Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Waymo robotaxi hits a child near an elementary school in Santa Monica

techcrunch.com

Post reply on HN
Print view View original

461–470 of 813 posts

Re: Waymo robotaxi hits a child near an elementary school in Santa Monica

#461

Post by smarnach »

NewJazz wrote: post #444
smarnach wrote: post #415

Earlier quoted context omitted.

> we should default to the calculation of 2-4x the rate. No we should not. We should accept that we don't have any statistically meaningful number at all, since we only have a single incident. Let's assume we roll a standard die once and it shows a six. Statistically, we only expect a six in one sixth of the cases. But we already got one on a single roll! Concluding Waymo vehicles hit 2 to 4 times as many children as…

Uh, the miles driven is like rolling the die, not hitting kids.

Sure, but we shouldn't stretch the analogy too far. Die rolls are discrete events, while miles driven are continuous. We expect the number of sixes we get to follow a binomial distribution, while we expect the number of accidents to follow a Poisson distribution. Either way, trying to guess the mean value of the distribution after a single incident of the event will never give you a statistically meaningful lower bound, only an upper bound.

Re: Waymo robotaxi hits a child near an elementary school in Santa Monica

#462

Post by Aloisius »

mlyle wrote: post #72
scarmig wrote: post #45

Earlier quoted context omitted.

It depends on the situation, and we need more data/video. But if there are a bunch of children milling about an elementary school in a chaotic situation with lots of double parking, 17 mph is too fast, and the Waymo should have been driving more conservatively.

> But if there are a bunch of children milling about an elementary school in a chaotic situation with lots of double parking, 17 mph is too fast Hey, I'd agree with this-- and it's worth noting that 17^2 - 5^2 > 16^2, so even 1MPH slower would likely have resulted in no contact in this scenario. But, I'd say the majority of the time it's OK to pass an elementary school at 20-25MPH. Anything carries a certain level of…

> Hey, I'd agree with this-- and it's worth noting that 17^2 - 5^2 > 16^2, so even 1MPH slower would likely have resulted in no contact in this scenario.

Only with instant reaction time and linear deceleration.

Neither of those are the case. It takes time for even a Waymo to recognize a dangerous situation and apply the brake and deceleration of vehicles is not actually linear.

Re: Waymo robotaxi hits a child near an elementary school in Santa Monica

#463

Post by aleksiy123 »

Bratmon wrote: post #432
matt-attack wrote: post #71

Earlier quoted context omitted.

Exactly. That’s why I’ve always said the driving is a truly AGI requiring activity. It’s not just about sensors and speed limits and feedback loops. It’s about having a true understanding for everything that’s happening around you: Having an understanding for the density and make up of an obstacle that blew in front of you, because it was just a cardboard box. Seeing how it tumbles lightly through the wind, and formi…

Slightly off topic, but it's endlessly funny to me watching people set the bar for AGI so high that only a small percentage of humans count as AGI.

humans aren't even a general intelligence at these requirements.

Re: Waymo robotaxi hits a child near an elementary school in Santa Monica

#465

Post by mlyle »

Aloisius wrote: post #462
mlyle wrote: post #72

Earlier quoted context omitted.

> But if there are a bunch of children milling about an elementary school in a chaotic situation with lots of double parking, 17 mph is too fast Hey, I'd agree with this-- and it's worth noting that 17^2 - 5^2 > 16^2, so even 1MPH slower would likely have resulted in no contact in this scenario. But, I'd say the majority of the time it's OK to pass an elementary school at 20-25MPH. Anything carries a certain level of…

> Hey, I'd agree with this-- and it's worth noting that 17^2 - 5^2 > 16^2, so even 1MPH slower would likely have resulted in no contact in this scenario. Only with instant reaction time and linear deceleration. Neither of those are the case. It takes time for even a Waymo to recognize a dangerous situation and apply the brake and deceleration of vehicles is not actually linear.

> It takes time for even a Waymo to recognize a dangerous situation

Reaction time makes the math even better here. You travel v1 * reaction_time no matter what, before entering the deceleration regime. So if v1 gets smaller, you get to spend a greater proportion of time in the deceleration regime.

> linear deceleration.

After reaction time, stopping distance is pretty close to n^2. There's weird effects at high speed (contribution from drag) and at very low speed, but they have pretty modest contributions.

Re: Waymo robotaxi hits a child near an elementary school in Santa Monica

#466

Post by trhway »

jjav wrote: post #154
jobs_throwaway wrote: post #8

Earlier quoted context omitted.

Yup. And to add > Waymo said in its blog post that its “peer-reviewed model” shows a “fully attentive human driver in this same situation would have made contact with the pedestrian at approximately 14 mph.” It's likely that a fully-attentive human driver would have done worse. With a distracted driver (a huge portion of human drivers) it could've been catastrophic.

> It's likely that a fully-attentive human driver would have done worse. We'd have to see video of the full scene to have a better judgement, but I wouldn't call it likely. The car reacted quickly once it saw the child. Is that enough? But most humans would have been aware of the big picture scenario much earlier. Are there muliple kids milling around on the sidewalk? Near a school? Is there a big truck/SUV parked th…

>The car can react faster that I can after seeing someone

and that can potentially allow internal planning algorithm to choose more risky and aggressive trajectories/behavior, etc. say to reach target destination faster and thus deliver higher satisfaction to the passengers.

Re: Waymo robotaxi hits a child near an elementary school in Santa Monica

#467

Post by parl_match »

direwolf20 wrote: post #108
parl_match wrote: post #101

Earlier quoted context omitted.

"and thereby hitting the child ... is a success." > What else to you expect them to do, only run on grade–separated areas where children can't access? no, i expect them to slow down when children may be present

how slow?

how about 10-15 mph if directly adjacent to a school, especially during the bands before and after school stars or ends. route away from schools whenever feasible.

Re: Waymo robotaxi hits a child near an elementary school in Santa Monica

#468

Post by jillesvangurp »

alkonaut wrote: post #77
jillesvangurp wrote: post #64

Earlier quoted context omitted.

I think those figures are already starting to accumulate. Incidents like this are rare enough that they are news worthy. Almost every minor incident involving Waymo, Tesla's FSD, and similar solutions gets a lot of press. This was a major incident with a happy end. Those are quite rare. The lethal ones even rarer. As for more data, there is a chicken egg problem. A phased roll out of waymo over several years has reve…

> ~17 deaths per 100K drivers per year. 40K+ fatalities overall. I also think one needs to remember those are _abysmal_ numbers, so while the current discourse is US centric (because that's where the companies and their testing is) I don't think it can be representative for the risks of driving in general. Naturally, robotaxis will benefit from better infra outside the US (e.g. better separation of pedestrians) but i…

Also fun to calculate how this compounds over say 40 years. You get to about 1 in 150 drivers being involved in some kind of deathly accident. People are really bad at numbers and assessing risk.

Re: Waymo robotaxi hits a child near an elementary school in Santa Monica

#469

Post by belorn »

jobs_throwaway wrote: post #8
BugsJustFindMe wrote: post #2

From the Waymo blog... > the pedestrian suddenly entered the roadway from behind a tall SUV, moving directly into our vehicle's path. Our technology immediately detected the individual as soon as they began to emerge from behind the stopped vehicle. The Waymo Driver braked hard, reducing speed from approximately 17 mph to under 6 mph before contact was made. > Following contact, the pedestrian stood up immediately, w…

Yup. And to add > Waymo said in its blog post that its “peer-reviewed model” shows a “fully attentive human driver in this same situation would have made contact with the pedestrian at approximately 14 mph.” It's likely that a fully-attentive human driver would have done worse. With a distracted driver (a huge portion of human drivers) it could've been catastrophic.

If you drive in Sweden you will occasionally come up to a form of speed reduction strategy that may seem counterintuitive. They all add to make driving harder and feel more dangerous in order to force attention and lower speed.

One is to merge opposite directional roads into a single lane, forcing drivers to cooperate and take turn to pass it, one car at a time.

For a combined car and pedestrian road (max speed of 7km/h) near where I live, they intentionally added large obfuscating objects on the road that limited visibility and harder to navigate. This forces drivers to drive very slow, even when alone on the road, as they can't see if a car or person may be behind the next object.

In an other road they added several tight S curves in a row, where if you drive anything faster than 20km/h you will fail the turns and drive onto the artificial constructed curbs.

In other roads they put a sign in the middle of two way roads while at the same time drastically limiting the width to the curb, forcing drivers to slow down in order to center the car in the lane and squeeze through.

In each of those is that a human driver with human fear of crashing will cause drivers to pay extra attention and slow down.

Re: Waymo robotaxi hits a child near an elementary school in Santa Monica

#470

Post by akoboldfrying »

smarnach wrote: post #415
Veserv wrote: post #244

Earlier quoted context omitted.

Err, that is not the desirable statistic you seem to think it is. American drivers average ~3 trillion miles per year [1]. That means ~7000 child pedestrian injurys per year [2] would be ~1 per 430 million miles. Waymo has done on the order of 100-200 million miles autonomously. So this would be ~2-4x more injurys than the human average. However, the child pedestrian injury rate is only a official estimate (it is pos…

> we should default to the calculation of 2-4x the rate. No we should not. We should accept that we don't have any statistically meaningful number at all, since we only have a single incident. Let's assume we roll a standard die once and it shows a six. Statistically, we only expect a six in one sixth of the cases. But we already got one on a single roll! Concluding Waymo vehicles hit 2 to 4 times as many children as…

More data would certainly be better, but it's not as bad as you suggest -- the large number of miles driven till first incident does tell us something statistically meaningful about the incident rate per mile driven. If we view the data as a large sample of miles driven, each with some observed number of incidents, then what we have is "merely" an extremely skewed distribution. I can confidently say that, if you pick any sane family of distributions to model this, then after fitting just this "single" data point, the model will report that P(MTTF zero incidents so far.
Post reply on HN