BMW says it wasn't a self-driving car: https://www.focus.de/panorama/welt/b28-im-kreis-reutlingen-s...
It shouldn't come as a shock, since self-driving cars currently do not exist - not even in the prototype stage.
One dead and nine wounded after accident with self-driving car
31–40 of 99 posts
Re: One dead and nine wounded after accident with self-driving car
#32BMW has denied that it was an autonomous / self-driving test-car [1]. The article is from Reutlinger Generalanzeiger, a regional newspaper, and they probably have better sources than Der Spiegel. The B28 is a very scenic drive, the region is also not a densely populated area. I used to live close by. [1]: https://www.gea.de/blaulicht_artikel,-schwerer-crashnahe-zai...
A sticker on the vehicle clearly shows it was BMW autonomous test vehicle: https://twitter.com/AEONde/status/1559477911446589441
Those stickers don't prove anything, especially if BMW themselves deny it. You can find those stickers on support-vehicles without any autonomous features as well. There will be more press coverage anyway and we'll find out soon.
Re: One dead and nine wounded after accident with self-driving car
#33Serious question: will we demand perfection from autonomous cars before we allow them out of “test” status? How many autonomous miles without an accident will be sufficient to prove safety? I am not diminishing this tragedy; I can only imagine the grief of the families involved and my heart goes out to them. I am also NOT making the “have to break a few eggs to make an omelette“ argument; clearly something went wrong…
“Autonomous miles without accident” is not the only metric. What matters is also how the algorithm fails.
To give an example. If you have autonomous miles without accident that is two times better than humans, but every accident tends to involve dark-skinned children and women, then the algorithm can have better statistical numbers in safety but still be rejected by humans.
Safety is a high dimensional problem that cannot be collapsed into one metric.
Re: One dead and nine wounded after accident with self-driving car
#34Need a bit more regularization on the neural network.
Re: One dead and nine wounded after accident with self-driving car
#35Serious question: will we demand perfection from autonomous cars before we allow them out of “test” status? How many autonomous miles without an accident will be sufficient to prove safety? I am not diminishing this tragedy; I can only imagine the grief of the families involved and my heart goes out to them. I am also NOT making the “have to break a few eggs to make an omelette“ argument; clearly something went wrong…
“Autonomous miles without accident” is not the only metric. What matters is also how the algorithm fails. To give an example. If you have autonomous miles without accident that is two times better than humans, but every accident tends to involve dark-skinned children and women, then the algorithm can have better statistical numbers in safety but still be rejected by humans. Safety is a high dimensional problem that c…
So if there's fewer deaths overall, including among black kids, but all the deaths are black kids that's not ok? I get your larger point but this one seems like a really poorly reasoned example.
Re: One dead and nine wounded after accident with self-driving car
#36Serious question: will we demand perfection from autonomous cars before we allow them out of “test” status? How many autonomous miles without an accident will be sufficient to prove safety? I am not diminishing this tragedy; I can only imagine the grief of the families involved and my heart goes out to them. I am also NOT making the “have to break a few eggs to make an omelette“ argument; clearly something went wrong…
Serious question: will we demand perfection from autonomous cars before we allow them out of “test” status?
I don't think anyone is demanding perfection, but they are quite reasonably demanding that autonomous driving doesn't cause any accidents that wouldn't have happened if a human was driving (not that this is clear from the accident in the article).
The bar issimply 'better than a person'. If a car apparently changes lane and drives into another car without the cause being clear it's not there yet.
Re: One dead and nine wounded after accident with self-driving car
#37Serious question: will we demand perfection from autonomous cars before we allow them out of “test” status? How many autonomous miles without an accident will be sufficient to prove safety? I am not diminishing this tragedy; I can only imagine the grief of the families involved and my heart goes out to them. I am also NOT making the “have to break a few eggs to make an omelette“ argument; clearly something went wrong…
> will we demand perfection from autonomous cars before we allow them out of “test” status? Not perfection, just no human injuries for the limited duration of the tests, that is, for a few billion km or so.
That's a standard no human driver could possibly achieve, for the simple fact that no human drove a billion miles. I think setting such standards, which will be close to impossible to achieve (and/or take centuries) is helpful.
Re: One dead and nine wounded after accident with self-driving car
#38Serious question: will we demand perfection from autonomous cars before we allow them out of “test” status? How many autonomous miles without an accident will be sufficient to prove safety? I am not diminishing this tragedy; I can only imagine the grief of the families involved and my heart goes out to them. I am also NOT making the “have to break a few eggs to make an omelette“ argument; clearly something went wrong…
> will we demand perfection from autonomous cars before we allow them out of “test” status? Not perfection, just no human injuries for the limited duration of the tests, that is, for a few billion km or so.
How many human injuries are there per few billion km when humans are driving? Why isn't that the baseline we compare against?
Re: One dead and nine wounded after accident with self-driving car
#39BMW says it wasn't a self-driving car: https://www.focus.de/panorama/welt/b28-im-kreis-reutlingen-s...
It shouldn't come as a shock, since self-driving cars currently do not exist - not even in the prototype stage.
Actually, they are existing in prototype stage.
EDIT: They exists since the 1970s on closed tracks. I remember during my work in 1990s at DASA (Daimler Chrysler Areo Space) they had them on the airfield in Friedrichshafen.
Re: One dead and nine wounded after accident with self-driving car
#40Earlier quoted context omitted.
“Autonomous miles without accident” is not the only metric. What matters is also how the algorithm fails. To give an example. If you have autonomous miles without accident that is two times better than humans, but every accident tends to involve dark-skinned children and women, then the algorithm can have better statistical numbers in safety but still be rejected by humans. Safety is a high dimensional problem that c…
So if there's fewer deaths overall, including among black kids, but all the deaths are black kids that's not ok? I get your larger point but this one seems like a really poorly reasoned example.
It would depend on the details.
If black kids are dying because the car accelerates in school zones and swerves directly into black kids at high speeds, you can imagine that would obviously be not okay no matter what the statistics are.
My point is that a single metric like “deaths overall” is insufficient. The failure modes need to be reasonable and sympathetic to a human.