Also relevant, when it rains in Miami during the summer, it pours.
As in, zero visibility for 15-30 minutes, then it's past.
So if it can handle Miami tropical rain, it should be okay with all sorts of normal rain.
Out of curiosity, what's Waymo's current production sensor suite mix? I'd assume lidar and radar would also be very unhappy with the surrounding space suddenly being ~10%(?) liquid water droplets.
Far less than 10%. During a heavy downpour, by volume, about one part in a million is liquid. In a cubic meter of heavily rain, there are only a few tens of raindrops.
Not just visibility in the rain, but diminished or fully obstructed visibility due to ponding and full flooding. Then there are the physical navigational problems associated with that. They probably shouldn't be driving though a foot of sea water.
> So if it can handle Miami tropical rain, it should be okay with all sorts of normal rain.
I feel like a lot of "How well does it handle rain?" comes down to how the roads are built and maintained (Huge puddles, proper drainage, etc) rather than about the car itself, as the car you could test by blasting it with water from different directions and amounts.
I would expect service to be canceled while it is pouring down. Do we have reasons to believe that they have the ability to ride safely during heavy rain? I haven't been keeping up.
Interestingly is there a potential moral issue in the making here? What happens if/when self driving dependency is so prevalent that the majority of inhabitants in a city don't know how to drive. In addition add the fact that self driving cars don't have a steering wheel so even people who know how to take over driving can't actually take over.
What happens if there's an event that requires a mass evacuation such as a Category 5 hurricane and the major self driving car companies deem it too risky to drive in the conditions that precede the storm?
Plenty of south florida rain is not helped by windshield wipers. Anyway I wonder if waymo sensors actually have better visibility in such conditions than people do
I'm guessing you've never tried driving in a tropical rainstorm. It's as bad a driving in heavy fog. Sometimes you only really have visibility of a few feet.
When I took Waymo in Phoenix, I booked a ride from a suburban hotel to a restaurant in a strip mall. One of the things I noticed was that I was picked up far away from the entrance of the hotel (eg, not under the overhang that protects from sun and rain, where every Uber has picked me up and dropped me off). I recall thinking that it was good there was no weather in Phoenix b/c I had to walk far enough I'd have gotten soaked in a decent rainstorm.
Also Waymo will not pick you up on private streets. I live in a small community with a private street and I have to walk to the nearest public one (2 mins).
One thing I've noticed about the SF deployment is that it's slowly gotten better at this. At first it was very cautious about where it would pick you up/drop you off, but now it offers much closer options (from a menu -- a bit like Uber at airports).
I suspect this might be something that is human-added from data collected in past trips.
Good point! Though also worth mentioning that they're already in Atlanta, which gets ~50in (and 59in so far this year, despite the mind-bending "first October without rain in recorded history")
Good point, I guess I missed when that happened. Looking at some news sites and Waymo's blog, it seems that they are testing in Atlanta and will start accepting customers in 2025.[0]
>It currently operates fleets of driverless cars in San Francisco, Los Angeles, Austin, and Phoenix. It also plans to launch a robotaxi service in Atlanta in an exclusive partnership with Uber.[1]
The rainfall can pose serious visibility risks that will be as much of a challenge as picking up and dropping off customers on a rainy day. Extreme high tides do still flood some roads on Miami Beach with brackish water, which isn't something you want to drive through in an electric car.
On the less challenging side, the city has zero snow, no road ice to worry about.
I think the most interesting part is that the article says that Waymo's handing its operations to Moove. It seems like Waymo's trying to become a software provider while having other companies handle the capital-intensive parts.
> having other companies handle the capital-intensive parts.
Waymo definitely wants to outsource the areas where they don't have special expertise (i.e. Waymo is 100x better at driving, but not 100x better at washing and vacuuming cars). I'm not sure how capital-intensive regional operations are. The vehicles are definitely the largest capital expense. This is more like an AirBnB property owner hiring a cleaning service.
Also, contracting out the menial labor makes Waymo's labor practices look much better. They can tell their engineers that all employees make a living wage and get excellent health insurance.
When the actual labor is done by part-timers with no health insurance making not much over minimum wage.
The pivot has already happened. They’re handing over Austin and Atlanta to Uber, and now Phoenix and Miami to Moove. The only places they will continue to own operations for at least the next year are SF and LA.
Capital-intensive, or labour-intensive? If I were a provider of 'special smart sauce' that goes on a common piece of equipment, I'd be trying to focus on making it so I could provide the sauce rather than dealing with all the real-world issues that come with all the real-world people using the saucy equipment.
I design motherboards for industrial computers for living. Last gem: radio module draws 5 amps while transmitting instead of specified 2 amps. Trust nobody!
This makes sense. If they don't outsource, they need to run millions of cars. This will cost Alphabet hundreds of billions capex, which is not cheap even for them. This is not just the money problem, but also has significant implications on their speed of business expansion. Let's say Google decides to pour tens of billions every year on Waymo, it will takes tens of years to expand into all of the major US cities. They probably don't want to give the competitors that much time.
This seems much more scaleable. Car share services (eg. Evo in Vancouver) seem like good partners as they already have the fleet management services and a recognizable (and hopefully trusted) brand.
I'm not sure about other car share services work, but in the case of Evo they have existing relationships with the cities that make up Metro Vancouver. I wonder if this would ease rollout as you'd already know all the required people to talk to within municipal government?
our service – which already provides over 150,000
trips per week across Phoenix, Los Angeles, San
Francisco, and Austin
Interesting. That's about 8 million rides per year.
I wonder how close they are to being profitable? As soon as they are getting close to being profitable, they will probably scale this up super fast.
I don't know how much Google invested into Waymo so far. Something like $10B?
If they at some point make $10 per ride, they would only need something like 50 million rides per year to justify that investment with a p/e ratio of 20.
To go from 8M rides to 50M in 5 years they would have to increase their capacity by 50% per year. Might be possible?
How many rides are there every day in developed countries?
Their internal business case probably has them targeting not 50 million rides per year, but per week… at an absolute minimum
Regardless; at some point specialised vehicles will be developed which are ultra small and lightweight - less than $1,000 to produce - to take care of short downtown rides, for example.
Profitable from an operations perspective? Surely close since they charge the same order of magnitude as an Uber/Lyft and have fewer than one driver per vehicle (monitoring the vehicles).
Google has invested and also drawn external funding. From what I've seen in the last 15 years since founding the lower bound for their cost seems to be higher than $12bn and I can only imagine expenses will only accelerate.
Assuming they don't license the technology to everybody else and keep running their own cars. How much is this technology worth to every taxi company, every car manufacturer, every fleet operator in the world?
I imagine it's hard to quantify "profit" with such a research-driven org. It's like penciling out the profitability of the metaverse after years of $XX billion dollar losses. In general I get the sense that Waymo is more of a diverse investment than a pure ride-hailing play; for example, as of 2020 they were working for Volvo, Chrysler, Jaguar, and Nissan[1], presumably for $$$.
It's also worth remembering that Zoox exists (Amazon's more futuristic self-driving car play, no steering wheel at all), and has not at all gone the way of Alexa/the Dodo bird (yet). I expect them to make a big splash sometime in the coming decade, personally.
That is, of course, assuming they survive regulatory capture by Tesla, which would need a miracle or an unfair advantage to beat these two at this point, even if they finally follow the science on the need for LiDAR. Another big unknown is how the electorate will react to self-driving cars becoming more than a novelty; Elon Musk is absolutely correct that a backlash of some kind is inevitable even if the safety stats pencil out, IMHO. Trusting a machine is kind of inherently creepy - see Prof. Weinersmith's lectures on the topic:
There were three things in Waymo's latest CPUC filing that interested me. First, through the end of August, Los Angeles still irrelevant. Over 85% of their California rides were still in SF. Second, ridership in SF doubled in 90 days without a significant expansion in either vehicles or trips per service hour, but they had the cars out on the road more hours every day. Third, the geographic concentration of their rides is extreme, with a large fraction of trips starting near either the Ferry Building or Fisherman's Wharf, which suggests that it is popular with and useful to tourists.
Anyone know the timeline for Waymo expanding to northern areas outside urban centers? Or are these underserved populations forever stuck waiting hours for an Uber that still cancels half the time?
It's got to be the lowest priority; population density improves economics and utilization. Not to mention that it's hard enough to drive in a city with snow, compared to all the other kinds of situations that can manifest outside of urban centers.
At the end of the day that isn’t a technical problem but a unit economics one.
Removing the driver from a taxi doesn’t bring down costs that much. Self driving cars aren’t going to change the uber/taxi model at a fundamental level.
They have a finite fleet that they need to deploy. Urban centers mean that fleet utilization is high, and relatively less time and miles are spent driving with no one on board. In rural areas with little demand they will sit empty or have to drive empty for many more miles to their pickups. It just isn’t profitable to use your fleet that way no matter what you do.
I think ultimately the solution to this problem is the same as it was for electrification and telco: government funded mandates to provide service to populations where it's otherwise uneconomical.
One interesting thing today is that CoL can be as high in rural areas as urban areas in the same state, partly because the additional costs of things that don't scale (mostly transportation and healthcare). But we've given up on government helping people, apparently.
SF in 2015 (with passengers 2021), Phoenix in 2022, Miami in 2025. Northern urban centers are probably a decade out, let alone areas outside urban centers. There are a lot of cities in the Sun Belt to expand to first.
I'm so curious about how the internal dev teams feeling about all this scaling. Four cities across 3 states -- surely there are differences in road signs, lane markings, emergency procedures, etc. Let alone the sheer volume of data of doing hundreds of thousands of miles ever week!
Massive kudos to them if they are able to do all this without things being aflame on the inside...
Only they know for sure where the bottleneck is. How much mapping do they really need before their software can drive it. How cautious are they being? How slowed down by the capital cost of cars and needing to set up maintenance depot's are they? How much is held up by legal approval in each jurisdiction? Knowing Google, the software is rock solid, it's the rest of everything that's taking forever.
My understanding is they have to make extremely detailed maps of where they operate and at great expense, and everything sort of breaks when anything changes. So lots of work indeed!
1. Literally nobody in Florida can drive. Nobody indicates. People run red lights. They speed on the hard shoulder to overtake someone else who is speeding slightly less;
2. There's a lot of things that come down to timing, like when the bridges are up on the Venetian and over the Miami River. You can also get trains blocking the entire of downtown;
3. It seems like there's constant rerouting for closed roads, typically due to contruction;
4. Inclement weather. High winds and flooding. Biscayne Boulevard is often called Lake Biscayne. 30 minutes can be the difference between Miami Beach being dry and every road being 1 foot deep in water (not an exaggeration); and
5. What will be the covered area? I guess Phoenix and LA sprawl too but what constitutes "Miami" goes south, west and north pretty far. I mean there's no break between Miami, unincorporated Miami-Dade County and Fort Lauterdale.
Yeah, I dunno if the team understands just how crazy Miami driving is. Maybe they'll restrict it to Downtown, Miami Beach, the Grove, etc to limit the chaos?
That won't limit the chaos. FWIW I went to whole foods downtown today and nearly got hit twice. I live in Midtown, so it's quite literally a 27-block trip. Some of the worst driving (save for on 95) I've seen here has been either 1. idiots on the Beach 2. idiots in Brickell or 3. idiots in downtown.
Did you know that almost every city believes this about their local drivers? I've gotten it in LA, Boston, NYC, DC, SF, Philly, Atlanta, and Austin. Adding Miami to the list.
Having lived in most of those cities and in Miami I can say Miami is definitely the most dangerous driving I've seen. And my car insurance company certainly seems to agree.
Yes, that’s true, but in this case they’re all actually wrong. I’ve driven or lived in all of those cities, and they’re all placid next to Miami. It’s not even close.
LA Drivers aren't that bad, driving just is awful there because the amount of traffic. NYC Driving is an absolute pleasure compared to Miami driving. I have what should be an easy ten minute commute and every day I am avoiding an accident due to a driver doing something crazy you would almost never see in another US city.
This is exacerbated by the dysfunctional government which is happy to let developers do whatever they want without regard to impact on traffic flow, while doing no investment in infrastructure itself. I'm generally pro-growth, and I think California goes too far with its restrictions, but living in Miami has caused me to gain some appreciation for the reason behind some of what California does.
In my experience it is a State wide problem. I95 seemed to be a speed minimum of 95 if you didn’t want to get run off the road by everyone else for going too slow. I-4 (Daytona to Tampa, through Orlando) had over 400 deaths caused by traffic accidents one year. More than 1 death per mile of road. I don’t think 75 was any better, of course it was rough in Atlanta too, but still.
My experience with Florida driving was not a great experience for the few years I lived there.
Readit News
Miami: 57 in. (https://www.ncei.noaa.gov/access/us-climate-normals/#dataset...
SF: 25 in. (https://www.ncei.noaa.gov/access/us-climate-normals/#dataset...)
Phoenix: 7 in. (https://www.ncei.noaa.gov/access/us-climate-normals/#dataset...)
As in, zero visibility for 15-30 minutes, then it's past.
So if it can handle Miami tropical rain, it should be okay with all sorts of normal rain.
Out of curiosity, what's Waymo's current production sensor suite mix? I'd assume lidar and radar would also be very unhappy with the surrounding space suddenly being ~10%(?) liquid water droplets.
[1] https://ouster.com/insights/blog/lidar-vs-camera-comparison-...
I feel like a lot of "How well does it handle rain?" comes down to how the roads are built and maintained (Huge puddles, proper drainage, etc) rather than about the car itself, as the car you could test by blasting it with water from different directions and amounts.
What happens if there's an event that requires a mass evacuation such as a Category 5 hurricane and the major self driving car companies deem it too risky to drive in the conditions that precede the storm?
The jaguar i-pace does this independent of the waymo use case.
Have they changed this?
I suspect this might be something that is human-added from data collected in past trips.
>It currently operates fleets of driverless cars in San Francisco, Los Angeles, Austin, and Phoenix. It also plans to launch a robotaxi service in Atlanta in an exclusive partnership with Uber.[1]
[0] https://waymo.com/blog/2024/09/waymo-and-uber-expand-partner...
[1] https://www.theverge.com/2024/12/5/24313346/waymo-miami-robo...
On the less challenging side, the city has zero snow, no road ice to worry about.
Waymo definitely wants to outsource the areas where they don't have special expertise (i.e. Waymo is 100x better at driving, but not 100x better at washing and vacuuming cars). I'm not sure how capital-intensive regional operations are. The vehicles are definitely the largest capital expense. This is more like an AirBnB property owner hiring a cleaning service.
When the actual labor is done by part-timers with no health insurance making not much over minimum wage.
Chick-fil-A grew into a pretty big business by vertically integrating outside of just selling sandwiches to Waffle House.
So sometimes it's worth owning sauce distribution too. ;)
Mother nature OS is by far the worst to develop for.
I design motherboards for industrial computers for living. Last gem: radio module draws 5 amps while transmitting instead of specified 2 amps. Trust nobody!
I'm not sure about other car share services work, but in the case of Evo they have existing relationships with the cities that make up Metro Vancouver. I wonder if this would ease rollout as you'd already know all the required people to talk to within municipal government?
I wonder how close they are to being profitable? As soon as they are getting close to being profitable, they will probably scale this up super fast.
I don't know how much Google invested into Waymo so far. Something like $10B?
If they at some point make $10 per ride, they would only need something like 50 million rides per year to justify that investment with a p/e ratio of 20.
To go from 8M rides to 50M in 5 years they would have to increase their capacity by 50% per year. Might be possible?
Their internal business case probably has them targeting not 50 million rides per year, but per week… at an absolute minimum
Regardless; at some point specialised vehicles will be developed which are ultra small and lightweight - less than $1,000 to produce - to take care of short downtown rides, for example.
It’s going to be a wild world.
Plus I guess they need high resolution maps? Not sure if that is a significant cost factor.
It's also worth remembering that Zoox exists (Amazon's more futuristic self-driving car play, no steering wheel at all), and has not at all gone the way of Alexa/the Dodo bird (yet). I expect them to make a big splash sometime in the coming decade, personally.
That is, of course, assuming they survive regulatory capture by Tesla, which would need a miracle or an unfair advantage to beat these two at this point, even if they finally follow the science on the need for LiDAR. Another big unknown is how the electorate will react to self-driving cars becoming more than a novelty; Elon Musk is absolutely correct that a backlash of some kind is inevitable even if the safety stats pencil out, IMHO. Trusting a machine is kind of inherently creepy - see Prof. Weinersmith's lectures on the topic:
- https://www.smbc-comics.com/comic/decisions
- https://www.smbc-comics.com/comic/fsd
- https://www.smbc-comics.com/comic/self-driving-car-ethics
[1] https://www.forbes.com/sites/davidsilver/2020/06/29/waymo-an...
Dead Comment
Once they can do that -- and (I guess) can prove profitable -- they could expand non-stop across the country.
Removing the driver from a taxi doesn’t bring down costs that much. Self driving cars aren’t going to change the uber/taxi model at a fundamental level.
They have a finite fleet that they need to deploy. Urban centers mean that fleet utilization is high, and relatively less time and miles are spent driving with no one on board. In rural areas with little demand they will sit empty or have to drive empty for many more miles to their pickups. It just isn’t profitable to use your fleet that way no matter what you do.
One interesting thing today is that CoL can be as high in rural areas as urban areas in the same state, partly because the additional costs of things that don't scale (mostly transportation and healthcare). But we've given up on government helping people, apparently.
Massive kudos to them if they are able to do all this without things being aflame on the inside...
This is actually a scale up to five cities across four states:
> ...which already provides over 150,000 trips per week across Phoenix, Los Angeles, San Francisco, and Austin
Which of course only adds to your point!
And it's just a matter of the cars driving through each of the streets and working with local authorities.
What is “your understanding” based on?
1. Literally nobody in Florida can drive. Nobody indicates. People run red lights. They speed on the hard shoulder to overtake someone else who is speeding slightly less;
2. There's a lot of things that come down to timing, like when the bridges are up on the Venetian and over the Miami River. You can also get trains blocking the entire of downtown;
3. It seems like there's constant rerouting for closed roads, typically due to contruction;
4. Inclement weather. High winds and flooding. Biscayne Boulevard is often called Lake Biscayne. 30 minutes can be the difference between Miami Beach being dry and every road being 1 foot deep in water (not an exaggeration); and
5. What will be the covered area? I guess Phoenix and LA sprawl too but what constitutes "Miami" goes south, west and north pretty far. I mean there's no break between Miami, unincorporated Miami-Dade County and Fort Lauterdale.
People here suck at driving.
This is exacerbated by the dysfunctional government which is happy to let developers do whatever they want without regard to impact on traffic flow, while doing no investment in infrastructure itself. I'm generally pro-growth, and I think California goes too far with its restrictions, but living in Miami has caused me to gain some appreciation for the reason behind some of what California does.
My experience with Florida driving was not a great experience for the few years I lived there.