Comment by potatolicious
16 hours ago
> "Why does having launched in other cities matter if the new city brings up things that none of the other launched cities do?"
New requirements come up all the time in technology. The existence of a new requirement isn't in and of itself justification for skepticism - is there a particular reason to believe that Waymo is not capable of solving for the new requirement?
The answer may be yes, but simply "ahah! It would need to do [new thing]!" is insufficient. "[new thing] is likely intractable because [reason]" would be more justification for skepticism.
> "It's my understanding that self-driving cars don't really account their acceleration and braking for roads that could sometimes be very slippery due to snow and ice."
Sure, but like above - is there a reason this is an intractable problem?
I'll throw this out there: your human-driven car already accounts for acceleration and braking on slippery roads, without the need for the human. Traction control systems and electronic stability control systems exist! They're in fact incredibly common on modern cars.
The interesting snow & ice problem for me is that humans will drive in winter conditions that are unsafe -- for example white-out blizzards. Robocars won't be able to drive in a white-out blizzard, so they'll likely refuse to do so. Humans should also refuse to drive, but people drive anyways.
NYC doesn't generally get white-out blizzards, so refusing to drive in them is quite feasible.
> Robocars won't be able to drive in a white-out
My Subaru can lane keep in Wyoming blizzards better than I can because it follows the car in front with radar.
Also snow and ice in NYC is a rare event now, not a given like it used to be.
I come from way up top on that globe of ours. I have driven in frankly apocalyptic snowstorms. They're an insidious problem to solve, but I remain optimistic. Back home, they will close specific roads due to snowstorms, but what do you do about the cars already on the road? You can't stay put for 16 hours can you? So you move as slow as possible, sometimes as low as 5 kilometres an hour. Cause that's the thing about a snowstorm; it's about visibility. You're not risking your life if a dude in skis can go faster than you.
> You're not risking your life if a dude in skis can go faster than you.
Sure you are. You can still drive off the road and into the ditch where nobody can see you. People then die because they don't clear their tail pipe and get carbon monoxide poisoning or they try and walk for help and freeze to death.
If white-out visibility is the only problem to be addressed then machines seem pretty well placed because they can use very accurate positioning and non-visible light sensors. Unfortunately they probably wouldn't know that there's a 50 yard section of the road that always drifts in when the wind comes from the south and the snow is dry.
>I'll throw this out there: your human-driven car already accounts for acceleration and braking on slippery roads, without the need for the human. Traction control systems and electronic stability control systems exist! They're in fact incredibly common on modern cars.
These systems don't help with the problems I am talking about.
You have to drive completely differently in heavy snow, significantly slower, brake sooner, turn less sharp, accelerate much slower, leave significantly larger gaps, leave space to move out of the way and be ready to move if someone behind you is coming at you too fast and can't stop in time, etc. I've spend my entire life in the midwest.
The traction control system in my 2023 camry didn't help one bit when I applied the brakes on black ice and the car didn't react at all, it just kept sliding at the same speed across the ice.
That all sounds like something that should be easier for a robot to do than the typical human. If programmed for how to drive in heavy snow, a robot should be able to switch driving modes much easier than the typical human brain.
Waymo has been trained in Buffalo NY for winter conditions, unlike most NYC drivers.
Is it possible to train a machine to drive in snow? Yes. But consider that humans are trained to do so by means of things like: actually crashing, observing others crashing, talking to people who crashed, and all of the above is highly localized. Where I live there are many days in winter when someone not from the immediate area should not drive at all. But I might if there was a good reason because I have 25 years experience with the specific roads, conditions, how those conditions relate to wind and on and on. Training a machine to know all that seems feasible but unlikely to be commercially viable. It's just not a problem that can be solved with a simple closed loop control system like ABS or traction control.
1 reply →