|
Punkin Spunkin posted:So you're saying it'll be like now but safer? A person distracted by a phone can't be better than a car computer self driving cars won't be 100% fully autonomous for a long time, but before that point people are going to think "oh i can text now while the computer drives" and it will still cause accidents. this is because people will think of self-driving cars as a safety feature when really it's a feature of convenience, like cruise control
|
#
?
Jan 19, 2017 20:49
|
|
|
|
| # ? Jun 7, 2024 17:45 |
|
|
boner confessor posted:self driving cars won't be 100% fully autonomous for a long time Doesn't Tesla have a working level 5 system that they're hoping to roll out this year? How else do you define fully autonomous? edit: i misread a press release, apparently they're aiming for level 4 in q4 this year. Still, that's not "a long time". Dr. Fishopolis fucked around with this message at 20:58 on Jan 19, 2017 |
|
#
?
Jan 19, 2017 20:55
|
|
|
Dr. Fishopolis posted:Doesn't Tesla have a working level 5 system that they're hoping to roll out this year? How else do you define fully autonomous? fully autonomous (pay attention to the road because the car may kick control back over to you with short notice and we have no liability if you go off a cliff) there really aren't that many fully autonomous vehicles - even ones which nominally drive themselves usually have a human backup in the cabin for safety reasons. subway cars, farm tractors, even airliners. cars won't be any different, except there will be a more prevalent and dangerous assumption that you can watch tv or whatever when you should be (according to the terms of your insurance) keeping an eye on the road. this is what caused the fatal tesla crash, the driver was too distracted to notice an obvious hazard that the automation didn't detect boner confessor fucked around with this message at 20:59 on Jan 19, 2017 |
|
#
?
Jan 19, 2017 20:57
|
|
|
boner confessor posted:fully autonomous* That's not what level 4 or 5 autonomy means.
|
|
#
?
Jan 19, 2017 20:59
|
|
|
tesla can put whatever they want in a press release. i'll believe it when i see it in use, not killing people
|
|
#
?
Jan 19, 2017 20:59
|
|
|
Dr. Fishopolis posted:That's not what level 4 or 5 autonomy means. Do you want to provide a link to these definitions ?
|
|
#
?
Jan 19, 2017 21:00
|
|
|
poemdexter posted:Coworker and I won a hackathon building an IoT coffeepot that shames you on twitter for taking last cup and not making a new pot. We were then invited to Vegas to demo during AT&T keynote and won 2nd. We split 17k in prizes after the whole ordeal. IoT is stupid easy to get into considering EVERYONE has an API to do something. Everyone kept asking us what we were going to do after going forward, but we just laughed and said it was gonna sit on a table at work and we'd chuckle from time to time about it. twitter (and all social) integration was stupidly hot for a while, thank god that it's died back and people are no longer concerned with making their electric toothbrush post to facebook
|
|
#
?
Jan 19, 2017 21:01
|
|
|
Baby Babbeh posted:If you REALLY want to know what investor denial looks like, here's Tim Draper, who has invented a conspiracy to avoid admitting his investment in Theranos failed because he didn't bother to vet the science: I mean, if you want to send me your blood I'll do 100 tests on it for £10. You want to know your blood's IQ and Kinsey rating, don't you?
|
|
#
?
Jan 19, 2017 21:03
|
|
|
I don't disagree that it's fair to be skeptical about the whole thing, I'm just saying that full autonomy is way more likely to be 2 or 3 years away than 15. It's not just conceptual poo poo, there are multiple manufacturers with public prototypes racing to be the first to market with this right now.jre posted:Do you want to provide a link to these definitions ? http://www.techrepublic.com/article/autonomous-driving-levels-0-to-5-understanding-the-differences/
|
|
#
?
Jan 19, 2017 21:04
|
|
|
Dr. Fishopolis posted:I don't disagree that it's fair to be skeptical about the whole thing, I'm just saying that full autonomy is way more likely to be 2 or 3 years away than 15. It's not just conceptual poo poo, there are multiple manufacturers with public prototypes racing to be the first to market with this right now. bleeding edge technology always, always runs up against the social and regulatory use of that technology. tesla could unveil a fully automated car tomorrow and it's still going to be a decade or more before it's widely in use because of the ways people buy and use vehicles, as well as government restrictions on the use of vehicles, also the auto insurance market figuring out how to insure self driving cars etc. like there could be self drivers on the road tomorrow and the day after that the government's going to formulate rules about how exactly self driving they can be for example, the contraceptive pill was invented in 1957. it was only available by perscription at first to women with "severe menstrual trouble", then to married couples, then to single women with a good reason and a doctor's note, etc. because the technology had to wait for society to accept its use
|
|
#
?
Jan 19, 2017 21:07
|
|
|
boner confessor posted:this is what caused the fatal tesla crash, the driver was too distracted to notice an obvious hazard that the automation didn't detect while that's true, the NHTSA investigation determined that the autosteer software reduced crash rate by 40 percent over the entire deployment of model S and X cars. I mean, sucks for that guy but it's hard to argue that there isn't a net benefit to even partial automation.
|
|
#
?
Jan 19, 2017 21:10
|
|
|
quote:http://www.techrepublic.com/article/autonomous-driving-levels-0-to-5-understanding-the-differences/ Dr. Fishopolis posted:It's not just conceptual poo poo, there are multiple manufacturers with public prototypes racing to be the first to market with this right now. Yeah, but none of them actually work, and all the manufacturers are really really against having to share data with the regulator about how often a human had to intervene to stop the prototypes crashing.
|
|
#
?
Jan 19, 2017 21:13
|
|
|
boner confessor posted:for example, the contraceptive pill was invented in 1957. it was only available by perscription at first to women with "severe menstrual trouble", then to married couples, then to single women with a good reason and a doctor's note, etc. because the technology had to wait for society to accept its use There's no deep-seated puritan taboo against discussing automated cars in American society, so I'm not sure that analogy holds water. You might be right though, I don' t think either of us can look into a crystal ball and say for certain how long it'll take for this stuff to integrate into society. I do think there's a public excitement and willingness to adopt new technology though, and combined with the obvious safety benefits I hope there wouldn't be too much NIMBY foot dragging. But also, Trump is president so who the gently caress knows.
|
|
#
?
Jan 19, 2017 21:15
|
|
|
jre posted:Yeah, but none of them actually work, and all the manufacturers are really really against having to share data with the regulator about how often a human had to intervene to stop the prototypes crashing. I guess the first self-driving car to market will have to legally be labelled "For entertainment purposes only; cannot actually self-drive."
|
|
#
?
Jan 19, 2017 21:16
|
|
|
Dr. Fishopolis posted:while that's true, the NHTSA investigation determined that the autosteer software reduced crash rate by 40 percent over the entire deployment of model S and X cars. I mean, sucks for that guy but it's hard to argue that there isn't a net benefit to even partial automation. i didn't argue there isn't a net benefit? what i argued is that this technology - which is not really a technology for safety as it is for convenience - will be viewed as being safer, and thus make drivers complacent and less safe. some drivers will use the technology properly, and will be paying attention to the road while letting the car drive itself. some drivers will be completely distracted by porn, and thus be less safe than they would be manually operating a car (but more safe than if they were watching porn while operating a manual car). and until self driving cars are more robust at handling hazards and alerting the driver, which i doubt will be in the first few generations of the technology, then there's going to be a tradeoff where drivers overly rely on the convenient self-driving feature and treat it like a safety catch when really it's a safety enhancement that still demands your attention
|
|
#
?
Jan 19, 2017 21:17
|
|
|
Dr. Fishopolis posted:There's no deep-seated puritan taboo against discussing automated cars in American society, so I'm not sure that analogy holds water. You might be right though, I don' t think either of us can look into a crystal ball and say for certain how long it'll take for this stuff to integrate into society. I do think there's a public excitement and willingness to adopt new technology though, and combined with the obvious safety benefits I hope there wouldn't be too much NIMBY foot dragging. But also, Trump is president so who the gently caress knows. What obvious safety benefit ?
|
|
#
?
Jan 19, 2017 21:17
|
|
|
jre posted:What obvious safety benefit ? if you allow a self driving car to drive, and you pay attention to fill in the gaps, then you're safer probably if you allow a self driving car to drive, and you're catching up on anime reruns, then i argue you are not safer given that like i dunno, 10%-20% of people are currently habitual distracted drivers, i fear how much this group will increase once they have cars that "drive themselves"
|
|
#
?
Jan 19, 2017 21:21
|
|
|
boner confessor posted:i didn't argue there isn't a net benefit? what i argued is that this technology - which is not really a technology for safety as it is for convenience - will be viewed as being safer, and thus make drivers complacent and less safe. some drivers will use the technology properly, and will be paying attention to the road while letting the car drive itself. some drivers will be completely distracted by porn, and thus be less safe than they would be manually operating a car (but more safe than if they were watching porn while operating a manual car). and until self driving cars are more robust at handling hazards and alerting the driver, which i doubt will be in the first few generations of the technology, then there's going to be a tradeoff where drivers overly rely on the convenient self-driving feature and treat it like a safety catch when really it's a safety enhancement that still demands your attention If any of that were true, wouldn't Tesla drivers be careening all over the place and killing busloads of nuns by now? I mean, that was probably a valid concern a year ago but this poo poo has been available to consumers for quite some time now and nothing like what you're describing has happened. In fact, the only evidence I can find shows the opposite effect, like the investigation I linked earlier, and the Virginia Tech study from a year ago. I don't disagree what you're saying makes intuitive sense, but the technology is here and it's just not playing out that way.
|
|
#
?
Jan 19, 2017 21:25
|
|
|
Boner confessor ran away from the self driving car thread to start up his ranting about the evils of self driving cars in this thread. (USER WAS PUT ON PROBATION FOR THIS POST)
|
|
#
?
Jan 19, 2017 21:25
|
|
|
Dr. Fishopolis posted:If any of that were true, wouldn't Tesla drivers be careening all over the place and killing busloads of nuns by now? I mean, that was probably a valid concern a year ago but this poo poo has been available to consumers for quite some time now and nothing like what you're describing has happened. In fact, the only evidence I can find shows the opposite effect, like the investigation I linked earlier, and the Virginia Tech study from a year ago. do you have some kind of special insight into how often tesla drivers using prototype technology are ignoring safety protocols or something? because as far as i know they haven't made any of that data public. we can assume that the people testing this technology are paying closer attention to the terms of use than would otherwise be typical, i think we do know the one guy killed in a self driving tesla wasn't paying attention, so
|
|
#
?
Jan 19, 2017 21:27
|
|
|
NHTSA’s full final investigation into Tesla’s Autopilot shows 40% crash rate reduction https://techcrunch.com/2017/01/19/nhtsas-full-final-investigation-into-teslas-autopilot-shows-40-crash-rate-reduction/?ncid=rss
|
|
#
?
Jan 19, 2017 21:29
|
|
|
boner confessor posted:we do know the one guy killed in a self driving tesla wasn't paying attention, so Please. It's a Tesla marketed as self-driving, it's not actually self-driving but many of the people buying them don't realize that (and that's intentional).
|
|
#
?
Jan 19, 2017 21:35
|
|
|
Holy poo poo take it to the terrible self-driving car thread. What is it about this topic that makes people so obtuse.
|
|
#
?
Jan 19, 2017 21:40
|
|
|
Dr. Fishopolis posted:If any of that were true, wouldn't Tesla drivers be careening all over the place and killing busloads of nuns by now? The Tesla accident/mile comparisons always bug me because it seems like they're cherrypicking sunny highway miles against the national average that includes situations the system wouldn't even attempt like "light rain."
|
|
#
?
Jan 19, 2017 21:42
|
|
|
Owlofcreamcheese posted:NHTSA’s full final investigation into Tesla’s Autopilot shows 40% crash rate reduction Based entirely on the data supplied to them by tesla?
|
|
#
?
Jan 19, 2017 21:51
|
|
|
JawnV6 posted:The argument is that as the systems get better more drivers will ignore the road conditions and be unable to reassert control. There's a regression in performance at the next step, not necessarily present in the current systems that require driver attention now. We can ignore the data that suggests they're safer than regular cars, because once they get safer, they'll get more dangerous, because people will feel safer. Got it. edit: this is getting really dumb, please let's go to the already bad thread for this stupid argument
|
|
#
?
Jan 19, 2017 21:51
|
|
|
Dr. Fishopolis posted:We can ignore the data that suggests they're safer than regular cars, because once they get safer, they'll get more dangerous, because people will feel safer. Got it. without knowing the parameters of the study you can't conclude that the technology is just flat out safer. like, it may be safer when people are paying attention which is a big problem for a technology which will absolutely cause people to pay less attention. adding safety features to a device doesn't intrinsically make a device safer, especially not if the safety feature itself encourages people to use it in an unsafe way. like the safety catch on a gun, it helps if you still practice proper gun discipline but if you act like a fool with the gun because the safety is on then it's only a matter of time before you shoot someone
|
|
#
?
Jan 19, 2017 21:55
|
|
|
JawnV6 posted:The Tesla accident/mile comparisons always bug me because it seems like they're cherrypicking sunny highway miles against the national average that includes situations the system wouldn't even attempt like "light rain." Are you sure about that? A quick glance through the NHSTA report says they're comparing pre-autosteer and post-autosteer Tesla accident rates. I'm not sure if they cover the exact same distribution of season-miles, but at a first glance it seems to be a good comparison. That said, due to the nature of the comparison, it's not clear if the autosteer is responsible for the drop/what fraction of the drop it's responsible for.
|
|
#
?
Jan 19, 2017 21:57
|
|
|
moebius2778 posted:Are you sure about that? A quick glance through the NHSTA report says they're comparing pre-autosteer and post-autosteer Tesla accident rates. Dr. Fishopolis posted:We can ignore the data that suggests they're safer than regular cars, because once they get safer, they'll get more dangerous, because people will feel safer. Got it. The NHSTA praises Tesla's efforts to research this specific area, I can't fathom why you'd attempt to downplay and ignore it.
|
|
#
?
Jan 19, 2017 23:18
|
|
|
Mozi posted:Holy poo poo take it to the terrible self-driving car thread. What is it about this topic that makes people so obtuse. I want to be able to yank it, while drunk, while going from place to place at high speed in my own vehicle, and it infuriates me when people suggest that might not be possible in the imminent future!
|
|
#
?
Jan 19, 2017 23:18
|
|
|
edit: Didn't see the next page of replies. Never mind.
|
|
#
?
Jan 19, 2017 23:43
|
|
|
Market challenges are almost always more difficult than technology challenges when it comes to products that are sufficiently advanced. Even stuff that looks like it just stepped in and changed everything overnight generally has a long history of marginally better antecedent products that get the market closer and closer to the point at which the revolution could happen. Stuff changes slowly, then all at once, and there's no reason to think self-driving cars won't be the same. We're in the "stuff changes slowly" part of the adoption curve right now. For something that's as big a change as this, it could last a decade or more.
|
|
#
?
Jan 20, 2017 00:10
|
|
|
Baby Babbeh posted:Stuff changes slowly, then all at once, and there's no reason to think self-driving cars won't be the same. We're in the "stuff changes slowly" part of the adoption curve right now. For something that's as big a change as this, it could last a decade or more. i totally agree, but i think the timescales are pretty optimistic, and you can break the promises into two chunks self driving cars will be safer and more convenient - yeah, probably, over the next 10-20 years self driving cars will change society - not likely, they're still just cars, and if at all on like a 40 year timespan if not longer
|
|
#
?
Jan 20, 2017 00:12
|
|
|
Worth remembering that the average car on the road right now is 12 years old. Even if every car that was sold from tomorrow onward was fully autonomous, it would be likely to take at least 12 years for normal processes of car buying and junking to get us to a majority of cars on the road having the functionality.
|
|
#
?
Jan 20, 2017 00:19
|
|
|
Baby Babbeh posted:Market challenges are almost always more difficult than technology challenges when it comes to products that are sufficiently advanced. Even stuff that looks like it just stepped in and changed everything overnight generally has a long history of marginally better antecedent products that get the market closer and closer to the point at which the revolution could happen. Yeah, but that slow change started many years ago. Even some of the cheapest cars you can buy right this second have features like "lane keep" that will steer you in a lane and dynamic cruise control that will adjust your speed in relation to cars around you. It's driving assistance features that aren't intelligent or complicated at all but are the precedence you are talking about. Same with all the cars that sound alarms or flash lights if you are going to back up into something or if you are turning into a car in your blind spot or all the other super boring features that already exist that will just get rolled into the next step of "alarm sounds" into "it just does it for you". It's not the future where this stuff will exist, this is basic features on modern cars right now.
|
|
#
?
Jan 20, 2017 00:24
|
|
poemdexter posted:Coworker and I won a hackathon building an IoT coffeepot that shames you on twitter for taking last cup and not making a new pot. We were then invited to Vegas to demo during AT&T keynote and won 2nd. We split 17k in prizes after the whole ordeal. IoT is stupid easy to get into considering EVERYONE has an API to do something. Everyone kept asking us what we were going to do after going forward, but we just laughed and said it was gonna sit on a table at work and we'd chuckle from time to time about it. Not to derail this wonderful new conversation or anything, but would you be interested in my life savings?
|
|
|
#
?
Jan 20, 2017 01:19
|
|
|
JawnV6 posted:Be as blithely reductive as you want, boner confessor is referring to a specific failure mode that would not be present in the current data, hence the lack of bus-o-nun killings you bemoaned. Once the computer system is safer, the human will stop paying the requisite attention and violate the system's assumption that they can reassert control to avoid collision. It's happened multiple times in aviation: Aeroflot 593, Asiana 214, Eastern Air Lines 401. I'm not saying the systems should be halted, data should be ignored, or whatever other assignations you're tossing at me, simply noting that it's a genuine issue in all autopilot systems. To be clear, I was being snarky because boner confessor's position seems to be "there's this problem that invalidates the entire idea of automated cars and nobody is working on it and it's impossible to solve anyway, so everyone is going to crash and die". Sorry if you got caught in the crossfire. Yeah, of course it's a real problem. Like you say, it's inherent to the evolution of all autopilot systems, and I have a hard time believing that the tens of thousands of people spending billions of dollars to make automated cars a reality haven't considered it, or would implement a solution with such a predictable flaw.
|
|
#
?
Jan 20, 2017 01:58
|
|
|
Dr. Fishopolis posted:To be clear, I was being snarky because boner confessor's position seems to be "there's this problem that invalidates the entire idea of automated cars and nobody is working on it and it's impossible to solve anyway, so everyone is going to crash and die". Sorry if you got caught in the crossfire. i get that a lot, i dunno why when i point out problems with things that people think i'm trashing that thing and insulting everyone who likes it imo there is way too much hype about automated cars, which, if you just sort of vaguely glance at retrofuturism, is like the #1 thing people have gotten super hyped up over since like the 1930s - flying cars, robot cars, atomic cars, etc. cars are a huge and critical part of the north american lifestyle and they centrally feature in our visions of the better world ahead, so if you diss cars you kind of diss the future at the same time and some folks dont like that Dr. Fishopolis posted:Yeah, of course it's a real problem. Like you say, it's inherent to the evolution of all autopilot systems, and I have a hard time believing that the tens of thousands of people spending billions of dollars to make automated cars a reality haven't considered it, or would implement a solution with such a predictable flaw. there's a long history of people rushing a technology to market without regarding predictable safety flaws, or ignoring safety flaws. especially when it comes to automobiles, where automakers are willing to tolerate casualties https://en.wikipedia.org/wiki/General_Motors_ignition_switch_recalls boner confessor fucked around with this message at 02:18 on Jan 20, 2017 |
|
#
?
Jan 20, 2017 02:15
|
|
|
boner confessor posted:i get that a lot, i dunno why when i point out problems with things that people think i'm trashing that thing and insulting everyone who likes it Yeah, self-driving cars were prominently featured at General Motors' Futurama exhibit of a future city, in the 1939-1940 New York World's Fair. At that time GM was saying the cars would be widespread in just 20 years' time, 1960 or so.
|
|
#
?
Jan 20, 2017 03:04
|
|
|
|
| # ? Jun 7, 2024 17:45 |
|
|
boner confessor posted:imo there is way too much hype about automated cars, which, if you just sort of vaguely glance at retrofuturism, is like the #1 thing people have gotten super hyped up over since like the 1930s - flying cars, robot cars, atomic cars, etc. cars are a huge and critical part of the north american lifestyle and they centrally feature in our visions of the better world ahead, so if you diss cars you kind of diss the future at the same time and some folks dont like that For the record, I don't own or want to own a car. If it were up to me, we would have invested in federal public transportation system eons ago, but that is never, ever going to happen. A massive, distributed automated car system is the next best thing. I actually want car ownership to become a niche hobby, and for everyone to have cheap, reliable, safe access to transportation. I think automated vehicles can help that happen. boner confessor posted:there's a long history of people rushing a technology to market without regarding predictable safety flaws, or ignoring safety flaws. especially when it comes to automobiles, where automakers are willing to tolerate casualties I don't disagree, I just doubt the kind of problems we're likely to see are obvious. I'm way more concerned about software security than driver interaction problems. Dr. Fishopolis fucked around with this message at 03:28 on Jan 20, 2017 |
|
#
?
Jan 20, 2017 03:25
|
|