Autopilot 2.0

[MarkS22]

I would say that the primary reason for new cars depreciating is that buyers won't pay nearly as much for a "used" car as they will for a brand new car.

I assume your point is that EV drivetrains should last much longer and be more reliable than ICE drivetrains. However most people don't understand that or are not considering buying an EV, so EVs are going to depreciate in a similar way that ICEs depreciate.

We're transitioning to a time where technology (including UI response, autonomy/sensors, safety features, and battery technology) will be depreciating much faster than the mechanical parts in the minds of consumers. Uncharted territory.

 

[Bimbels]

As you probably know, cost for retrofitting with Tesla have always been huge even when retrofitting has (rarely) been possible. It has been cheaper to sell old car and buy new than retrofit.

But if this AP 2.0 only needs additional camera, I see that reasonable priced retrofit could be possible. But on the other hand, it is better for Tesla, that you buy new car instead, so...

I would not be surprised if they did a free/reasonably priced retrofit, if possible, because of safety. Tesla is very safety-conscious and I see that is the number one reason a retrofit will be available and reasonable. Again, if it is possible.

 

[MP3Mike]

I would not be surprised if they did a free/reasonably priced retrofit, if possible, because of safety. Tesla is very safety-conscious and I see that is the number one reason a retrofit will be available and reasonable. Again, if it is possible.

Well of course it would be possible, WK057 proved it was even possible to retrofit AP1.0 onto a classic Model S. (However, it wasn't cheap or easy. Though I guess it would be much easier for him to do it again.)

 

[msnow]

Well of course it would be possible, WK057 proved it was even possible to retrofit AP1.0 onto a classic Model S. (However, it wasn't cheap or easy. Though I guess it would be much easier for him to do it again.)

Although Jason said he would never do it again. Too much work, too expensive.

 

[mrElbe]

Recognizing traffic lights is a huge challenge due to different shapes and positions. Some are vertical, some are horizontal. Some are at the far end of the intersection, some hang from a cable in the middle of the intersection. And then there is Europe where the lights are at the nearside of the intersection and the car first in line has them directly overhead.

 

[MP3Mike]

Recognizing traffic lights is a huge challenge due to different shapes and positions. Some are vertical, some are horizontal. Some are at the far end of the intersection, some hang from a cable in the middle of the intersection. And then there is Europe where the lights are at the nearside of the intersection and the car first in line has them directly overhead.

And some where the actual lights are around a "blind" corner so they add an additional light inline with your direction of travel and if you passed under it you would cross the traffic and run into trees on the other side. So you actually end up with the straight through light on both the left and right side of the left hand turn signal. (That confuses enough human drivers that I suspect it will cause a problem for most AP systems as well.)

 

[RogerHScott]

"You say it's going to happen now.
But when, exactly, do you mean?
See, I've already waited too long.
And all my hope is gone."

 

[calisnow]

I do not follow the resale threads nor have the premium subscription of ev-cpo, do you think they are not holding value? I thought they were holding way above 50% residual value after 3 years. Frankly I assumed tesla would break ground on this depreciation tactics and 50% after 3 years just baffles me, isn't the primary depreciation reason engine and drive train?

I don't really know what the data says. I was just pointing out that removing the resale value guarantee does not seem logical if the cars are indeed holding value well, because the RVG is a great selling point.

 

[schonelucht]

Losing the RVG is puzzling to me. If Tesla is confident about the value of the cars then the rational move is to keep the guarantee in place since it provides a benefit to the consumer and also costs Tesla nothing. The only reason to remove the guarantee is if in fact the cars are not holding resale. I think.

Remember that Elon is underwriting the guarantee personally. Maybe the volume of cars has simply become to big to continue to do so

 

[Chopr147]

IDK how the re-sale will hold up but there are always more people who can afford the used car rather than new. And with EV's coming in the next few years there just may be millions more wanting a Tesla w/o having to pay premium for a new one. The S could actually hold it's value more than any other car out there "IF" the EV era comes roaring in!

 

[schonelucht]

I don't really know what the data says. I was just pointing out that removing the resale value guarantee does not seem logical if the cars are indeed holding value well, because the RVG is a great selling point.

We know the cars sold 3 year ago with the guarantee held up. Tesla recognized $12M additional revenue from expirations (ie when the trade in value was above the guarantee) in 2015. What we don't know (and what is relevent for Tesla in deciding the guarantee going forward) is how the prices will hold up in 3 year. A 3yo Model S CPO will then compete with a brand new model 3 while today it effectively competes with nothing in that price range. More competition means lower prices...

 

[Zybd1201]

Recognizing traffic lights is a huge challenge due to different shapes and positions. Some are vertical, some are horizontal. Some are at the far end of the intersection, some hang from a cable in the middle of the intersection. And then there is Europe where the lights are at the nearside of the intersection and the car first in line has them directly overhead.

This is already solved and current hardware and software is able to read signs.

 

[supratachophobia]

If Tesla continues to behave in the future like they always have in the past there will be no upgrade path. I do not think they have ever released a "wiring harness" that was later upgradeable to something else - but maybe I'm mistaken?

Maybe this time will be different, I dunno. If I were in your shoes and I had functional transportation I would definitely specify a late December/early January delivery, not September. We are *this* close - I think it's silly not to wait 90 more days for what is likely going to be a huge leap forward in functionality and safety that will pay dividends to you for years to come.

If anything, I believe the fatality will push Tesla to get the next generation out sooner rather than later. Tesla really needs some good news to get people past the Solar City/fatality/declining delivery mountain of negative press.

Considering the AP processing is all done on the mobile eye hardware, modular upgrades could be a very real possibility.

 

[xav-]

We know the cars sold 3 year ago with the guarantee held up. Tesla recognized $12M additional revenue from expirations (ie when the trade in value was above the guarantee) in 2015. What we don't know (and what is relevent for Tesla in deciding the guarantee going forward) is how the prices will hold up in 3 year. A 3yo Model S CPO will then compete with a brand new model 3 while today it effectively competes with nothing in that price range. More competition means lower prices...

brand new 3, brand new auto pilot.. More range, Likely more reliable, cheaper repairs, more durable.

 

[xav-]

Yep. Why would you need second camera for that?

To add redundancy.. Also each camera can focus on items at different distances enabling collecting more data, reducing the number of false positives/negatives

 

[MarkS22]

Although Jason said he would never do it again. Too much work, too expensive.

There was a lot more that needed to be done than "add sensors." For example, completely replacing the brakes to be CAN bus controlled. (Which he called "super annoying.") In theory, this step wouldn't be needed.

I think a lot will depend on how many new sensors are implemented on AP 2.0. If it's "simply" a trifocal camera cluster and multiple EyeQ3s controlling the same steering and braking hardware, the labor and expense won't be anywhere near what he had to do.

 

[Saghost]

There was a lot more that needed to be done than "add sensors." For example, completely replacing the brakes to be CAN bus controlled. (Which he called "super annoying.") In theory, this step wouldn't be needed.

I think a lot will depend on how many new sensors are implemented on AP 2.0. If it's "simply" a trifocal camera cluster and multiple EyeQ3s controlling the same steering and braking hardware, the labor and expense won't be anywhere near what he had to do.

Exactly. If they do a trifocal only (AP 1.5?) upgrade and the trifocal fits in the current windshield space, I'd expect Tesla to offer an easy and fairly cheap upgrade, pretty much just the cost of the trifocal module with the new chips. If it won't fit and you need a new windshield, they might still offer the upgrade for somewhat more money.

If they jump straight to 2.0 with full surround cameras and possibly more radar, I don't expect Tesla to offer an upgrade - but it'd still be easier to upgrade than what Jason did (because you shouldn't have to touch the brakes or any other systems.)

 

[Beryl]

The smart move, IMO, is for Elon to announce that there will be an upgrade path for existing AP 1.0 vehicles now. I'm saying that as a stockholder -- not just a owner who would benefit from such an announcement. I want to see as many Teslas ordered now as possible.

Given my current knowledge of the car, if I didn't already have a Model S, I would order a 60D ASAP. I wouldn't want to wait another 6-9 months for AP2 and miss my current driving enjoyment. I'd also be willing to pony up the $$$ for the retrofit just like I'd pony up to increase driving range on the 60D.

(After 7 months of EV, I cringe at the thought of pumping gas now.)

 

[tomp]

Recognizing traffic lights is a huge challenge due to different shapes and positions. Some are vertical, some are horizontal. Some are at the far end of the intersection, some hang from a cable in the middle of the intersection. And then there is Europe where the lights are at the nearside of the intersection and the car first in line has them directly overhead.

Doesn't autopilot have an advantage in that it should know (or be able to know based on GPS/road data) exactly where each traffic light is located? Thus it can know when to be looking for one, and also which way is 'straight' vs which way is 'turn left' or 'turn right'?

 

[sorka]

If Tesla continues to behave in the future like they always have in the past there will be no upgrade path.

No upgrade path? What do you call?:

1) Folding mirrors.
2) 3g to 4g.
3) Parking sensors.
4) Sun visor.
5) Software updates to add new features.

And probably another half dozen things I just can't think off the top of my head.

What other car company has EVER updated already sold model year cars in the past??