|
Condiv posted:he didn't base that 10x figure off this one death, he already came to it from other data He's comparing apples and oranges, and to be fair he admits that. We don't really know what percentage of disengagements would have turned into accidents. He thinks that 1 in 10 is a generous guess, and maybe it is, but it is just a number he made up. Not based on anything other than his hunch. The guy is tripping over himself to say "I told you so" before knowing whether or not Uber was even at fault.
|
# ? Mar 19, 2018 23:05 |
|
|
# ? Jun 13, 2024 01:11 |
|
Jose Valasquez posted:He's comparing apples and oranges, and to be fair he admits that. We don't really know what percentage of disengagements would have turned into accidents. He thinks that 1 in 10 is a generous guess, and maybe it is, but it is just a number he made up. Not based on anything other than his hunch. The guy is tripping over himself to say "I told you so" before knowing whether or not Uber was even at fault. in what way would you claim he's tripping over himself to claim uber was at fault? the post in the tweet was published a month before this also, would you argue that 1/10 disengagements turning into accidents is not being generous? remember, these are events are described as "a deactivation of the autonomous mode when a failure of the autonomous technology is detected or when the safe operation of the vehicle requires that the autonomous vehicle test driver disengage the autonomous mode and take immediate manual control of the vehicle." i'd argue that such events, in a system that should be autonomous and that people are saying could handle the road without a human driver are more likely than 1/10 to cause an accident. imo, it's pretty easy to draw a line to these events happening in a system meant for public use (not staffed with a professional test driver) and vehicular accidents.
|
# ? Mar 19, 2018 23:21 |
|
Condiv posted:in what way would you claim he's tripping over himself to claim uber was at fault? the post in the tweet was published a month before this quote:also, would you argue that 1/10 disengagements turning into accidents is not being generous? remember, these are events are described as "a deactivation of the autonomous mode when a failure of the autonomous technology is detected or when the safe operation of the vehicle requires that the autonomous vehicle test driver disengage the autonomous mode and take immediate manual control of the vehicle." i'd argue that such events, in a system that should be autonomous and that people are saying could handle the road without a human driver are more likely than 1/10 to cause an accident. imo, it's pretty easy to draw a line to these events happening in a system meant for public use (not staffed with a professional test driver) and vehicular accidents. The number of times that any given person does something dangerous while driving and doesn't cause an accident is way more than 10 times per accident. I see close calls literally every single time I drive but I see relatively few actual accidents on a daily basis.
|
# ? Mar 19, 2018 23:25 |
1/10 disengagements turning into accidents is at least an order of magnitude off, probably more. I'd guess it's more like 1/10k disengagements is an accident, but we don't have the data.
|
|
# ? Mar 19, 2018 23:33 |
|
quote:The number of times that any given person does something dangerous while driving and doesn't cause an accident is way more than 10 times per accident. I see close calls literally every single time I drive but I see relatively few actual accidents on a daily basis. that's not really comparable to a disengagement event though? how many times do you see people going unconscious behind the wheel and things turning out fine? when i think disengagement event i think things like that or https://www.youtube.com/watch?v=ZBaolsFyD9I not someone swerving in and out of traffic haphazardly. Jose Valasquez posted:By tweeting an "I told you so" tweet as soon as he found out about it. so, why are you so certain that uber was not at fault then? what exactly have you seen that points to that conclusion? them halting self-driving tests? the testimonials in thread about how pedestrians dart in and out of heavy traffic like squirrels? the reason he's claiming "i told you so" is because self driving cars have been getting into accidents at rates above human accident rates already, and now they've gotten someone killed way earlier than they statistically should've. Condiv fucked around with this message at 23:42 on Mar 19, 2018 |
# ? Mar 19, 2018 23:39 |
|
Condiv posted:that's not really comparable to a disengagement event though? how many times do you see people going unconscious behind the wheel and things turning out fine? when i think disengagement event i think things like that or quote:so, why are you so certain that uber was not at fault then? what exactly have you seen that points to that conclusion? them halting self-driving tests? the testimonials in thread about how pedestrians dart in and out of heavy traffic like squirrels? Of course they stopped self-driving tests, they are performing an investigation and if it turns out they were at fault it's going to look really bad if they ignored it and kept their cars out there. They would halt the tests during the investigation whether they are at fault or not. quote:the reason he's claiming "i told you so" is because self driving cars have been getting into accidents at rates above human accident rates already, and now they've gotten someone killed way earlier than they statistically should've. Jose Valasquez fucked around with this message at 23:56 on Mar 19, 2018 |
# ? Mar 19, 2018 23:53 |
|
Jose Valasquez posted:Except he doesn't know that. He is basing this off the same articles everyone else is reading and there is not enough public information to definitely say that the self driving car was at fault (yet) Actually, there is! http://abc30.com/3234630/ quote:The vehicle was going 40 mph when it hit the pedestrian. There was no indication the car attempted to slow itself before the collision.
|
# ? Mar 20, 2018 00:11 |
|
Tempe police said the car didn’t try to slow down.
|
# ? Mar 20, 2018 00:14 |
|
Condiv posted:Actually, there is! this doesn't mean what you think it means but you're going to argue about it anyway no indication doesn't mean "crashed into an object with plenty of time to stop" like you are asserting it does here, she could have walked in front of the vehicle so suddenly it had no time to stop and i know this is where you will say something goofy about reaction times or w/e and you're wrong about that too
|
# ? Mar 20, 2018 00:14 |
|
boner confessor posted:this doesn't mean what you think it means but you're going to argue about it anyway it means literally what it says boner confessor. no attempt to stop. not "didn't manage to stop in time". not "had no time to stop". literally "There was no indication the car attempted to slow itself before the collision" if the car was functioning properly, there would have been an attempt to stop. it did not. hence, it is at fault.
|
# ? Mar 20, 2018 00:19 |
|
Condiv posted:it means literally what it says boner confessor. no attempt to stop. not "didn't manage to stop in time". not "had no time to stop". literally "There was no indication the car attempted to slow itself before the collision" yeah this just means there's a lack of skidmarks on the road and nothing in the police report saying the driver hit the brakes. you're reading this completely incorrectly as "the uncaring machine plowed directly into the woman as she plead for her life" but that is because of the position you're trying to push here. you're bending the facts in a pretty silly way and it's sort of funny i guess to watch you tantrum about this. i do enjoy watching people continually demonstrate how much they don't understand what they're trying to argue but, eh, 4/10
|
# ? Mar 20, 2018 00:20 |
|
boner confessor posted:yeah this just means there's a lack of skidmarks on the road and nothing in the police report saying the driver hit the brakes. you're reading this completely incorrectly as "the uncaring machine plowed directly into the woman as she plead for her life" but that is because of the position you're trying to push here. you're bending the facts in a pretty silly way and it's sort of funny i guess to watch you tantrum about this i'm not bending the facts at all. if the machine were working correctly there would have been an attempt to stop. that it hit her going 40mph indicates it was malfunctioning
|
# ? Mar 20, 2018 00:22 |
|
Condiv posted:i'm not bending the facts at all. if the machine were working correctly there would have been an attempt to stop. that it hit her going 40mph indicates it was malfunctioning. you can stop trying to stick up for the robot car now haha ok it's pretty good you're arguing here about an unthinking tool not knowing when to stop lol like, the irony, etc.
|
# ? Mar 20, 2018 00:24 |
|
Condiv posted:i'm not bending the facts at all. if the machine were working correctly there would have been an attempt to stop. that it hit her going 40mph indicates it was malfunctioning Was the human in the seat also malfunctioning?
|
# ? Mar 20, 2018 00:24 |
|
Owlofcreamcheese posted:Was the human also malfunctioning? in all likelihood, he was not paying enough attention. i guess you could say yes he was in that case. but people have warned that autonomous systems tend to lead to such outcomes as they lull people into a false sense of security boner confessor posted:haha ok it's pretty good you're arguing here about an unthinking tool not knowing when to stop lol why are you melting down so hard about this?
|
# ? Mar 20, 2018 00:26 |
|
I think this will be one of many cases that proves even if Uber is at fault, we as a society don’t care about human life enough for this to be a meaningful block to implementation.
|
# ? Mar 20, 2018 00:27 |
|
Trabisnikof posted:I think this will be one of many cases that proves even if Uber is at fault, we as a society don’t care about human life enough for this to be a meaningful block to implementation. the government can and will get involved if self driving car technology isn't "safe enough", they already regulate emissions and other safety features as well and if there's a hint the technology might be defective or "defective" then the NHSTA can and will step in with a whole new pile of regulations. im guessing they haven't yet because the technology is still infant
|
# ? Mar 20, 2018 00:29 |
|
boner confessor posted:the government can and will get involved if self driving car technology isn't "safe enough", they already regulate emissions and other safety features as well and if there's a hint the technology might be defective or "defective" then the NHSTA can and will step in with a whole new pile of regulations. im guessing they haven't yet because the technology is still infant i wouldn't count on it with president trump, but at least the NHTSA is looking into it
|
# ? Mar 20, 2018 00:30 |
|
boner confessor posted:the government can and will get involved if self driving car technology isn't "safe enough", they already regulate emissions and other safety features as well and if there's a hint the technology might be defective or "defective" then the NHSTA can and will step in with a whole new pile of regulations. im guessing they haven't yet because the technology is still infant I think history is on my side on this one. When automobiles first caused massive pedestrian deaths we responded by criminalizing pedestrian behavior that didn’t comply with the needs of automobiles. Of course they are and will continue to be regulated, but when push comes to shove I think as a society we’d select more pedestrian deaths if it still reduced labor costs for businesses.
|
# ? Mar 20, 2018 00:35 |
|
Condiv posted:in all likelihood, he was not paying enough attention. i guess you could say yes he was in that case. but people have warned that autonomous systems tend to lead to such outcomes as they lull people into a false sense of security Or maybe the fact both the car and the testing engineer failed to avoid the woman is some indication that it wasn't a simple situation that is the one weird trick that proves self driving cars are fake and dumb.
|
# ? Mar 20, 2018 00:40 |
|
second source on the "car did not attempt to stop" claim: https://twitter.com/ceodonovan/status/975857228442775552 Owlofcreamcheese posted:Or maybe the fact both the car and the testing engineer failed to avoid the woman is some indication that it wasn't a simple situation that is the one weird trick that proves self driving cars are fake and dumb. so you think she teleported directly in front of the car and that's why it didn't slow down? cause if the car was not malfunctioning, it would have slowed at least somewhat in all realistic scenarios. but it did not. sorry, this was a malfunction, not an entirely unavoidable accident. also, i'm sure prosecutors are looking at possible charges because there was no fault on uber or their autonomous vehicle's part in this case oocc Condiv fucked around with this message at 00:45 on Mar 20, 2018 |
# ? Mar 20, 2018 00:40 |
|
Condiv posted:so you think she teleported directly in front of the car and that's why it didn't slow down? cause if the car was not malfunctioning, it would have slowed at least somewhat in all realistic scenarios. but it did not. sorry, this was a malfunction, not an entirely unavoidable accident. Does someone need to post some 4chan style video compilation of hundreds and hundreds of clips of people stepping into traffic and getting immediately hit by a car? Because you seem to be acting really skeptical that that happens.
|
# ? Mar 20, 2018 00:47 |
|
Owlofcreamcheese posted:Does someone need to post some 4chan style video compilation of hundreds and hundreds of clips of people stepping into traffic and getting immediately hit by a car? Because you seem to be acting really skeptical that that happens. self driving cars have many multitudes higher reaction times than humans oocc. if it were not malfunctioning, stepping in front of it would trigger it attempting to slow down. it did not attempt to slow down. that is extremely indicative of malfunction of the autonomous system.
|
# ? Mar 20, 2018 00:52 |
|
Trabisnikof posted:I think history is on my side on this one. When automobiles first caused massive pedestrian deaths we responded by criminalizing pedestrian behavior that didn’t comply with the needs of automobiles. oh i'm not disputing you there. i'm just saying that self driving tech clearly falls under the jurisdiction of NHSTA and im wondering what they've done so far, and what they will do when self driving cars are in wider usage and killing more people Condiv posted:self driving cars have many multitudes higher reaction times than humans oocc. if it were not malfunctioning, stepping in front of it would trigger it attempting to slow down. it did not attempt to slow down. that is extremely indicative of malfunction of the autonomous system. it is not at all lol boner confessor posted:this doesn't mean what you think it means but you're going to argue about it anyway
|
# ? Mar 20, 2018 00:57 |
|
Condiv posted:self driving cars have many multitudes higher reaction times than humans oocc. if it were not malfunctioning, stepping in front of it would trigger it attempting to slow down. it did not attempt to slow down. that is extremely indicative of malfunction of the autonomous system. Or an indication that they stepped out so very close to the car, so close that neither the human or the driver could act. Instead of some weird theory that the human driver sat and watched the car plow down the woman but the car had mind controlled him into complacency and not caring he was going to kill someone.
|
# ? Mar 20, 2018 00:59 |
|
Condiv posted:self driving cars have many multitudes higher reaction times than humans oocc. if it were not malfunctioning, stepping in front of it would trigger it attempting to slow down. it did not attempt to slow down. that is extremely indicative of malfunction of the autonomous system. Autonomous cars don't have instant reaction times. The best estimate I found was about 0.5 seconds. That gives about 30ft at 40mph. It is not unreasonable that someone could step out in front of a car that is 30ft away if they aren't paying attention. I know that if the investigation comes back and Uber is in fact at fault you're going to be smug and think you won this argument, but nobody is arguing that Uber is not at fault, they are arguing that there isn't enough public information to definitely say that Uber is at fault. You are wrong even if Uber is at fault.
|
# ? Mar 20, 2018 01:06 |
|
Personally, I'm sad that Uber's lovely tech murdered someone
|
# ? Mar 20, 2018 01:07 |
|
Owlofcreamcheese posted:Or an indication that they stepped out so very close to the car, so close that neither the human or the driver could act. Instead of some weird theory that the human driver sat and watched the car plow down the woman but the car had mind controlled him into complacency and not caring he was going to kill someone. oocc, there have been studies done that have shown that autonomous vehicles with disengagement systems tend to worsen reaction times in humans significantly, cause they stop paying attention to them. i invite you to read up as for your theory that the woman darted out in front of the car faster than it could react, well, that's not based in anything realistic. the fact of the matter is that if the car was functioning properly, it would have begun braking as soon as an obstacle was detected in its path. it detected no obstacle, even though there was one there. that is indicated by it hitting her at 40mph. considering that the latest sensors in autonomous vehicles can image at 120fps, they have a maximum reaction time of 8ms. even if the vehicle's reaction time was 16ms, that's far faster than human movement can put itself into the path of the car. basically, there's no way it could have hit her like it did cause of her actions. that puts the machine at fault.
|
# ? Mar 20, 2018 01:08 |
|
Jose Valasquez posted:Autonomous cars don't have instant reaction times. The best estimate I found was about 0.5 seconds. That gives about 30ft at 40mph. It is not unreasonable that someone could step out in front of a car that is 30ft away if they aren't paying attention. 0.5 seconds would put autonomous cars at below human reaction times. where are you getting that number? 500 milliseconds is an eternity computationally. 16ms or so sounds much more likely human reaction time averages at around .25 seconds btw
|
# ? Mar 20, 2018 01:10 |
|
Condiv posted:as for your theory that the woman darted out in front of the car faster than it could react, well, that's not based in anything realistic. i can only imagine you've never driven a car around pedestrians or something
|
# ? Mar 20, 2018 01:10 |
|
So so far we have the Tempe police saying there's no indication the vehicle attempted to slow down, probably looking for skid marks and the like. Given that this car was instrumented out the wazoo, I'm sure we'll have a lot more information to argue about later about whether the car should have seen the pedestrian. Seems way early to be saying what the car should or should not have done.
|
# ? Mar 20, 2018 01:10 |
|
boner confessor posted:i can only imagine you've never driven a car around pedestrians or something are pedestrians superhuman now? i'd like to see one that can dart into the path of a car in less than 100ms.
|
# ? Mar 20, 2018 01:12 |
|
fordan posted:Seems way early to be saying what the car should or should not have done. absolutely, unless you're hell bent on arguing that self driving cars are bad using the worst possible arguments i'm like the designated "i hate self driving cars" guy itt and even i think uber probably isnt at fault here. pedestrians get themselves hit by cars in dumb ways all of the time, especially at night i'll post this again and quote it for those choosing not to click the link https://www.npr.org/2017/03/30/522085503/2016-saw-a-record-increase-in-pedestrian-deaths quote:According to the GHSA report, 74 percent of pedestrian fatalities happen at night, and 72 percent of those killed were not crossing at intersections. both of which are true in this case
|
# ? Mar 20, 2018 01:12 |
|
boner confessor posted:absolutely, unless you're hell bent on arguing that self driving cars are bad using the worst possible arguments you're trying real hard to blame a woman walking her bike across the street for her own death. the fact of the matter is she was not moving fast enough that the car couldn't react to her if it wasn't malfunctioning. and it's almost certain that no human could move fast enough to produce the result we saw here in a non-failure state of the automated system.
|
# ? Mar 20, 2018 01:17 |
|
fordan posted:So so far we have the Tempe police saying there's no indication the vehicle attempted to slow down, probably looking for skid marks and the like. Given that this car was instrumented out the wazoo, I'm sure we'll have a lot more information to argue about later about whether the car should have seen the pedestrian. Seems way early to be saying what the car should or should not have done. i have absolute faith that Uber, a company which spent time and money developing a new, better way to destroy subpoenaed evidence, will give all of the info.
|
# ? Mar 20, 2018 01:19 |
|
Condiv posted:0.5 seconds would put autonomous cars at below human reaction times. where are you getting that number? Human reaction times while driving are not anywhere close to 0.25 seconds while driving. http://www.visualexpert.com/Resources/reactiontime.html posted:Expected: the driver is alert and aware of the good possibility that braking will be necessary. This is the absolute best reaction time possible. The best estimate is 0.7 second. Of this, 0.5 is perception and 0.2 is movement, the time required to release the accelerator and to depress the brake pedal. This is the only reference I've found for self driving car reaction times https://www.wired.com/story/self-driving-cars-perception-humans/ posted:Machines can react faster than humans, in about 0.5 seconds on a dry road compared to 1.6 seconds for the meatbags.
|
# ? Mar 20, 2018 01:22 |
|
It's hilarious that Condiv has successfully driven SV-skeptics into defending Uber and self-driving cars because he has to dig his heels in with a dumb argument about reaction times.
|
# ? Mar 20, 2018 01:24 |
|
Slanderer posted:i have absolute faith that Uber, a company which spent time and money developing a new, better way to destroy subpoenaed evidence, will give all of the info. This is a very valid point, although I suspect that this happening well away from Uber HQ and that the police are talking potential criminal charges might help keep the data safe, especially the video data which they'd have a hard time arguing doesn't exist. Also kinda hoping the police just took the car as evidence.
|
# ? Mar 20, 2018 01:30 |
|
Jose Valasquez posted:Human reaction times while driving are not anywhere close to 0.25 seconds while driving. even in your article, the human reaction time while driving is 500 milliseconds. we are not concerned with human muscle-reaction in this discussion, merely recognition of stimulus. the idea that an autonomous vehicle is about as slow at signal processing as a human is ridiculous. according to nvidia, they can do the processing at 30fps. that's 33ms reaction time, or able to react to something appearing in its sensors within 2 feet at the speed this car was going when it hit the pedestrian. https://devblogs.nvidia.com/deep-learning-self-driving-cars/ and their self-driving car is probably not as far along in development as uber's was (also, this tech is approximately 2 years old now!) as i said before, if the car was not malfunctioning, it would have attempted to stop. a human walking a bike across a street cannot suddenly appear with no time for the car to react in a non-failure state. it's just not possible for humans to move that quick in 33ms Condiv fucked around with this message at 01:39 on Mar 20, 2018 |
# ? Mar 20, 2018 01:31 |
|
|
# ? Jun 13, 2024 01:11 |
|
Police Say Uber Is Likely Not at Fault for Its Self-Driving Car Fatality in Arizona http://fortune.com/2018/03/19/uber-self-driving-car-crash/ https://www.sfchronicle.com/busines...m_medium=social
|
# ? Mar 20, 2018 02:14 |