Live data from Hacker News

Replaying real life: how the Waymo Driver avoids fatal human crashes

blog.waymo.com

11–20 of 182 posts

Re: Replaying real life: how the Waymo Driver avoids fatal human crashes

#11
1) Would a random driver (random steering and throttle) also have avoided these crashes? It seems like they were fatal because several variables coincided very exactly, and really any other input would have avoided them.

2) What input is the Waymo driver using here? If these were not originally Waymo cars, then is it perceiving simulated video or getting raw access to ground truth vehicle positions?

Re: Replaying real life: how the Waymo Driver avoids fatal human crashes

#12
post #5

Earlier quoted context omitted.

Well, devil's advocate here, but teaching a car to drive in good weather is hard enough, and has lots of applications! It doesn't seem totally unreasonable to focus on that first?

Yes, if basing your entire model on visible road markings that is broken by a light coating of crystallized water.

Not if the roads are mapped. You have prior knowledge of exactly where the markings are even if it’s not entirely visible due to rain.

Re: Replaying real life: how the Waymo Driver avoids fatal human crashes

#13
post #5

Earlier quoted context omitted.

Well, devil's advocate here, but teaching a car to drive in good weather is hard enough, and has lots of applications! It doesn't seem totally unreasonable to focus on that first?

Yes, if basing your entire model on visible road markings that is broken by a light coating of crystallized water.

If the road markings are visible, do you use them? Or do you ignore them and never use them since they may sometimes be covered?

The former?

So makes sense for an automatic system to do the same?

> a light coating of crystallized water

Do you mean snow?

Re: Replaying real life: how the Waymo Driver avoids fatal human crashes

#15
post #5

Earlier quoted context omitted.

Well, devil's advocate here, but teaching a car to drive in good weather is hard enough, and has lots of applications! It doesn't seem totally unreasonable to focus on that first?

Yes, if basing your entire model on visible road markings that is broken by a light coating of crystallized water.

Oh. Personally I learned how to drive in good weather first.

Re: Replaying real life: how the Waymo Driver avoids fatal human crashes

#16

Figure 3 has the responder slowing unusually soon. Makes me wonder if their model is overly fit to the sample. Still, if it's genuinely that good at responding and mitigating then the future looks bright.

Figure 3 almost looks like the responder is lagging. Feels like it slows down and speeds up in several places.

Assuming that’s not a gif/video rendering bug, it’s a great way to get rear-ended.

Re: Replaying real life: how the Waymo Driver avoids fatal human crashes

#17
This is really cool.

I suspect that when we have self driving cars, it's going to take off very quickly due to insurance. In particular, if you are in a non self driving car and in an accident with a self driving one, it'll be hard to not be shown as at fault. So insurance I would guess go up based on the percentage of self driving cars, and likely be really cheap for self driving cars.

I wonder if there's a period though where people cut off self driving cars, knowing they won't get hit. I suspect pedestrians in cities will jaywalk a lot more (and maybe that's ok).

Re: Replaying real life: how the Waymo Driver avoids fatal human crashes

#18
You can read the California DMV's set of autonomous vehicle accident reports.[1] Almost all the Waymo reports are "vehicle was entering intersection, detected cross traffic, stopped, was rear-ended by human driver". There's one Waymo report where someone ran a stop sign and hit them.

The rear-ending problem will be solved as automatic emergency braking becomes standard on cars. Already, it's shipping on almost all high-end cars and 70-80% of midrange cars. There's a US auto industry goal of it being standard by 2022. That's probably the main feature needed for self-driving cars to coexist with human-driven cars.

[1] https://www.dmv.ca.gov/portal/vehicle-industry-services/auto...

Re: Replaying real life: how the Waymo Driver avoids fatal human crashes

#19
post #2

What do you call that when you subconsciously suppress any tests or scenarios you know will break your code? Like you notice somethings kinda racey but it involves other components you don't control and so you don't do the parallel tests you probably should, that kinda thing. Because that is exactly the bad feeling I get with Waymo never actually testing anywhere that doesn't have mm-precise maps and the all importan…

Even if you require precise mapping, that's not at all a deal breaker. Google went and mapped rural Kenya to make Google Maps.

Mapping wealthy cities with >1 million population for a product with vastly more revenue potential is a trivial hurdle.

Re: Replaying real life: how the Waymo Driver avoids fatal human crashes

#20
post #18

You can read the California DMV's set of autonomous vehicle accident reports.[1] Almost all the Waymo reports are "vehicle was entering intersection, detected cross traffic, stopped, was rear-ended by human driver". There's one Waymo report where someone ran a stop sign and hit them. The rear-ending problem will be solved as automatic emergency braking becomes standard on cars. Already, it's shipping on almost all hi…

>The rear-ending problem will be solved as automatic emergency braking becomes standard on cars. Already, it's shipping on almost all high-end cars and 70-80% of midrange cars. There's a US auto industry goal of it being standard by 2022. That's probably the main feature needed for self-driving cars to coexist with human-driven cars.

I really wish we could just enforce proper education of being observant.

Most times when I hear about people getting rear-ended it's because the driver behind them was looking at their phone or otherwise distracted. We shouldn't invent features that allow that to continue.

On the other hand, I think this is fine in some rare cases, like maybe an accident or some other incident happens up ahead, a driver stomps their brakes, and then the person behind them who doesn't know what's happened doesn't stop soon enough. But even for the latter case, you could argue that the DMV says to keep enough distance where in most cases you should be able to still come to a stop with barely a moment's notice as long as you're paying enough attention. Of course, most people don't leave that much space much, if ever.

Post reply on HN