Live data from Hacker News

Cruise told by CA DMV to reduce robotaxi fleet 50% following crash

techcrunch.com

101–110 of 205 posts

Re: Cruise told by CA DMV to reduce robotaxi fleet 50% following crash

#101
post #17

> “Over one hundred people lose their lives every day on American roadways, and countless others are badly injured, Cruise said in a statement sent via email. I’m so tired of this angle being accepted at face value. Current driverless cars are not even capable of safely navigating situations that below-average human drivers rarely struggle with, like avoiding driving into clearly marked wet concrete, or not crashing…

Just because they have weird failure cases does not mean they fail more often than human drivers. FWIW as a biker in SF i feel much more comfortable around driverless than actual drivers, especially in areas like the Mission where people don’t check their right before turning onto streets like Guerrero or Valencia. The self driving ones always do

Evaluating safety of self driving cars is not just a matter of comparing number of failures of number of accidents, or any one scalar number.

Just to illustrate my point, imagine a case where self driving cars have same number of fatal deaths (normalised by whatever you need to convince yourself), but self driving cars disproportionately fail on dark skinned children. Then it is obvious that self driving cars are worse.

Weird failure cases, cases where a bad human driver would do better, are bad for self driving cars.

Re: Cruise told by CA DMV to reduce robotaxi fleet 50% following crash

#102
post #77
post #17

> “Over one hundred people lose their lives every day on American roadways, and countless others are badly injured, Cruise said in a statement sent via email. I’m so tired of this angle being accepted at face value. Current driverless cars are not even capable of safely navigating situations that below-average human drivers rarely struggle with, like avoiding driving into clearly marked wet concrete, or not crashing…

You’re missing the point. It’s an appeal that we need to change the status quo. They may not be there yet. But AVs are the only serious contender to solve the traffic fatality problem.

Nah, walkable cities, physically separated bike lanes, public transit, and trains. Also cheaper, leads to better health outcomes (more walking and biking), and more sustainable. And as a bonus, you avoid having thousands of cameras trolling around at all times.

Re: Cruise told by CA DMV to reduce robotaxi fleet 50% following crash

#103

Earlier quoted context omitted.

Trains. Buses. Bikeable areas. How did we land in a spot where the “only serious contender” to solve this problem is VC funded companies using the roads and people’s lives as beta testing grounds? Absolutely absurd.

I get where you’re coming from but SF is full of busses, it has trams, light rail, (even some cable cars!) and a subway. A couple of years ago it was stuffed full of electric scooters. Bikeable areas are constrained by the hills or practically impossible due to the road layout (bus lanes, tram tracks, crazy Victorian street plan) pedestrians who go literally anywhere they want at any time, questionable grid system th…

Some of these things are basically solved (e-bikes chew through hills, physically separated bike lanes solve the "no space for bikes" issue, etc) but SF is still super car friendly. It has to be because of road and zoning regulations, so roads have to be maintained at a certain level and must be a certain width, and there must be a certain amount of parking, etc. If we gave walkability and biking the same amount of deference and funding our cities would be cyclist/pedestrian paradises.

Re: Cruise told by CA DMV to reduce robotaxi fleet 50% following crash

#104

I've said it before when a self-driving Uber killed a woman, and I'll say it again: I don't understand why authorities allow self-driving cars to beta test out in public. I thought Waymo was the best with something like millions of hours on the road but when they started accepting passengers, it failed to proceed past a traffic cone: https://youtu.be/zdKCQKBvH-A?t=742 I wouldn't trust any of the others like Tesla and…

> why authorities allow self-driving cars to beta test out in public

because that's the definition of a beta test?

they have gone through a number of test stages over the years, and if the goal is that they drive in public, the penultimate step is test driving in public.

if your criticism is that they didn't do enough prior testing, could be, but you need to make that case.

Re: Cruise told by CA DMV to reduce robotaxi fleet 50% following crash

#105
post #35
post #31

Earlier quoted context omitted.

but just because they have weird failure cases does not mean they fail less often than human drivers either.

Right, but the OP seems to be skipping that part entirely. The only metric we should care about is overall accident rates (controlling for reasonable variables of course), not "safely navigating situations that below-average human drivers rarely struggle with".

Given that driverless cars are slim minority on our roads, if they exist at all, failure modes actually do matter a lot. Human drivwr failure modes are what everybody is used to, and trained by everyday life to cope with. The radom, unpredictable stuff driverless cars do is in itsrlf dangerous from that perspective.

Re: Cruise told by CA DMV to reduce robotaxi fleet 50% following crash

#106
post #17

> “Over one hundred people lose their lives every day on American roadways, and countless others are badly injured, Cruise said in a statement sent via email. I’m so tired of this angle being accepted at face value. Current driverless cars are not even capable of safely navigating situations that below-average human drivers rarely struggle with, like avoiding driving into clearly marked wet concrete, or not crashing…

Better cognition? Maybe. As long as planners contain a single if statement in C++, totally.

But sensors? These cars scan the environment with dozens of sensors, much better than what a human can do.

Better perception? Yes, although the perception models have come a long way but they are still quite primitive compared to what a human can do.

Re: Cruise told by CA DMV to reduce robotaxi fleet 50% following crash

#107
post #77

Earlier quoted context omitted.

You’re missing the point. It’s an appeal that we need to change the status quo. They may not be there yet. But AVs are the only serious contender to solve the traffic fatality problem.

Trains. Buses. Bikeable areas. How did we land in a spot where the “only serious contender” to solve this problem is VC funded companies using the roads and people’s lives as beta testing grounds? Absolutely absurd.

Driver training, road design, car design and safety regulations... The list is basically endless.

Re: Cruise told by CA DMV to reduce robotaxi fleet 50% following crash

#108
post #17

> “Over one hundred people lose their lives every day on American roadways, and countless others are badly injured, Cruise said in a statement sent via email. I’m so tired of this angle being accepted at face value. Current driverless cars are not even capable of safely navigating situations that below-average human drivers rarely struggle with, like avoiding driving into clearly marked wet concrete, or not crashing…

> Current driverless cars are not even capable of safely navigating situations that below-average human drivers rarely struggle with, like avoiding driving into clearly marked wet concrete, or not crashing into an emergency vehicle that has its sirens and lights blazing.

Those are both very common accidents for human drivers. It's fine if you want to argue with numbers about whether or not Cruise/Waymo/Tesla are more or less likely to get into these accidents. But please don't spin and claim that people don't do this too. That is simply incorrect and you know it.

Re: Cruise told by CA DMV to reduce robotaxi fleet 50% following crash

#109
post #99

Earlier quoted context omitted.

Absolutely insane that your society allows children to experiment on unwitting people without consent.

What are you talking about?

You have untrained people using multi-ton motorized equipment on streets where people cross. And you don't even ask them if they'd be okay with this experiment. At least a bit unethical.

Re: Cruise told by CA DMV to reduce robotaxi fleet 50% following crash

#110
post #101

Earlier quoted context omitted.

Just because they have weird failure cases does not mean they fail more often than human drivers. FWIW as a biker in SF i feel much more comfortable around driverless than actual drivers, especially in areas like the Mission where people don’t check their right before turning onto streets like Guerrero or Valencia. The self driving ones always do

Evaluating safety of self driving cars is not just a matter of comparing number of failures of number of accidents, or any one scalar number. Just to illustrate my point, imagine a case where self driving cars have same number of fatal deaths (normalised by whatever you need to convince yourself), but self driving cars disproportionately fail on dark skinned children. Then it is obvious that self driving cars are wor…

Your point is that we shouldn't have self-driving vehicles because... we can't prove they're not racist? I'm not following.

I mean, yes, you're correct that there are edge cases in any moral analysis and it's not just about comparing two numbers (though in safety arguments, yeah, it's almost always about simple numbers).

But you don't get to wave that fact as a get-out-of-jail-free card in an argument! You need to show us the edge case. And the best you can do is "the cars might be racist"?

Post reply on HN