deploying even partial solutions at a rather large scale, like Tesla is doing, provides a great source of real world data and experience for the software developers. Sure things are bound to go wrong sometimes in the autopilot in its current state, some things that are simple oversights or bugs, other things that may have been unforeseen such as a truck that almost perfectly matches the colour of the sky behind it.

Re:Short-term numbers versus long-term

[Anonymous Coward]

Technology from 10 years ago? Please. The smartphones were in their infancy 10 years ago and look where they are today. You cannot begin to make comparisons to 10 year old tech, especially with the explosion of GPU-powered processing.

Re:Short-term numbers versus long-term

[catchblue22]

I'm not up on state of the art on computer image/object recognition but the experience I have from about 10 years ago leads me to believe that there are still challenges to be solved, especially when it comes to recognizing movements and intentions.

Neural networks have come a LONG way in ten years, due in large part to the exponential growth in processing power in GPU's. Neural nets can perform the same or better as humans in a variety of image recognition tasks. For example, neural nets have been trained to give the prognosis for cancer patients based on images of tumors. The networks were trained on thousands of known images of previous cancer patients along with medical histories. When new images were passed through the network, the prognosis, including likelihood of survival was given, based upon the images of previous patients.

If you really think about it, when humans drive, we are largely doing simple image recognition. White line, yellow line, double line, car in front, car in back, pedestrian about to cross, bicyclist riding on side of road, etc. There is some context, yes, but really in most cases the decisions we make in driving are quite automatic and shallow. They are in large part rule based, learned from long habit. There are still likely to be cases at times when simple image recognition and habitual rule based behaviour will not suffice; in such cases, yes self driving cars might have trouble. However I would assert that such corner cases are likely to be rare. The advantages of having a computer driving will be that they won't get drowsy or distracted, and will have far more information input about the surroundings of the car, gleaned from eight cameras, several ultrasound sensors, and a radar system that can detect cars in front of the car in front of you. To a certain extent, driving is simple. If there is an object in front of you, or beside you, don't hit it; stay in the correct lane; don't go too fast into corners. I think that on the whole, computers are likely to be better at this than humans.

Why not 100%

[Anonymous Coward]

Why don't they target a 100% reduction in crashes?

Re: Why not 100%

[Anonymous Coward]

Im targeting 108%

Exactly how was the study done?

[paradigm82]

I can't completely wrap my head around how they could do this study and how the statistics were calculated.. The article states this is a before/after test so presumably involves comparing the the crash rate before auto pilot with that after. But it is somewhat difficult to do because did they really include drivers that crashed (before getting into AutoPilot) to see if they would crash again (after AutoPilot)? I would imagine many people might even stop driving after a crash or at least have gotten a new car which might not be a Tesla. In other words, it seems that from the point on where AutoPilot is bought then the group is already suffering from self-selection bias since many of those who suffered accidents (reflecting a prone-ness to accidents) were selected out in the pre-AutoPilot phase, thus only getting to count for the pre-AutoPilot data. And they can't know how many, of those who crashed, who would have bought AutoPilot, had they not crashed, and what their accident rate would have been. It seems this would tend to make it seem as if AutoPilot increased safety eve if it didn't.
Another way of comparison could be to compare drivers buying AutoPilot vs not buying it from a given point in time. But that causes another problem namely that those buying the AutoPilot might not have the same risk/accident-rate as those not buying it. It could be they are richer, older, more careful etc.
The only fair way to do such a study would be to, at some point in time, randomly select a group of drivers who will get AutoPilot for free and another group who won't be able to use it. Then you can study exactly what is the safety effect of giving a driver AutoPilot.

Re:Exactly how was the study done?

[Calydor]

It's called a control group.

They literally have the same car - except for AutoPilot being enabled.

They are completely randomly selected across the nation based on one criteria: Buying a new car.

Then they compare crash rates per mile driven for the two groups: The ones with and the ones without AutoPilot.

You don't have to check if the guy that crashed without AutoPilot also crashes with it.

Re:Exactly how was the study done?

[paradigm82]

No because it specifically says: "[W]hile all Tesla vehicles come with the hardware necessary for Autopilot, you need a software upgrade that costs thousands of dollars to make it work. Since buyers can add Autopilot features after purchase, this provides a perfect before-and-after comparison."
There is absolutely no reason to think that the group voluntarily electing to spend a lot of money on AutoPilot has the same overall risk as those who doesn't. So you can't compare "Drivers who bought AutoPilot after it became available" with "Drivers who didn't buy AutoPilot after it became available". So what could they compare? It seems they are compaing "Risk of drivers before they bought AutoPilot" vs "Risk of drivers after they bought AutoPilot". I.e. you only consider those drivers who ultimately bought AutoPilot and compare their risk before and after. But that is not a valid comparison either because if people have a crash they may exit the whole car-driving thing and are hence not available at all for consideration into the study as someone who would/could AutoPilot. If they hadn't crashed at that time they might have gone on and bought an AutoPilot and crashed there. So the people buying AutoPilot already represents a survivor group. Hence studying that group might not give an indication of the true impact of AutoPilot on new Tesla buyers (who haven't proven themselves as survivors yet).
Randomly assigning AutoPilot to a random half of some randomly selected pool of drivers, and not to the other half, would have allowed for a true comparison.

Re: Exactly how was the study done?

[mspohr]

They compared crashes with autopilot on to crashes with autopilot off. Same car and driver. Car crashes. .. was autopilot on or off.

Re: Exactly how was the study done?

[fluffernutter]

But people drive in way more dangerous circumstances with autopilot off. That isn't seriously how they measured it, is it?

Re: Exactly how was the study done?

[Anonymous Coward]

You still don't get what paradigm82 is saying. If someone crashed and therefore they got rid of their car, then they weren't around to buy autopilot. So such people could not have been a part of the study. Therefore the study could only contain people who didn't crash and destroy their car right away. They might be better drivers. Yes, you're comparing their crash rates from before and after, but "they" are (slightly) different group from "all Tesla buyers".

Re: Exactly how was the study done?

[esperto]

To me Tesla lost a great opportunity to make it 100% reduction, just turn autopilot off immediatly before any crash, then you would never have autopilot on... profit /sacarsm

Re: Exactly how was the study done?

[mspohr]

They do?

Re:Exactly how was the study done?

[enrique556]

Came here to say something similar. Men pay higher car insurance premiums (at least in my country) because they crash more than female drivers. Are they worse drivers than women, or do they enjoy driving more, so take more risks? People who buy autopilot upgrade would not be the ones who want to race their tesla. A person who has bought autopilot but then has it turned off is more likely to be taking risks. To put it another way, if you have the option of not having to drive your car, what reasons (besides cost) do you have left to drive it?

Re: Exactly how was the study done?

[BasilBrush]

No they didn't. The statistics cover Teslas without autopilot vs Teslas with autopilot. The second group including all miles, whether or not autopilot was actually engaged. This is because:

a) Some of the safety features of autopilot are on all the time even when the full autopilot isn't engaged.

b) If they did what you suggest, they'd be comparing mostly urban travel vs mostly highway travel. As autopilot is only really designed for the highway at this stage.

Re: Exactly how was the study done?

[Anonymous Coward]

You are slightly misunderstanding.

They compared crashes with the autopilot feature turned on in the software delivery (but maybe or maybe not actually enabled at the time of the crash) with crashes with the autopilot feature turned off in software (and so the driver had no option to use it).

Just having the option to use the autopilot software reduced crashes by 40%. Note, that this means that some of the crashes that happened to people who had paid for autopilot may have happened because they chose not to enable the autopilot even when they "should" have done.

Re: Exactly how was the study done?

[Carewolf]

They compared crashes with autopilot on to crashes with autopilot off. Same car and driver. Car crashes. .. was autopilot on or off.

And the autopilot is only supposed to be on when driving on highways where the rate of crashes is 10x smaller than on average. So does that mean it crashes 4x more often than humans on highways?

Re: Exactly how was the study done?

[fluffernutter]

Sure.. Autopilot won't back out of your driveway for you, or out of a parking spot. It only works when the conditions are clear.

Re: Exactly how was the study done?

[mspohr]

You can turn the autopilot on at any time. Divided highway or narrow two lane roads. It's especially good in traffic.

Re: Exactly how was the study done?

[mspohr]

Autopilot can be active at any time. It's especially good in heavy traffic. Works well on two lane roads as well as highways.
They compared accidents driving with AP ON vs AP OFF.

Re: Exactly how was the study done?

[BasilBrush]

Autopilot can be active at any time. It's especially good in heavy traffic. Works well on two lane roads as well as highways.

I said it was only really designed for highways. Not that it can't do other roads.

They compared accidents driving with AP ON vs AP OFF.

You are completely wrong. Consult the report. They compared cars without AP installed vs those with it installed. Using the data for cars with AP installed, regardless of whether it was active or not.

This is marketing tripe

[Anonymous Coward]

I'll bet the sample size is in the dozens, and autopilot had only been it for a few years. There isn't enough data for me to believe anything. If I cause a wreck and I'm behind the wheel, it's my fault. If I die, I'm fine with that, I could have prevented it. If a computer kills me because some idiot programmer forgot a semicolon, I'll be pissed.

Can't wait for the lawsuits

[xxxJonBoyxxx]

>> installing Autopilot prevents crashes -- by an astonishing 40 percent...you need a software upgrade that costs thousands of dollars to make it work

This kind of read like: "So...you want the software that shuts off your fuel value during a crash or the one that opens it full on impact. It's a $5K option...your choice, really."

If it's really just a software option, doesn't this sound a lot like the VW software "option" that cheated on emissions tests?

Re:Can't wait for the lawsuits

[tempo36]

No, it's a fancy tech feature that you're paying for (or not) which is optional. That's like complaining that one ski jacket has a built in avalanche transponder (some do) and another model by the same manufacturer doesn't and that somehow the manufacturer is liable for your injury when you choose to not purchase the additional safety feature.

At some point if the NTSB mandates autopilot then it will be on all cars, but until then it's just a feature just like any number of others. Seats that reduce whiplash by moving in a crash, pre-crash seatbelt tensioners, car hoods that actuate to minimize pedestrian injury, emergency braking, blind spot warning...etc.

Re:Can't wait for the lawsuits

[fluffernutter]

That's what bugs me about this. Corporations are on their way to determining who lives and who dies based on how much one can pay. Not passively like having a better airbag, but actively.

Re:Can't wait for the lawsuits

[Anonymous Coward]

You realize that airbags were originally only available on high-end cars? Same with anti-lock brakes, and most other things we all take for granted.

Let the rich folks pay for the development of stuff the rest of us will get for much cheaper in a few years. Think if them as guinea pigs.

 

Re:Can't wait for the lawsuits

[wvmarle]

Isn't that the case in the US for a long long time already, somewhat solved a few years ago but since yesterday back to how it was?

ER staff: "Sorry, this patient can't pay for his treatment, please take him out again."

Re:Can't wait for the lawsuits

[fluffernutter]

You realize auto speed control has been available in high end cars for ten years now and without a legal motivation to do it, car manufacturers have filtered it down to very few affordable cars? Automation won't be filtered down either.

Re:Can't wait for the lawsuits

[fluffernutter]

This compounds the problem. Now you're hoping your automaker thinks you're important enough and your family is important enough to look out for you and your price point. When will we have to pay more money if we want to drive at highway speed?

Wile E. Coyote

[Anonymous Coward]

Could someone paint fake road markings and cause a Tesla to crash? Perhaps into a wall with a fake/painted tunnel?

Someone needs to try this out.

Re: Wile E. Coyote

[Anonymous Coward]

It has radar, so the painted tunnel likely wouldn't work. It may be possible to trick it, but you'd likely also end up tricking human drivers too. The difference is that human drivers can't all be fixed with a single software update. Of course all human drivers can't be hacked, but hacking all cars to become killing machines would require so much development, testing, espionage, coordination, all without ever being discovered, I just don't see it happening. And even if it did, cars aren't made to go everywhere, so once word got out, people would be able to get to safety and the cars would be easier pickings than various species we've driven into extinction. It's hard for humans to wipe out all humans because effective destruction requires the kind of large long term cooperative embetterment that isn't conducive to people with destructive desires. In order to become destructive they need to be peaceful for generations, but if they're peaceful for generations they'll likely no longer want to be destructive. The only only way to be destructive over the long term is to continuously foster a destructive culture and a destructive culture isn't able to progress sufficiently to become effectively destructive. It wasn't always this way but the cumulative enrichment through the progress of technology and science have made the world this way. There may eventually be doomsday devices that can't be countered, but considering that the vast majority of people strongly want to preserve the human species, I think there will always be sufficient precautions in place to prevent any such hypothetical device from getting into the hands of someone wanting to annihilate mankind.

To someone dumb as fuck, you mean.

[Anonymous Coward]

And, frankly, someone that dumb was going to crash sooner or later anyway, so this at least delays the inevitable.

Re:To someone dumb as fuck, you mean.

[Mr D from 63]

The data does not tell us the new software is safer. It tells us the accident rate has dropped. That may have happened even without the software update simply from all of the publicity and attention on how previous crashes were tied to negligent use. There has been a ton of reinforcing communication to Tesla drivers regarding the limits and we no longer see videos of people showing how cool their hands free driving experience is.

Also, we were not supplied with the raw data so we don't even know if the result is statistically significant.

Go on, list them

[Anonymous Coward]

Story after story? List them, with the dates. Lets check it off against the number of people dying after putting "cruise control" on and taking a nap, hmm?

Re:Go on, list them

[murdocj]

Really? Maybe YOU could cite some examples of people thinking cruise control actually drove the car? If you mean "people becoming drowsy on long trips" that has nothing to do with Tesla story.

Re:Regulatory Solutions

[saloomy]

But the AutoPilot in an aircraft and and in this case the Tesla do the same thing. They assist in control of the aircraft. AutoPilots do not replace the pilot or relegate his/her duties to that of a passenger, you still have to have a pilot or you aren't going anywhere. The name is appropriate.

The Tesla Autopilot is pretty good for how early in car automation we are. It would be a terrible inconvenience to require you to put your hands on the wheel when you are driving thru some stretches of the American south-west, like going between Phoenix and LA on i10. Just like it would be a terrible inconvenience to expect the auto-pilot of a passenger jet liner to require input every three seconds before it disengages between Heathrow and JFK. There is no reason for it. There is no need for a law.

Will accidents happen? Sure. Just like aircraft accidents still happen. The goal of the technology is to reduce the chance of an accident happening, not eliminate it altogether, while providing some amount of relief for the pilot / driver. It is pretty hard to keep at attention for many hours on end when the task is so mundane. As the technology gets better and better, and more wide-spread, we will see more and more benefits from this technology, like reduced energy consumption by drafting closely to the car infant of you, or reduced traffic congestion by synchronized acceleration, intersection control, etc.. etc..

Re:Regulatory Solutions

[Anonymous Coward]

Wouldn't disengaging the autopilot when the driver's hands are off the wheel cause more crashes? It sounds like that would be negligent in the extreme. The autopilot should not disengage until the driver's hands are ON the wheel to assure continuous control of the vehicle through the transition.

Media Reminder:

[BigBuckHunter]

Just a reminder,

Expect media outlets whose owners will benefit financially from Tesla's success to report this in a positive light and pimp it hard.

Expect media outlets whose owners will benefit financially from Tesla's failure to report this in a negative light, bury it, or begin advertising sponsored competitor's autopilot as being superior.

Does anyone have a working link to the actual report? It was supposed to be at static.nhtsa.gov/odi/inv/2016/INCLA-PE16007-7876.PDF

I coulda used this

[rsilvergun]

when I got creamed out on my road bike, assuming it can detect pedestrians. I was carrying the damn thing across a cross walk (with the little green walky man no less) and somebody ran the light. They clipped the wheel of my bike or I wouldn't be here right now. The lady stopped. It as broad daylight but somehow she didn't "see" me. Folks tune out when driving. That light almost never turns red so it didn't occur to her to stop. Put another way they're driving with their lizard brains. I'd rather they do it with an electronic brain instead.

Re:I coulda used this

[david_thornley]

My new car is supposed to see pedestrians and brake to stop them. I haven't actually tested this yet, due to a lack of experimental subjects.

Re:Regulatory Solutions

[Dutch Gun]

People keep making this argument about the analogy between Tesla's feature and the autopilot feature of airplanes, but how many normal people know an airplane's autopilot works? It's a pointless comparison unless it's a widely known fact among the general public, which I'd argue it is not. I'll bet most people have the same mis-impression of an airline's autopilot feature.

Even so, I think we're just in a collective learning curve regarding semi-autonomous vehicles. Eventually, the cars will become fully autonomous anyhow, so I'm not terrible concerned. The fact that collisions are down by 40% validates what many of us long believed, which is that computers are going to be much safer drivers than humans. And this is just a very early and flawed first iteration of the technology to come.

Lies, damn lies and statistics

[Anonymous Coward]

How do we define a "crash", and how do we know that for example major crashes increased in frequency but minor ones decreased?

So you don't know what they tell you.

[Anonymous Coward]

Maybe you need to find out first, eh?

I laugh at you.

Ha ha ha ha ha ha.

You first, cupcake.

[Anonymous Coward]

I see a hard lack of list off you, yet you were thinking "story after story", seems that is the only story you got, cupcake.

Re:Regulatory Solutions

[superdave80]

I don't understand why you are comparing pilot terminology to everyday person terminology. 99% of the population doesn't speak pilot-ese.

'astonishing'?

[superdave80]

installing Autopilot prevents crashes -- by an astonishing 40 percent...

I think everybody knew that a solid autopilot system was going to be better than human drivers on average. Why is the author astonished by this?

Re:'astonishing'?

[Anonymous Coward]

Probably because Tesla's driver assist functions (aka "Autopilot") are anything but "solid", and are actually worse than the equivalent gear in a lot of similarly priced European cars.

The Tesla automatic emergency braking system didn't even detect a truck right in front of the car at all, even though the truck was there for approx 7 seconds. Even when the Tesla drove directly under it, killing the driver, the AEB still didn't kick in, and in fact the car only stopped - after driving through a fence - when it couldn't physically keep going.

So personally I'm astonished that Autopilot prevents any accidents, given that apparently the car won't even stop itself when there are easily detected objects right in front of it...

Re:Regulatory Solutions

[saloomy]

But I'm willing to bet 100% of Tesla owners have flown SOMEWHERE before, where an AutoPilot was used, but never without a PILOT. The fact that it's industry jargon doesn't matter. It's becoming closer and closer to our natural understanding every day. The vehicle warns you anyhow, and even if you thought "well there wasn't a pilot in the airplane who flew me to LAX last time I traveled, this car must be able to run sans-driver!", the car tells you strait up that you must keep at attention.

If a car tells you "5 star safety rating!" and you elect not to strap the car seat in properly *despite* all the warning labels and instructions, does that mean you can sue the auto manufacturer, or fault them because "5 star means perfect"? No. Of course not. Implying otherwise is obtuse.

Re:Regulatory Solutions

[saloomy]

People keep making this argument about the analogy between Tesla's feature and the autopilot feature of airplanes, but how many normal people know an airplane's autopilot works?

That doesn't matter. The car specifically tells you to pay attention. So even if your perception was "The one time I flew, I didn't see a pilot or hear an announcement." You are still told that you have to maintain control of your car. Warning label argument still applies.

It's a pointless comparison unless it's a widely known fact among the general public, which I'd argue it is not.

But the general public doesn't matter. Its the segment of the public that can afford $80,000 on a car. I'd make the argument they tend to fly quite a bit and notice the $150,000/year employee the airlines keep on staff who sit at the front.

I'll bet most people have the same mis-impression of an airline's autopilot feature.

They have that warning label again, and anyone who has purchased one of these vehicles knows pretty fast the features and limits of its "autopilot".

Even so, I think we're just in a collective learning curve regarding semi-autonomous vehicles. Eventually, the cars will become fully autonomous anyhow, so I'm not terrible concerned. The fact that collisions are down by 40% validates what many of us long believed, which is that computers are going to be much safer drivers than humans. And this is just a very early and flawed first iteration of the technology to come.

Here I absolutely agree with you, 100%.

Re:Regulatory Solutions

[mysidia]

Wouldn't disengaging the autopilot when the driver's hands are off the wheel cause more crashes?

Well, they can do a forced transition back to driver control by having the system refuse to maintain speed and start applying brakes, until either the vehicle reaches a complete stop and remains stopped, or the driver puts their hands back on the wheel, and transition is then completed.

Re: Regulatory Solutions

[Anonymous Coward]

I've been on a plane for more hours than you have been alive, most of that in the seats at the front, yet I have not seen in detail how pilot and autopilot interact. Especially since your Saudi allies toppled a few buildings im NYC 16 years ago.

Airplanes are slower than cars

[aberglas]

If an airplane autopilot goes bad it would take many tens of seconds if not minutes for disaster to ensue. They are basically dumb things that maintain direction and altitude. If the pilot looks out the window, it takes many seconds for an on coming plane to arrive. More likely they rely on TCAS, the transponder based alerting. But it all happens quite slowly during a cruise, which is why pilots can study a map while flying.

Cars on the other hand are often only a second or two from disaster. Swerve into oncoming traffic and things happen very fast indeed.

Re:Airplanes are slower than cars

[mlyle]

> If an airplane autopilot goes bad it would take many tens of seconds if not minutes for disaster to ensue.

Tell that to the guys in the bizjet who were killed when the autopilot trimmed to the maximum stop fighting a small control column force and then disconnected/"let go" and killed everyone in the back from head trauma.

Re:Airplanes are slower than cars

[mrclevesque]

Of course there is the rare exception.

Airline pilot checking in...

[PilotKnob]

Just to chime in about the "Autopilot" name: The name is irrelevant. You're the one driving the car, and you alone are assuming the responsibility to know how it operates AND when to use it properly to both the limits of yourself and the system, nomenclature be damned. But if you ask me, "Autopilot" is just a bit aspirational. Does one "pilot" a car, or "drive" it? I've always "driven" cars and "piloted" airplanes. But YMMV, naturally. I'm not mad about it, just sayin'.

What I actually came here to ask is whether there is any data separating out airline pilots using these Tesla systems to see if mandatory training for proper use of appropriate levels of automation in varying conditions has any outcome on the incidence of crashes. What I'd wager is that this training and experience might carry over from one area to the other, seeing reduced incidences of crashes by airplane-autopilot trained drivers. This could mean that until we get to truly autonomous cars, it'd be wise to at least have some sort of minimum training requirement before enabling levels of automation on an individual driver basis.

Re: Airline pilot checking in...

[WindBourne]

Actually, every new Tesla is delivered to your door and then they spend several hours with you teaching you everything about the car. Our Tesla is a 2013 so does not have this. But from what I have heard, on the new cars, they spend about an hour on the AP getting customer comfortable. And consider the fact that this is only used on the most dangerous conditions ( highways/rural roads vs slow urban ), and still has way less than average, I would suggest that is proof enough that Tesla has done enough training.

Re: Airline pilot checking in...

[PilotKnob]

You know, I'm not even saying that Tesla should be responsible for providing this training, except maybe to have an easy way to make sure the Autopilot functions remain disabled if the customer isn't comfortable with it or doesn't feel adequately trained in its use. Not likely with a Tesla purchaser, I realize. Again, I'm putting the responsibility directly on the operator of the vehicle to ensure they have the necessary skills to appropriately use the automation. How they gain those skills is up to them.

From the beginning I've been a huge proponent of this type of technology. We've all seen how our populace is more interested in surfing than driving, and it's now proven to be saving lives and reducing collateral damage. I couldn't be happier, and this is the way we need to be encouraging all manufacturers to go. Tesla has the balls to push out this technology when it can be a safety asset, even in Beta. I give them all kinds of kudos for that, and will consider their vehicles first during my next round of shopping. I'm currently rocking a '13 Leaf, and it meets 90% of my driving needs, and I have a gas-banger for my long-range road trip car. The pickup truck only gets driven when my wife needs a load of mulch or I'm planning a big day at Home Depot or Lowes.

Re: Regulatory Solutions

[Dog-Cow]

Your post makes so little sense that it's worth removing your head from your neck so we can take a peak at what's going on inside.

Wouldn't you be more cautious

[Anonymous Coward]

If your trying something like a auto pilot system. Wouldn't you as a human be more cautious and aware? You can't program the auto pilot to drive aggressively or make foolish moves on the highway. Its whole purpose is to drive conservatively and with caution. I do not think we have anywhere near enough auto pilot like vehicles on the road compared to human drivers to make a claim that they are safer. Only that we can safely create technology that is trying to prove that. For example, let's put a autonomous race car on the track and see how it does against human drivers?

Re: Regulatory Solutions

[Anonymous Coward]

Thanks for the chuckle! Mod this guy up!

Re:Regulatory Solutions

[Highdude702]

I cant do what you did.. i dont have the patience so i applaud you because i would have said "YOURE A FUCKING MORON BUT YOURE KINDOF RIGHT ON ONE PARRT!! RARWWW! lol. either way hopefully one day i will learn to be diplomatic

Re: Regulatory Solutions

[Highdude702]

Wish i had mod points +2 Funny!

Re:Regulatory Solutions

[Highdude702]

I'm sure after one or two 'Impeding Traffic' Tickets they will learn. Sounds like a lovely idea, I believe they already do something like this though. seeing as they have buzzers that tell you to put your hands on the wheel. there has to be some kind of a time limit where after X seconds decelerate

Re:Regulatory Solutions

[Dutch Gun]

And you'd be yelling obscenities at someone who probably agrees with you in most aspects on this topic. All I'm saying is that comparing Tesla's 'autopilot' it to it's original moniker isn't a very compelling argument - nothing more, nothing less.

Contrary to what you and saloomy probably believe (based on the vociferous rebuttals) I don't actually believe that Tesla needs to change it's 'autopilot' name. Even if they called it "minor driver assist but please still pay attention to the road", stupid people are stupid, and will do stupid things regardless, while most people will use it responsibly after learning how it works. Of all the people driving Teslas, how many people were honestly confused by the name? I'm guessing not a lot, if any. So, the entire argument seems sort of silly to me.

See, that's why it's not a bad thing to remain civil.

Re:Regulatory Solutions

[Highdude702]

I actually agree 100% with that. I don't feel everything I do should be regulated due to a few stupid people that don't care enough to read or learn something.

Re:Regulatory Solutions

[mrclevesque]

Maybe you would care more if a stupid person killed someone close to you.

Re:Regulatory Solutions

[mrclevesque]

But the average traffic a plane encounters isn't at all like the average traffic a car encounters.

Re:Regulatory Solutions

[Highdude702]

I'm talking about warning label stupid. Them drinking a gallon of bleach isn't going to kill anybody in my family unless they were one of the dumb ones. We have tampered with natural selection with our regulations and safety labels. And now we have an overpopulated mess of a world. I wasn't talking about the "I wonder if this gun can really kill somebody" stupid. Those people should be killed and their parents jailed for child abuse.

Re:Regulatory Solutions

[mrclevesque]

"there has to be some kind of a time limit where after X seconds decelerate"

It seems like it takes from 1 to 5 minutes for a warning depending on road conditions, and then a series of 3 warnings every 15 seconds after that.

Re:Regulatory Solutions

[mrclevesque]

"I'm talking about warning label stupid"

That is very common, even in higher iq people, and anyone with an iq of 70 can drive a car. So if you don't want to get hit by a car driven by a 'stupid' person it's good to tone down the 'autopilot' analogy.

"We have tampered with natural selection with our regulations and safety labels"

Some low iq people are more intelligent than some high iq people in many situations. Natural selection seems to prefer a large spread of iq. Human's are part of the process of natural selection, we can't step out of it, i.e. in a general sense, if we tamper, the tampering has been naturally selected.

Re:Regulatory Solutions

[Highdude702]

but now kind of slowing of the vehicle? that seems stupid.

Re:Regulatory Solutions

[Highdude702]

you went from we have to do something, to "in a general sense, if we tamper, the tampering has been naturally selected." Basically saying whatever is done is fate. so in that case why do anything? right?

Re:Regulatory Solutions

[mrclevesque]

"so in that case why do anything? right?"

If one thinks natural selection implies we can't choose or our choices are meaningless, maybe, but I don't, and natural selection doesn't imply it.

My point second point was directed towards arguments in general that use the idea that x is wrong because it is tampering with natural selection, which I disagree with for the reasons I wrote.

Re:Regulatory Solutions

[superdave80]

But I'm willing to bet 100% of Tesla owners have flown SOMEWHERE before, where an AutoPilot was used, but never without a PILOT.

Well, no shit. Why would you think there wouldn't be a pilot, even if there was some form of 'autopilot' in use? Even if there was a point A to point B, fully autonomous airplane autopilot, there is still going to be a pilot on board. A pilot still being on board tells us NOTHING about the level of autonomy of the plane.

BS statistics

[sentiblue]

So the 40% claim is deceiving.... Let's be clear about this: Let's say for example last year there were 2 crashes and this year there was 1.... that doesn't mean it's a 50% drop... the percentage is only high because the number of total incidents is low.

Now if last year had 200 crashes and this year 100.... I will DEFINITELY believe it's a 50% drop....

Similar to how a product is newly rated ... there are 2 ratings: 1x5stars and 1x1star.... this averages to 3 stars... that doesn't mean anything until there's more ratings.