Comment by ryukoposting
3 days ago
SV is the most cloistered place I've ever seen. I'm comfortable assuming that nobody in any position of power at Waymo ever thought to themselves "gee maybe the weather is different in this new city we're deploying to, perhaps we should test that"
As a former engineer and manager at Waymo I can say with the confidence and sincerity of firsthand experience that this is not the case. People at all levels of the company think deeply about how different locations have present different challenges, including different weather.
Also it's not like we never have flooded roads here in the valley.
Whatever is going on, I'm confident it's not a result of straightforward parochialism in the way that you say you're comfortable assuming.
How many years for Waymo to work in Mumbai?
I worked in the field, not at Waymo. Everyone in the business is acutely aware of weather, along with hundreds of other factors, many much less obvious.
The engineers whose expertise you assume away are actually debating corner cases like the one we saw of someone carrying a bicycle on their skateboard.
In fact the companies run test campaigns in shitty weather all over the country on purpose, at great expense.
Yeah maybe we should just stop doing that and invest in public transit infrastructure instead.
Public transit is a function of city design, less so much the presence of public transit. If you can’t walk to a stop, or if your destination isn’t reasonably accessible from that network, it won’t be used for that trip.
While it sucks for many other reasons, autonomous vehicles are actually a very good solution to public transit in most American cities. What I envisage is a dense grid of virtual bus stops in N square miles surrounding a rapid transit stop. You hail using an app, and a minibus (8-20 pax) adjusts its route to collect you and get you to that rapid transit station. The inverse happens for people arriving at that station, where routes are planned as the train approaches, so people heading to the same general area can be directed to the same minibus.
Last mile is still a thing. We need long distance public transit, regional public transit, local public transit (buses, trams, cable cars, ...) and we also need hyperlocal public transit (taxis, autonomous vehicles/"peoplemovers").
Who is "we?" The cost to develop self driving cars is not exactly being felt by society at large.
It certainly isn't stopping anyone from improving public transit, but it seems like you believe it's this and not any one of a bajillion actual factors to blame.
if you want people to use public transit, you need to make it not be a mobile homeless shelter. otherwise everyone who can afford to will insist on a private transport
1 reply →
Roads are public transit infrastructure.
These are companies. They can invest where they please.
Call your government reps.
3 replies →
True. While we're at it, let's not fixing roads as well. Also electric cars. Also what's the deal with space exploration? Fix what's on earth first please.
There's a long tail of unpredictable events in the AV industry that you end up seeing, especially since the cars in aggregate end up driving more than one could over a lifetime.
At a previous employer, we've seen anything from cars getting mooned, a SUV slowly driving past the AV, the rear window roll down, and someone poke their head out and start throwing dollar bills at the AV, a convention of people dressed up in animal costumes, the "Miami left," and so on.
So it's much less of "maybe we should test that" and more of "we don't know what we don't know, so let's gather some data." In practice, the cars have lidar so they won't crash into solid objects that aren't recognized, they just end up getting stuck in embarrassing situations like these.
I used to work for an AV startup.
One of my favourite things to see were the random encounters that our data annotators would flag up.
Unusual agricultural vehicles, large to-scale images on the sides of vehicles, cars facing backwards being carried by a vehicle transporter.
It's a wildly long tail of things that automated vehicles need to handle.
A flooded road is a very predictable event, though.
Is it? I have been driving for 25 years and never encountered one.
Waymo seems to accept they can’t predict everything so they built a system that’s safe enough to operate in the real world and learn from experience.
5 replies →
My guess is this was brought up but getting the product out there was more important to the business so it got ignored.
Now that it's a problem for them, they get to hide behind an "oops sorry, let's fix the really obvious thing now", almost like taking "if it ain't broke, don't fix it" to malicious levels.
This jives with CRUD software in general, where people are not usually rewarded for preventing future issues and instead rewarded for waiting until it's a visible problem and then fixing it.
This seems silly -- they roll the service out to individual cities in different regions, one at a time. Why do you think they do that? I'm pretty sure this is exactly that testing that you're referring to.
Surely Waymo can afford a test track.
They can, and I bet they have! But they cannot afford a test track that accurately reproduces every condition exactly as it will be encountered in the real world. At some point, it is judicious to test with real-world conditions, and simulating only gets you so far.