How Robotaxis Might Mitigate Electric Car Depreciation (robohub.org)

Hallie Siegel writes: Autonomous car expert Brad Templeton argues that we're in for a period of about 5 years in electric cars where each year's new model is a lot better, and that could be a problem for people trying to sell them. Further exacerbating Moore's Law for cars is that autonomous features (like traffic jam assist) rely heavily on computers. Unfortunately cars cost a lot more than computers or cell phones, so throwing them away before the end of their lifespan is a bit of a problem. How do get over the depreciation problem while autonomous cars and electric cars are going through this period of rapid development? Templeton suggests that a taxi model could be the answer, since use is so much more intense that with a private ownership model, that the cars are likely to wear out before they become worthless from a resale perspective.

Comments

Well

[DrYak]

I am sure people will protest "but Google has a car that works now!". No. They really don't. They have a car that can navigate through heavily pre-mapped city streets.

Well, not exactly. If you pay attention to Google latest TED talks:
- that used to be the case at beginning.
- but even heavily pre-maped city streets aren't perfect: new construction works that haven't been mapped, new pot hole, etc.
- thus this ultra-high detail maps would need to be constantly update with very tiny details (which would have been a very fastidious and expensive work if actually done).
- one could better automatically crowd-source the updates: cars automatically submits them as they encounter them.
- this requires automatic interpretation of the data (to avoid overloading the human curator with constant influx of data) and also requires a bunch of very high detail sensors on each car.
- if you have the sensors and the automatic processing power, you might as well decide to directly feed the gathered data to car's autonomous driving system, instead of having to wait for a whole rountrip to the mother ship and the data coming back a map update.

In other ways:
- efforts and technological challenge to keep ultra the insane-level-of-detail maps as up-to-date as possible ~= efforts and techonological challenge to have the car aware of its environment and properly react on changes like a new construction work.

So in fact, this can work with high quality maps similar to high-range GPS devices (the kind that can make distinction between lanes, and knows which lanes (are supposed) to lead where), combined with a fuckton of sensors and a skynet-worth of signal processing and interpretation.

In lay terms: you don't actually need a map of all the pot-holes, when you need a technology able to autonomously detect them to update pot-hole map and keep it as close to the reality as possible. You can as well skip the middle step and directly use your pot-hole detecting equipement to avoid the pot-holes, "map free".

So ...

They have a car that can navigate through heavily pre-mapped city streets.

To be more correct: Google has car that *used to* rely on heavily pre-mapped city street maps. Google has now a car that might not get and manage to react to unexpected situation, in the small subset of all things that the cars has seen up until now and was programmed to handle.

That means that the car already can react in some really weird and unexpected situation, some times bordering on surrealism:
- The TED presentation showed the car correctly detecting a cyclist coming out of simingly no where and cycle completely across a road crossing, passing in front of cars, right just when the traffic lights turn green. The google car stayed and waited the cycling to pass along his erratic part. That wasn't the case of the meat-controlled cars nearby (luckily none ran over the cyclist).
- The TED presentation showed the car correctly reacting to a lady on a motorized wheelchair, chasing a duck in circles in the middle of a street. (again both the grand ma and duck end up safe).

That also means that the car has NOT being tested and will probably fail (not gracefully) on tons of other situations which weren't accounted for and designed-in yet.
- snow is often given as an exemple. Google Cars are currently being tested in California, and have clocked KMs of experience in situation where the street markings are not visible (e.g.: due to weather).

That's also the argument behind Mercedes Benz current research in autonomous cars: they are trying to test their cars in a as many different conditions as possible accross whole Europe.

In short: Google Cars aren't ready yet. But it's not because we need more detailled map of all potholes, but because the cars are only able to see and detect the potholes by themselves, they might be completely oblivious to snowmen as far as we know, and we still need to test (and develop) to that.

But the idea of using electric cars for human-driven taxis is a good one. In fact that is one of the reasons why so many taxis are hybrids already.

Same here around.

Not that much

[DrYak]

That is because of dealer markup.

Not quite. The dealer don't have actually that high margins on new cars.

It's really the market price falling with the perception of the cars being in pristine conditions or not.
(It's the same kind of consideration as that to a fan or a collector, it's REALLY important if a good was unboxed / if the packaging has already been cut open or not.)

On the other hand: you can count the dealers to milk your wallet as much as they can get with it. Specially for periodic controls. But also at the slightest warning light going orange / the slightest warning message on the dashboard. Or even for simple firmware update. Or even if the offered services don't make any sense (e.g.: car dealer selling care packages including a few oil changes. On an electric vehicle. (which has a sealed electric motor and thus no possibility to touch the oil).
That's why car dealers are pissed by dealership-free cars companies like Tesla Motors (With automatic OTA update that don't require a "technician" and are offered for free instead of being charged. And with fewer parts requiring actual service)