Though ... how good a job are humans actually doing in heavy precipitation? I know that under normal circumstances our brains constantly do a bunch of work to create the illusion of a comprehensive high res visual field even though we really only have detail at the fovea. When it's raining heavily, and we think we can see "enough" to drive ... are we right? Or are we just lucky and pedestrians and cyclists are more likely to be off the road at those moments and so accidents increase but not to the point of disaster?

Agreed - I think the "but can it drive in a whiteout blizzard" question is best redirected to: "can a human?"

I suspect there are operating conditions the AVs won't solve acceptably - some of those conditions IMO are also conditions where we should not accept a human to solve acceptably. In general I feel we have a lackadaisical culture around driving that encourages/excuses unsafe behavior, and is overoptimistic about people's ability to drive well.

I grew up in a part of Ontario where total whiteouts can happen fairly frequently on both major highways leading in and out of my small town. It is in fact possible to drive dozens of km in near or total whiteout conditions simply by the hazard lights of the car ahead of you. You very frequently will see lines of cars km long, all going <20km/h, white knuckled and crawling home. Maybe one in every couple thousand goes in the ditch.

I don't think vision-based FSV will ever reliably handle winter conditions like this. The engineering and QA effort just isn't worth the cost-benefit when you factor in the very small amount of drivers who are consistently exposed to conditions like that. My father, who spent his career commuting to the city on that highway, was disappointed when I explained this to him.

I was once in the passenger seat in a downpour. My father was driving to the nav, and it seemed like we were traversing Mekong underwater. It was a complete instrument driving condition, except at most a feet of road markings were visible. The car was on local roads. He made cautious turns and drove slow, because it was obviously scary. Suddenly the nav said "Ding! you have reached your destination" in what seems to be middle of a road, and we immediately started making noises at the nav.

Then a person knocked on a window through the brown wall. It was someone we were to meet at the destination. He greeted us, and told us to come out. We tried to explain we can't just walk all perhaps a quarter mile to the place in this heavy rain, leaving the car left at a roadside. He insisted it'll be a short walk, and gave us no choice. Only when we stepped out, we realized that the car is right in the middle of the premise we were looking for, just couple feet from the main door.

This memory surfaces to my mind in the context of human drivers and inclement weathers; I'm still one piece, but maybe that has more to do with my luck, not necessarily due to myself playing every games extra safe.

Humans certainly don't reliably handle these conditions. ;)

It does seem like "something else" is needed for these kinds of low-visibility scenarios -- frankly, when nobody should be on the road.

The reason all that works is people drive to what they expect. In such conditions you might hit a human standing on the road, no human would be there in the first place, only other cars with flashing lights. As such so long as you stay in the correct lane for your direction of travel and go slow you don't need to see because there is no real danger most of the time. Most of the time...

The worst is when one goes into a ditch and the car following them follows into the ditch because their main indicator of where to go was the running lights and tire tracks of the car in front.

Where I grew up people would go together off the side of the mountain this way in very heavy fog.

rumble strips are amazing in whiteouts. its a relief when you hear it because you now know where the side of the road is...

people are actually pretty good at driving in blizzards in locales where they happen often. snow tires (and possibly chains), good clearance, and great caution can get you a long way.

obviously you try to avoid driving in these conditions when possible, but sometimes a moderate storm is much more intense than forecast and you get caught out. pulling off to the side of a snowy mountain pass doesn't guarantee your survival either.

I know next to nothing about lidar engineering but 60GHz band radars can still function out to several hundred meters in rain. It is significantly attenuated as the rain rate (in mm/hour) increases, but it takes a lot of rain to make it completely useless.

This depends on how powerful your antenna are, $200 wigig transmitters will struggle with much range over those distances.

And how directional and narrow the gain pattern is.

Summers in South Florida will put that to the challenge.

People can't see in that weather either

This has been my argument all along. People are driving in unsafe conditions when they should be stopped.

In florida, I will say when it truly pours like that people do tend to drive extremely slowly and turn on their hazards. Also, sometimes these storms just appear out of nowhere. Over the summer I was driving from the Vero beach area to fort lauderdale and on the way to Vero beach, clear skies and on the way back there was an ENORMOUS storm that flooded streets and you couldn't see crap. It just happens

Moreover, contrary to some other commenters:

Unless it truly is "once-in-a-lifetime", or very brief, you cannot just stop and give up. To be a viable replacement, a vehicle must (as a human driver would) continue to make progress even in extremely adverse conditions. The progress might be much slower than usual, might be a re-routing (back up away from floodwater and go elsewhere), etc.

just because you can't write code to handle it doesn't mean a human can't. thousands and thousands of people drive in rain, sleet, snow and hail daily and do it just fine.

Humans do not do just fine - look at the statistics of car accidents. Humans refuse to admit how bad they really are at driving.

Airplanes couldn't fly in inclement weather for decades. Took a while but we solved it.

"Oops it's pouring, can't get a Cruise, gotta fall back to an Uber" doesn't sound that terrible to me, for now.

Cruise could even offer a product that compensates/insures you for that eventuality, if for example Cruise was your primary vehicle.

But if self-driving succeeds then there won't be enough gig-drivers ready-to-go to cover a self-driving outage.

Taxis, available with e.g. a simple phone call, have existed since before telephones and cars really, and they still exist now even as ride-sharing has taken over. They will exist as Cruise rises.

They exist because they're a viable business because there is enough approximately consistent demand. That's not guaranteed to continue.

There might be taxis, but not enough for everyone who wants to catch one. It’s already difficult enough on rainy days.