SoylentNews is people
SoylentNews
While it may not be all that surprising to SN readers, some data on "self driving" cars has now hit the big time, WaPo reports: https://www.washingtonpost.com/technology/2022/06/15/tesla-autopilot-crashes/
Tesla vehicles running its Autopilot software have been involved in 273 reported crashes over roughly the past year, according to regulators, far more than previously known and providing concrete evidence regarding the real-world performance of its futuristic features.
The numbers, which were published by the National Highway Traffic Safety Administration for the first time Wednesday, show that Tesla vehicles made up nearly 70 percent of the 392 crashes involving advanced driver-assistance systems reported since last July, and a majority of the fatalities and serious injuries — some of which date back further than a year. Eight of the Tesla crashes took place before June 2021, according to data released by NHTSA on Wednesday morning.
And 5 of 6 fatalities were linked with Tesla cars, the other was one of the competing Level 2 systems offered by other automakers.
WaPo continues,
The new data set stems from a federal order last summer requiring automakers to report crashes involving driver assistance to assess whether the technology presented safety risks. Tesla's vehicles have been found to shut off the advanced driver-assistance system, Autopilot, around one second before impact, according to the regulators.
The NHTSA order required manufacturers to disclose crashes where the software was in use within 30 seconds of the crash, in part to mitigate the concern that manufacturers would hide crashes by claiming the software wasn't in use at the time of the impact. [Ed: Emphasis provided by the submitter.]
(Score: 4, Interesting) by JoeMerchant on Monday June 20 2022, @02:36PM (17 children)
>The comparison is only relevant assuming folks use autopilot all the time (I guess most do not).
a) most Teslas have not been delivered with autopilot activated (correct me if I'm wrong...)
b) if autopilot is used in less than 3% of Tesla miles driven, then it is on-par with human drivers
c) the people who have access to the "real" data know figures for crashes per mile driven, and can probably break those down between high and low crash areas / times of day. If they wanted to share, they would - I'm going to guess that puts the Tesla autopilot at more or less equivocal standing with human driver performance, possibly +/- a factor of 2 which is within the error bars of the available data. If it were dramatically better Musky would be all a tweeter about it. If it were dramatically worse, Musky's detractors would be pounding out press releases on an hourly basis, just after shorting TSLA stock.
d) there are few enough autopilots on the road that autopilot on autopilot interactions are quite rare. Most accidents are caused by one party, and assignment of blame in the official reports can be motivated by all kinds of arbitrary things having little or nothing to do with the correctness of operation of the involved drivers. All this to say: you wouldn't expect autopilot accident statistics to be too much different (factor of 2) from human drivers, because they are out there in a sea of human drivers with the same (thankfully rare, but not rare enough) bad actors that all drivers deal with putting them in the same unavoidable circumstances that human drivers find themselves in in most accidents.
🌻🌻 [google.com]
(Score: 2) by legont on Monday June 20 2022, @03:01PM (14 children)
In addition, most folks are actually sane enough to use autopilot in better conditions while taking over in difficult ones.
"Wealth is the relentless enemy of understanding" - John Kenneth Galbraith.
(Score: 3, Insightful) by ikanreed on Monday June 20 2022, @03:13PM (13 children)
What experience have you had with human beings that leads you to believe that to be true of most of them?
(Score: 2) by JoeMerchant on Monday June 20 2022, @03:32PM (12 children)
The remarkable thing about people is that the vast majority of them are relatively good. The bad actors make the news, cause the damage, but if you pick a random cross sample of 100 humans from most populations, you'll find 99% of them are pretty O.K.
Think of it this way: sit yourself in a "high crash" area and count cars passing by between crashes. Even if you're in a remarkably bad spot that gets 3 crashes a week, you'll probably see tens or hundreds of thousands of cars pass without significant incident between crashes.
🌻🌻 [google.com]
(Score: 2) by ikanreed on Monday June 20 2022, @04:35PM (1 child)
"Relatively good" doesn't mean concerned about others, sensible, and aware of relevant facts to their choices. It just means the first one.
(Score: 3, Interesting) by JoeMerchant on Monday June 20 2022, @05:29PM
Oh, hell yeah. Saints are rare. But, by and large, people aren't out to screw each other, especially at potentially high cost to themselves - a fair number want to screw with others, but most of them stop short of causing serious harm.
And: are the other drivers paying attention? I will say this, before the "no texting while driving" laws passed here, you could watch passing traffic and over 60% of drivers would be holding a cellphone while driving by. After the law passed, that number is down under 5% - but the number of people staring at their lap has increased from about 2% to maybe 5%. Still, remarkably, even so impaired, most of these good people still keep it in their lane and stop before rear-ending the car in front most of the time. I say most, one fine rainy morning a neighbor rear-ended me on the main road very hard. My 77 GMC actually came out slightly improved from the collision, it straightened a kink that had been recently put in the rear bumper. Her SUV was totalled, and she had to take off her seat belt to climb over the steering wheel to retrieve her cellphone from the windscreen (90s cab forward style huge dashboard). I also have had a couple of near misses where the driver who slammed their brakes at the last moment simultaneously lost grip on their phone and it flew into the windshield-dashboard meeting point. But, that's one rear-end collision and maybe three near misses in a million miles of driving... they usually get it right.
Oh, and I love classic cars, but I will NEVER drive one without headrests in traffic. It only takes one rear-end collision in a 1960s style seat without headrests to end up paralyzed for life - or, slightly better, dead.
🌻🌻 [google.com]
(Score: 3, Touché) by DeathMonkey on Monday June 20 2022, @08:45PM (9 children)
Well driving is basically the most dangerous thing you could possibly be doing at any given point in time so "good" isn't exactly how I would describe our performance!
If you have the most dangerous job in the world you're still more likely to die driving to it than working on it!
(Score: 2) by JoeMerchant on Monday June 20 2022, @09:16PM (8 children)
>If you have the most dangerous job in the world you're still more likely to die driving to it than working on it!
Almost: try deep ocean construction welder.
🌻🌻 [google.com]
(Score: 3, Informative) by DeathMonkey on Monday June 20 2022, @09:50PM (6 children)
Commercial diving has a fatality rate of 3.4 to 4.2 deaths per 100,000 divers. [marineinjurylaw.com]
Driving resulted in 11.7 deaths per 100,000 people [iihs.org]
So driving a vehicle is about 3x more fatal than commercial diving, per capita.
(Score: 2) by pvanhoof on Tuesday June 21 2022, @12:44PM (1 child)
Non-professional Scuba diver here: the reason why there are fewer accidents with diving is because more passionate people will train harder to be good at what they are passionate about. Diving requires passion. Similarly, open source software development used to yield better programmers on average. As you had to be passionate about it to get involved. Let average people, who do drive a car, do daily diving: most of them will die within a month. At around their 100th dive. The 100th dive is in the diving community considered as your most dangerous dive. That's because then you feel like you can do it fluently.
(Score: 2) by DeathMonkey on Tuesday June 21 2022, @03:15PM
Good. I'm passionate about workplace safety!
(Score: 2) by JoeMerchant on Tuesday June 21 2022, @06:18PM (2 children)
Or. a 15% fatality rate, depending on where you get your statistics.
https://www.offshoreinjuryfirm.com/offshore-injury-blog/2015/october/underwater-welding-accidents-how-common-are-they/#:~:text=According%20to%20statistics%2C%20there%20is,occupation%20underwater%20welding%20can%20be. [offshoreinjuryfirm.com]
🌻🌻 [google.com]
(Score: 2) by DeathMonkey on Tuesday June 21 2022, @06:48PM (1 child)
15% of what? Incident rates aren't usually communicated in percentages so I'm not even sure what they're referring to. I did see that link though when I was looking for the fatality rate.
(Score: 2) by JoeMerchant on Tuesday June 21 2022, @09:38PM
Birth 100% ends in death, so scope matters.
I think what the shock tactic is trying to say is: 15% of workers in deep ocean welding end up dying on the job - how many years do they do the job? Or, maybe they've stretched it to assume a 30 year career would end in death 15% of the time, one death per 180 man-years. Or maybe they pulled the number out their ass - I'd give better than 50% odds that's the case.
Anyway, commercial diver ranges from the guy who scrubs the bottom of my boat in the marina every 60-90 days up to these nutjobs who saturation dive ALONE for weeks on end at insane depths while they do underwater welding on seriously heavy stuff. If you focus on the nutjobs, the death rate is certain to rise.
🌻🌻 [google.com]
(Score: 0) by Anonymous Coward on Tuesday June 21 2022, @09:28PM
It's elevator mechanics, all the way down:
https://www.safetyandhealthmagazine.com/articles/18054-elevator-related-fatalities-in-construction-industry-increasing-cpwr [safetyandhealthmagazine.com]
I'm one of the lucky ones that got shafted and survived...
(Score: 0) by Anonymous Coward on Tuesday June 21 2022, @09:00AM
If a deep ocean construction welder were to drive to work, they'd still be more likely to die (drown) during their commute.
(Score: 0) by Anonymous Coward on Monday June 20 2022, @04:52PM
> ... because they are out there in a sea of human drivers with the same (thankfully rare, but not rare enough) bad actors that all drivers deal with putting them in the same unavoidable circumstances that human drivers find themselves in in most accidents.
I don't think you've been looking at the data? Yes, there are bad actors on the roads, but that's NOT "most accidents". Most accidents are single vehicle, here's a page from PA on that topic (which subtly looks at different slices of the data), but you won't have to look far to find similar data,
https://www.edgarsnyder.com/car-accident/types-of-accidents/single-vehicle/single-vehicle-statistics.html [edgarsnyder.com]
I've copied the text from that page --
(Score: 2, Informative) by steveg on Tuesday June 21 2022, @07:57PM
What are we talking about when we say "autopilot"? Maybe I should ask about what the original article is talking about when they say "autopilot".
What Tesla calls autopilot is the cruise control system. It's (slightly) more sophisticated than most cruise control systems, but not not hugely. It's basically an adaptive cruise control with lane-keeping. That's all it does, and it is enabled on every Tesla car.
It's NOT the same was what they call FSD (Full Self Drive) which is a *very* expensive add-on.
If the stats in the report are misreporting FSD as Autopilot, then the total number of cars (and miles driven) are at least an order of magnitude (or two) lower than the number of cars and miles that would be involved if they really mean Autopilot.