[mcrady]

I also had a magical autonomous ride when I visited Phoenix a couple years ago.

I asked a friend who works at waymo why it's taking forever to build up a critical mass of vehicles to get the wait times down to a more reasonable level.

He said the issue was cost. I guess all those sensors are actually more expensive than the cost of a human driver at this point?

[hwillis]

> I guess all those sensors are actually more expensive than the cost of a human driver at this point?

Waymo inhoused their sensors and claimed a 90% cost reduction, but before that they were at $200k+.

Still, even at $200k+ that's not terribly high compared to a human driver once you amortize it over 8+ years. Higher than minimum wage, but really it's probably a matter of the upfront cost more than anything.

I'll also point out that even if they 100% solved self driving... somebody has to fill the things up with gas in most states, lol.

[vilhelm_s]

The New Jersey and Oregon policy of not allowing self-service gas turns out to be the visionary policy for the future of transportation!

[tomjakubowski]

Getting off topic now but Oregon does allow self-serve gas in some places now. I've pumped my own gas at the Chevron on Tomahawk Island, Portland.

[voxadam]

Oregon altered or suspended some rules regarding self-service early during the pandemic.

https://www.oregon.gov/osp/programs/sfm/Pages/Self-Service-R...

[ugh123]

Or they use electric vehicles and the cars just park on top of charging pads. But there's always some kind of maintenance in involved

[hwillis]

I don't think wireless EV charging will really take off- it's always going to be MUCH slower because of the I^2 losses. It's always going to be much more expensive and heavier than a plug. Solutions like battery swaps[1] or even just automatic plugs will win out at the large scale.

It really only has anything to offer to individual consumers, specifically saving the seconds it takes to just plug in your car. That seems like an insane trade to make, to me. Maybe in theory some city could invest in embedding them into street parking, but I just don't think that will happen. Even if vandals started cutting plugs or something, you can buy a LOT of replacement cables for the price of one wireless charging pad.

[1] https://www.wired.com/story/china-ev-infrastructure-charging...

[JumpCrisscross]

> the issue was cost

It has to be deeper. Google has deep pockets. Even it doesn’t want to touch them, one could sell interests in individual cars’ cash flows, a quasi-debt which neither dilutes the company’s equity nor puts its survival at risk.

[toast0]

They're spending a bunch of money on this already. Deep pockets doesn't mean infinite pockets. Google Fiber over promised, and the infinite pockets evaporated.

It makes a lot of sense to roll something like this out slowly. Taxi service is a commodity, putting out a bunch of cars that are have negative ROI doesn't get you much over having a few cars.

[UncleMeat]

Google has a shitload of money. But it also has 150,000+ employees. If everybody has loose pockets it becomes a disaster. Supply chain is also a problem. Even if a single piece of hardware is cheap, buying a gazillion of them is hard.

I find resource planning to be difficult and frustrating here at Google, despite billions and billions in the bank.

[judge2020]

High quality LIDAR suitable for autonomy is still pretty expensive, and it’s tough to take claims of “the price is coming down” at face value.

[BurningFrog]

I think they're still in learning mode.

You don't want huge numbers of cars on the road until you're sure they're incredibly safe.

[i_am_proteus]

Profit scales, but do does risk.

An improperly-rolled-out software update could kill a lot of people, including people not riding in Waymo vehicles. That bug could be in every Waymo vehicle on the road right now, waiting for the environmental conditions that make it turn deadly.

Slow is safe, safe is smooth, smooth is fast.

[zizee]

Is it possible that he was talking about a different cost than that of vehicle parts? Even at a million dollars a piece, I'd imagine google would be happy to spend a hundred million or two to get dozens/hundreds of higher visible vehicles out in aive environment.

Perhaps the expense he was talking about is that it would be super expensive to create a factory to be able to increase the volume whilst the vehicles are still rather custom built.

I will say that I have very little knowledge about anything here, just that Google has already spent billions on this effort and it would seem strange that they are shying away from spending a few hundred thousand per vehicle.