The case is believed to be the first time that U.S. prosecutors have brought felony charges against a motorist who was using a partially automated driving system.
I have had the adaptive cruise control brake on multiple Hondas and Subarus in similar situations. Not like slamming on the brakes, but firm enough to confuse the hell out of me.
Every time it was confusing and now I just don’t use it if the road is anything but open and clear.
Honda’s sensing system will read shadows from bridges as obstructions in the road that it needs to brake for. It’s easy enough to accelerate out of the slowdown, but I was surprised to find that there is apparently no radar check to see if the obstruction is real.
My current vehicle doesn’t have that issue, so either the programming has been improved or the vendor for the sensing systems is a different one (different vehicle make, so it’s entirely possible).
i barely trust the lane-keeping assistant in my friend’s car. imagine going 70+km/h and suddenly the car decides to jerk the steering to the left/right because you weren’t exactly in the middle of your lane.
fuck modern assistants IMO. i can use the steering wheel just fine, and people have been able to for a hundred years.
Considering that driving is (statistically) the most dangerous thing the average person does, I wouldn’t really say that people use the steering wheel just fine.
It’s just that computers are currently worse at it than humans.
In not entirely sure if I trust the statistics that are available for a couple reasons (and feel free to correct me if I’m wrong):
They are self reported by the manufacturer
Systems like autopilot will revert to manual control when it detects a situation it can’t handle, which means it has the luxury of “not being at fault during the crash” when it may have caused the situation 5 seconds before
It’s comparing to all vehicles instead of just vehicles that have similar non-self-driving but effective safety features
I wouldn’t even say that without seeing statistics to back it up. The news doesn’t cover routine traffic accidents, but one Tesla screws up one thing and that story is front page. Don’t rely on anecdotes and emotions.
i can use the steering wheel just fine, and people have been able to for a hundred years.
People have been bad at it for a hundred years. I’m not saying that people should necessarily be using auto-steering that keeps them in the middle of their lanes, but they should at least be using systems that beep at them when they stray out of their lane.
The bar for self-driving technology isn’t some amazing perfect computer that never makes a mistake. It’s the average driver. The average driver is bad.
we can do two things (these are not mutually exclusive):
-take further control away from the drivers and make them dependent on a computer, which can always misunderstand a situation and make the driver responsible for it.
-educate drivers properly, at least in the US. americans have been historically bad at driving and have also been known to be undereducated.
I’m all for more driver education, and for stricter licensing requirements like they have in Europe. Having said that, eventually computers are going to have to take over.
It’s pretty absurd that we’re handing control over multi-ton devices traveling at tens of meters per second to fallible, bored, easily distracted humans. The safer cars get, the safer drivers feel. The safer drivers feel, the less they feel they need to concentrate on driving.
Safe driving just will never be a skill that humans will be good at. The tasks that humans are good at that require concentration are tasks that are challenging and remain challenging. Think playing a sport where there’s always action and you have to react. Humans are bad at tasks that are mostly routine and boring, but if your concentration lapses you can cause a catastrophe. Those are the kinds of tasks where people get bored so they start glancing away, reading a book or looking at a smartphone, or whatever. For driving to be engaging, it has to be non-boring, which means non-safe. The safer it gets, the more boring it gets, so people stop paying the required attention. There’s just no winning.
That’s the bar that automatic driving has. It messes up once and you never trust it again and the news spins the failure far and wide.
Your uncle doing the same thing just triggers you to yell at him, the guy behind him flips you off, he apologizes, you’re nervous for a while, and you continue your road trip. Even if he killed someone we would blame the one uncle, or some may blame his entire class at worst. But we would not say that no human should drive again until it is fixed like we do with automated cars.
I do get the difference between those, and I do think that they should try to make automated drivers better, but we can at least agree about that premise: automated cars have a seriously unreasonable bar to maintain. Maybe that’s fair, and we will never accept anything but perfect, but then we may never have automated cars. And as someone who drives with humans every day, that makes me very sad.
There is a big difference between Autopilot and that hypotethical uncle. If the uncle causes an accident or breaks shit, he or his insurance pays. Autopilot doesn’t.
By your analogy, it’s like putting a ton of learner drivers on the road with unqualified instructors, and not telling the instructors that they are supposed to be instructors, but that they are actually taking a taxi service. Except it’s somehow their responsibility. And of course pocketing both the instruction and taxi fees.
The bar is not incredibly high for self driving cars to be accepted. The only thing is that they should take the blame if they mess up, like all other drivers.
Yeah, for sure. Like I said, I get the difference. But ultimately we are talking about injury prevention. If automated cars prevented one less death per mile than human drivers, we would think they are terrible. Even though they saved one life.
And even if they only caused one death per year we’d hear about it and we might still think they are terrible.
The difference is that Tesla said it was autopilot when it’s really not. It’s also clearly not ready for primetime. And auto regulators have pretty strict requirements about reliability and safety.
While that’s true that autonomous cars kill FAR less people than human drivers, ever human is different. If an autonomous driver is subpar and that AI is rolled out to millions of cars, we’ve vastly lowered safety of cars. We need autonomous cars to be better than the best driver because, frankly, humans are shit drivers.
I’m 100% for autonomous cars taking over entirely. But Tesla isn’t really trying to get there. They are trying to sell cars and lying about their capabilities. And because of that, Tesla should be liable for the deaths. We already have them partially liable: this case caused a recall of this feature.
Except Tesla’s uncle had brain damage and doesn’t really learn from the situation so will go it again, and had clones of him driving thousands of other cars.
Holy shit. My car doing that once and I’d be a nervous wreck just thinking about using it again.
I give teslas more room because I have been brake checked by them on empty roads before. These ghost brake problems are prevalent.
I have had the adaptive cruise control brake on multiple Hondas and Subarus in similar situations. Not like slamming on the brakes, but firm enough to confuse the hell out of me.
Every time it was confusing and now I just don’t use it if the road is anything but open and clear.
Honda’s sensing system will read shadows from bridges as obstructions in the road that it needs to brake for. It’s easy enough to accelerate out of the slowdown, but I was surprised to find that there is apparently no radar check to see if the obstruction is real.
My current vehicle doesn’t have that issue, so either the programming has been improved or the vendor for the sensing systems is a different one (different vehicle make, so it’s entirely possible).
i barely trust the lane-keeping assistant in my friend’s car. imagine going 70+km/h and suddenly the car decides to jerk the steering to the left/right because you weren’t exactly in the middle of your lane.
fuck modern assistants IMO. i can use the steering wheel just fine, and people have been able to for a hundred years.
Considering that driving is (statistically) the most dangerous thing the average person does, I wouldn’t really say that people use the steering wheel just fine.
It’s just that computers are currently worse at it than humans.
agreed. if “autopilot” becomes a better driver than the average person, then it has a right to exist.
Despite autopilot’s flaws, this is already true, if we are speaking statistically.
In not entirely sure if I trust the statistics that are available for a couple reasons (and feel free to correct me if I’m wrong):
I wouldn’t even say that without seeing statistics to back it up. The news doesn’t cover routine traffic accidents, but one Tesla screws up one thing and that story is front page. Don’t rely on anecdotes and emotions.
People have been bad at it for a hundred years. I’m not saying that people should necessarily be using auto-steering that keeps them in the middle of their lanes, but they should at least be using systems that beep at them when they stray out of their lane.
The bar for self-driving technology isn’t some amazing perfect computer that never makes a mistake. It’s the average driver. The average driver is bad.
we can do two things (these are not mutually exclusive):
-take further control away from the drivers and make them dependent on a computer, which can always misunderstand a situation and make the driver responsible for it.
-educate drivers properly, at least in the US. americans have been historically bad at driving and have also been known to be undereducated.
I’m all for more driver education, and for stricter licensing requirements like they have in Europe. Having said that, eventually computers are going to have to take over.
It’s pretty absurd that we’re handing control over multi-ton devices traveling at tens of meters per second to fallible, bored, easily distracted humans. The safer cars get, the safer drivers feel. The safer drivers feel, the less they feel they need to concentrate on driving.
Safe driving just will never be a skill that humans will be good at. The tasks that humans are good at that require concentration are tasks that are challenging and remain challenging. Think playing a sport where there’s always action and you have to react. Humans are bad at tasks that are mostly routine and boring, but if your concentration lapses you can cause a catastrophe. Those are the kinds of tasks where people get bored so they start glancing away, reading a book or looking at a smartphone, or whatever. For driving to be engaging, it has to be non-boring, which means non-safe. The safer it gets, the more boring it gets, so people stop paying the required attention. There’s just no winning.
That’s the bar that automatic driving has. It messes up once and you never trust it again and the news spins the failure far and wide.
Your uncle doing the same thing just triggers you to yell at him, the guy behind him flips you off, he apologizes, you’re nervous for a while, and you continue your road trip. Even if he killed someone we would blame the one uncle, or some may blame his entire class at worst. But we would not say that no human should drive again until it is fixed like we do with automated cars.
I do get the difference between those, and I do think that they should try to make automated drivers better, but we can at least agree about that premise: automated cars have a seriously unreasonable bar to maintain. Maybe that’s fair, and we will never accept anything but perfect, but then we may never have automated cars. And as someone who drives with humans every day, that makes me very sad.
There is a big difference between Autopilot and that hypotethical uncle. If the uncle causes an accident or breaks shit, he or his insurance pays. Autopilot doesn’t.
By your analogy, it’s like putting a ton of learner drivers on the road with unqualified instructors, and not telling the instructors that they are supposed to be instructors, but that they are actually taking a taxi service. Except it’s somehow their responsibility. And of course pocketing both the instruction and taxi fees.
The bar is not incredibly high for self driving cars to be accepted. The only thing is that they should take the blame if they mess up, like all other drivers.
Yeah, for sure. Like I said, I get the difference. But ultimately we are talking about injury prevention. If automated cars prevented one less death per mile than human drivers, we would think they are terrible. Even though they saved one life.
And even if they only caused one death per year we’d hear about it and we might still think they are terrible.
The difference is that Tesla said it was autopilot when it’s really not. It’s also clearly not ready for primetime. And auto regulators have pretty strict requirements about reliability and safety.
While that’s true that autonomous cars kill FAR less people than human drivers, ever human is different. If an autonomous driver is subpar and that AI is rolled out to millions of cars, we’ve vastly lowered safety of cars. We need autonomous cars to be better than the best driver because, frankly, humans are shit drivers.
I’m 100% for autonomous cars taking over entirely. But Tesla isn’t really trying to get there. They are trying to sell cars and lying about their capabilities. And because of that, Tesla should be liable for the deaths. We already have them partially liable: this case caused a recall of this feature.
Except Tesla’s uncle had brain damage and doesn’t really learn from the situation so will go it again, and had clones of him driving thousands of other cars.