San Francisco cops pull over a Cruise driverless car for no lights on • Hacker News Board

Re: San Francisco cops pull over a Cruise driverless car for no lights on

#251

Post by kadoban » Mon, Apr 11, 2022, 2:49 PM UTC

xadhominemx wrote: post #101

jwr wrote: post #72

I am really worried by the fact that I am the unwilling tester in the Great Driverless Car Experiment. Tradition has it that when you load-test a new bridge, you put the architect underneath. I feel like this, except I didn't design those driverless cars, somebody else did. Being an experienced software engineer, my trust in the software in these cars is pretty low. And yet they are testing them on me, because I can…

We are all unwilling testers of every new teenage driver who hits the road.

That teenage driver goes through a training and certification process, and we have ways of stopping them from driving if they fuck up too badly. They also have liability for their actions.

Most of those are lacking or at least inconsistent, for AI driving.

Re: San Francisco cops pull over a Cruise driverless car for no lights on

#252

Post by fatnoah » Mon, Apr 11, 2022, 2:49 PM UTC

swozey wrote: post #229

fatnoah wrote: post #226

Earlier quoted context omitted.

I'm also not sure why this is a thing. With all of the various sensors, etc. in cars, there's safeguards to prevent me from typing an address in the GPS while moving, so why can't there be something that turns your headlights on while driving at night above a certain speed. On Saturday, I drove from NJ to MA and saw 3 vehicles on the highway at night with only daytime running lamps, and one with no lights on whatsoev…

I haven't driven a modern car in a decade or more that didn't have light sensors in the dashboard to know when to automatically turn on lights. They even turn them on when the wipers are engaged. Everytime you go through a tunnel you'll see headlights pop on automatically. My 2009 BMW is the oldest car I have that has auto lights, but it doesn't have wiper sensors. Older cars? Of course not.

The problem with auto lights is that they can be turned off or set into other modes that prevent the auto part from working. Throw in drivers who don't seem to notice that their headlights are not or barely illuminating the road and we get what I experienced over the weekend.

Maybe there's hope: https://www.edmunds.com/car-news/make-sure-your-headlights-a...

Re: San Francisco cops pull over a Cruise driverless car for no lights on

#253

Post by dylan604 » Mon, Apr 11, 2022, 2:50 PM UTC

tremon wrote: post #239

dylan604 wrote: post #227

Earlier quoted context omitted.

The teenager's ability to drive safely isn't just a mechanical ability of driving a car or of just being aware of the laws/rules. Safe driving comes from years of experience on being able to decide how long it will take to stop at certain speeds in certain conditions. Being able to recognize their car's brakes are a little soft and might not stop on a dime any more. Being able to recognize that the driver in front of…

experience vs book learning Why would you assume that a driver's education is a matter of book learning? Is that how you were taught?

Yes it was. I am of a certain age that we had a certain number of classroom hours as well as a certain number of hours behind the wheel with an instructor as well as certain number of hours in a car as an observer with an instructor.

We had book(lets) that we had to read from to learn the rules/laws of the road. There were quizzes/tests to be taken during those classes in order to pass the class. Once the class was passed, I then had to take a written exam issued by the state. That was followed by an actual driving test with a state official in the car.

So assuming based on my personal experience. I know the times have changed, and you no longer have to do silly things like learn how to drive or be required to wear shoes while driving in today's world.

Re: San Francisco cops pull over a Cruise driverless car for no lights on

#254

Post by eru » Mon, Apr 11, 2022, 2:50 PM UTC

bee_rider wrote: post #237

eru wrote: post #173

Earlier quoted context omitted.

Investigated, yes. Grounding the whole fleet? Not necessarily. It's a cost-benefit trade-off. We don't ground the whole fleet of human drivers, when one of them crashes their car, because the benefit doesn't outweigh the costs. It's not a given that the answer will change that much for self-driving cars. Especially for extremely rare failure modes.

Humans all have different versions of the 'driving' software installed, there's no way of identifying which ones will misbehave. If Teslas are malfunctioning, we shouldn't ground Chevy Cruise. We exist on a spectrum of competence. If a type of car has a widespread issue that unexpectedly gets it into a state where a minimally competent non-impaired person can't operate it safely, that type of car should be pulled fro…

Well, the obvious incentive with such a rule would then be to install slightly different versions of the software in every vehicle.

(Or perhaps to have separate versions on file, and just load that different version when the first version is grounded.)

> Without a human in the loop, what's the unreliable part to point the finger at?

It's a cost benefit analysis. If the problem is sufficiently rare, it might make more sense to leave the cars running while you investigate. Even human lives aren't infinitely valuable.

Typical traffic engineering in the US assumes a cost of human life of about 6 million USD or so.

Re: San Francisco cops pull over a Cruise driverless car for no lights on

#255

Post by argiopetech » Mon, Apr 11, 2022, 2:50 PM UTC

fho wrote: post #200

ISL wrote: post #137

Earlier quoted context omitted.

And every drunk/elderly driver -- the car's decision to move after a traffic-stop felt very much the behavior of someone with impaired cognition.

Anecdote: There was an old lady living nearby who despite being 80+ years still driving around in her 20 year old car. She literally ran over traffic cones and ... a child on a tricycle [1] and never got prosecuted. She recently passed away ... sad, but the streets around here are definitely safer now. (This was in Germany just for reference) [1] The child was fine it was a low speed collision ... but still ...

I caution against age as a sole proxy for driving ability. As a contradictory datum, my 80+ year old grandmother (in the USA) maintains a 100+ acre farm. She daily drives a (impeccably maintained) 2006 3/4 ton pickup, to include multiple yearly 10+ hour trips hauling a 10k+ lbs trailer. She almost always brings a passenger on these trips for companionship and safety. I've never heard any complaints, and never personally observed her driving unsafely.

It won't be this way forever, and I'm under no illusion that she is the norm for her age, but I'd hazard she's a better drive R than most people on the road.

I would like to see a move toward more frequent driver testing for all drivers in the USA, decreasing in frequency after the first few years of driving and increasing in frequency for older drivers. Testing after initial licensing is not the norm in any state I've inhabited, so this would probably require a significant uphill battle on both the individual and government fronts.

Re: San Francisco cops pull over a Cruise driverless car for no lights on

#256

Post by eru » Mon, Apr 11, 2022, 2:50 PM UTC

joshribakoff wrote: post #192

eru wrote: post #170

Earlier quoted context omitted.

Depends. In principle you can log exactly what data the camera sends to the decision making unit. And you could replay that data. The camera itself might not be deterministic enough to reproduce the same data with the same beetle, but replaying the log should work. (Keep the log in a 'black box' on the car, and perhaps only keep the last x hours of data, if necessary.) I think Waymo already does something like these…

what you’re referring to is just simulating, at least after the point your modified algorithm does something slightly different in the simulator than the real car did in real life.

I wondered about the last point, too. But Waymo apparently worked out a satisfactory solution to that.

Re: San Francisco cops pull over a Cruise driverless car for no lights on

#257

Post by AviationAtom » Mon, Apr 11, 2022, 2:51 PM UTC

sseagull wrote: post #65

DennisP wrote: post #58

Earlier quoted context omitted.

Not necessarily, if you make your intentions clear, e.g. by slowing down and putting your hazards on.

Well, most of the time. If the cop feels like it, they can make you have a very bad time. https://amp.usatoday.com/amp/7645474002 “An Arkansas woman says she had no safe place to pull over in July 2020 when a state trooper tried to stop her for speeding, so she turned on her hazard lights and slowed down. Moments later, the officer rammed her vehicle, causing it to flip over and injuring the woman”

Assuming you aren't driving far, the police will generally understand, and appreciate your regard for their safety. I have seen people choose to pull over in the worst of spots, where I am almost certain the police would have appreciated them moving to a safer spot.

As for the Arkansas lady, did you watch the video? IIRC, she went quite a ways before he PIT'd her. I would agree he shouldn't have done the PIT on her so soon, but it wasn't the case of her going a few hundred feet more.

Re: San Francisco cops pull over a Cruise driverless car for no lights on

#258

Post by ISL » Mon, Apr 11, 2022, 2:52 PM UTC

argiopetech wrote: post #255

fho wrote: post #200

Earlier quoted context omitted.

Anecdote: There was an old lady living nearby who despite being 80+ years still driving around in her 20 year old car. She literally ran over traffic cones and ... a child on a tricycle [1] and never got prosecuted. She recently passed away ... sad, but the streets around here are definitely safer now. (This was in Germany just for reference) [1] The child was fine it was a low speed collision ... but still ...

I caution against age as a sole proxy for driving ability. As a contradictory datum, my 80+ year old grandmother (in the USA) maintains a 100+ acre farm. She daily drives a (impeccably maintained) 2006 3/4 ton pickup, to include multiple yearly 10+ hour trips hauling a 10k+ lbs trailer. She almost always brings a passenger on these trips for companionship and safety. I've never heard any complaints, and never persona…

Triggering testing/education after traffic violations might be a middle-ground.

Re: San Francisco cops pull over a Cruise driverless car for no lights on

#259

Post by lolc » Mon, Apr 11, 2022, 2:52 PM UTC

jillesvangurp wrote: post #238

IgorPartola wrote: post #88

Being used to riding motorcycles in the US, why are headlights not always on for all vehicles at this point? LED bulbs last a very long time and there is little reason to not have them on during the day. I know, it’s much more fun to anthropomorphize the car in this case, but the simple solution is to just hard wire them.

It's been required to do so in some northern European countries. The practical reason there is that dawn and dusk can take many hours during much of the year at higher latitudes thus creating a weird situation where it's not really clear when exactly you should have the lights on or off in exactly the kind of situation where a lot of accidents can happen if you don't have them on. The simple rule of "always have your…

I've gotten so used to lights being on always that I classify cars with lights off as parked. I won't focus on them. I'm very surprised when they move.

It should really be hardwired, because people forget.

Re: San Francisco cops pull over a Cruise driverless car for no lights on

#260

Post by eru » Mon, Apr 11, 2022, 2:52 PM UTC

ocdtrekkie wrote: post #196

eru wrote: post #164

Earlier quoted context omitted.

I wonder whether that would just lead to people using slightly tweaked variants of the same algorithm branded to be different? (Of course, you don't want to fork too much, because there's probably a certification per algorithm needed. But it might make sense for Google to officially field a thousand 'different' algorithms, so that a single incident doesn't ground the entire fleet?)

Engineers often fail to realize you can't cheese a judge and if you try to, the law will come down much, much harder on you. Either one of two things will be the case with this approach: 1. The algorithm will be too imperceptibly different to be treated as a different program. Adding some tiny decimals somewhere won't make it unlikely to make the same mistakes, and the entire class of algorithms will be treated as on…

Judges are human indeed. But things like tax dodges still exist and other loopholes. You just need good enough lawyers to structure things for you.

Your second point would lead to a ban on all but one self-driving car systems, when taken seriously?