14
votes
Waymo outsources fleet operations to African fintech Moove in Phoenix and, soon, Miami
Link information
This data is scraped automatically and may be incorrect.
- Authors
- Rebecca Bellan, Tage Kene-Okafor
- Published
- Dec 5 2024
- Word count
- 715 words
Anyone know if these first few cities are being chosen because they're easier to navigate? Tesla autopilot used to try to drive into the monorail pillars in Seattle, for example, so I'd expect to to be categorized as difficult. (They did eventually fix this.)
Miami is weird. Driving is a nightmare given the sheer number of cars and the crazy drivers, but I would bet that this sort of thing is only going to be in very limited areas for the foreseeable future where high-paying tourists are likely to be (Miami Beach mainly).
In terms of navigation: streets are wide, no elevation change, and the whole city is a giant grid. It's honestly one of the easiest cities for humans to intuitively navigate, but I'm not sure if that translates well for self-driving cars since they rely on GPS. But being a completely flat massive grid probably lends itself well to self-driving cars. Plus you don't ever have to worry about snow or ice.
Though I'm sure there are other cities that could work. Miami is probably on the list because of the money and because public transportation there is laughable at best, so an alternative to that might actually tempt quite a few people.
edit: "Waymo would not share which neighborhoods the company is targeting in Miami" 100% it's going to be Miami Beach, Brickell, Coral Gables, maybe Downtown too. And then a few of the areas in between or with significant recent development like Wynwood, Dadeland, or Midtown.
I don’t know for sure, but I’d be surprised if any given US city is meaningfully more difficult than another when you’re training the system to just that area, so I’d suspect regulatory environment and good old profit projections have more influence on the decision.
Situations like that autopilot glitch tend to be from general case systems getting thrown off by a situation they don’t recognise, and Waymo’s got the advantage of any given vehicle being limited to a specific area, so they should have pretty much 100% coverage for training data. Failure cases shift towards dynamic situations that could happen anywhere (construction, accidents, people behaving unexpectedly) rather than features of the city itself.
Weather conditions for one can make a huge difference, most cities they have been trained in have fairly stable weather. There are also plenty of other issues with self driving cars. Not all of them are technical in nature, here is a previous discussion on Tildes
Weather's a good example of a dynamic situation that does depend significantly on location, actually - I was thinking of differences in fixed geography and architecture when it comes to navigation, but yeah it'd be totally fair to throw weather into the bucket as a feature of the city rather than just an external event.
The rest is... a much bigger topic that I honestly don't have the energy to get into as deeply it deserves. I very much see the problems that self driving cars could (and probably will) cause, but as with most automation I see them primarily as the same problems we've already got but bigger and faster. It's a catch 22: if we the had the organisation and political will to block the use of a given technology, that power would be better wielded by regulating in such a way that the technology actually benefits the community. If we don't have that power, it's already a foregone conclusion. Either way, I see it as a political failure more than anything.
That's fair :) I honestly feel the same way, mostly because it has been a recurring topic on a fairly regular basis on Tildes recently.
I wouldn’t say Seattle is particularly more difficult than SF.
A lot of these companies train their systems in the Bay Area. Having driven in both, I’d say they’re equally difficult, but with different difficulties.
Speaking of snow, I’d expect the Duluth/Superior area to be one of the last chosen. Hills, cobblestones, regular heavy snowfall and ice, and they used to have to declare a state of emergency most years so they could be allowed to use more salt on the roads. (I believe they’ve switched to something else.)
From the article:
…
Also, Waymo previously announced expansion to Austin and Atlanta. For some reason it will only be via the Uber app in those cities.
Seems like Waymo is trying to transition to be more of a service provider. They provide the self driving software, another company handles day-to-day operations like fetching stuck cars or handling customer funds. It’s a smart pivot.
Uber gets the contract for Atlanta and Austin, it seems.
See, to me it just seems like a way of offloading the blame when the cracks in the system get worse over time. "You just bought our software, blame Moove for running it badly!"
And then Moove can blame Waymo.