Working on Autopilot v2 hardware suite - Confirmation by Elon

[satheesh.net]

So as you probably know by now, Elon has confirmed that they are working on a new hardware suit, let's call it Autopilot v2.


It's a shame that existing TMX owners and near future owners (including yours truly) will not have this new set of features already installed and up and running. The current Autopilot actually gives me the features I need today and probably for a little while longer. Summoning the car from another city isn't really something I need per say.


So the question is; wait for the new hardware or be happy with the current one that's already here?


(From the conference call on Sunday 11th. of January. TOCN Live! Elon Musk on firmware 7.1)

 

[spentan]

I wonder if they learnt their lesson and made it retrofittable this time. Model X especially Signature owners are gonna be really pissed that they waited 4 years for a car that was obsoleted in 6 months

 

[satheesh.net]

I wonder if they learnt their lesson and made it retrofittable this time. Model X especially Signature owners are gonna be really pissed that they waited 4 years for a car that was obsoleted in 6 months

I totally agree! I'm also hoping that they learned their lesson but not holding my breath on this one... Maybe we should start a petition to be sure that we get their attention as early as possible.

 

[J1mbo]

It is possible that the X already has two rear corner radars. See the screenshot from 7.0 developer mode in this thread.

(Important to note that S had a bunch of settings for Autopilot hardware in the Developer screen for years before AP was launched).

Rear corner radar helps with overtaking, as the car has a better view of cars that are coming up from behind than the ultrasonic sensors can provide. I would guess that this is a critical feature for the sort of autonomy that Elon was talking about.

 

[ankitmishra]

He said that the new hardware suite will take time. And why would a feature that will make people want to upgrade their cars would be retrofittable? If people want the new hardware, upgrade to the latest model.

 

[satheesh.net]

Btw, a cool thing with v7.1 and the current hardware, is that it appears to "see" this cyclist and slows down before passing it and then speeds up again. (YouTube)

 

[timf]

The way I read it, the upgraded sensor set is still 2-3 years away. It may be ready about the time of Model 3 release or shortly thereafter. Is it worth waiting that long in order to achieve full autonomy? I would say no, because there's no guarantee the software and regulatory approvals will be in place to allow it. The amount they are able to achieve with the current feature set is impressive enough and by the time full autonomy is readily available I'll likely be ready for a new car anyway.

 

[MXinCT]

I would venture to guess the new hardware suite is a solid year away as they have just now really started pushing autopilot software and promised features on the original sensor suite. The longer they can support a single sensor suite in the wild the better as far as development and implementation goes. I would bet they don't make the same mistake the second time around and roll out the hardware a year before anyone gets to use it. All of the features Elon mentions required for truly autonomous driving are going to add some cost to each car for sure and with their push for profitability i doubt they will throw a couple grand of hardware in the cars without getting paid for it.

 

[hockeythug]

The 8-camera suite and the new EyeQ4 chips is not going to be a retrofit.

 

[Congo Line]

I don't know, this is seriously making me think about putting off my Model X purchase. My wife could still drive her current SUV for another year or two. I already did the upgrade thing with my Model S when autopilot was announced; don't feel like doing it with Model X again once v2 comes out.

 

[Vitold]

So the question is; wait for the new hardware or be happy with the current one that's already here?

How long are you willing to wait? Tesla has one of the best systems at the moment and they are still improving it. Considering that Tesla's AP engineer quit and EM announced that he is assembling new AP team I personally think that they are at least one year away.

 

[roblab]

How long are you willing to wait? Tesla has one of the best systems at the moment and they are still improving it. Considering that Tesla's AP engineer quit and EM announced that he is assembling new AP team I personally think that they are at least one year away.

First off, we are pretty sure EyeQ from Mobileye will not be the hardware. Or the software.

Second, at least in my car, there is room by the rear view mirror, on the right, for a bunch of cameras, radar, who knows what else.

Elon is building his own version of AP. I think a lot of it will work in my current car, and I expect hardware upgrades. As to a "year", I wouldn't conjecture that far out. There are times it seems Tesla takes a long time, but other times when, "poof", it downloaded over the air and is waiting for you to install.

I wouldn't wait. The perfect is the enemy of the good. You will likely buy another Tesla and upgrade, and find the prices have dropped again.

 

[ohmman]

First off, we are pretty sure EyeQ from Mobileye will not be the hardware. Or the software.

Second, at least in my car, there is room by the rear view mirror, on the right, for a bunch of cameras, radar, who knows what else.

Elon is building his own version of AP. I think a lot of it will work in my current car, and I expect hardware upgrades. As to a "year", I wouldn't conjecture that far out. There are times it seems Tesla takes a long time, but other times when, "poof", it downloaded over the air and is waiting for you to install.

I wouldn't wait. The perfect is the enemy of the good. You will likely buy another Tesla and upgrade, and find the prices have dropped again.

^ Yes. Shop the car you know, not the car you're extrapolating. If you need or even really, really want full autonomous driving, Tesla doesn't meet your needs, so just wait until it does. If not, they have a pretty great product available today.

You could easily make the same argument about waiting for another manufacturer's EV. Those are all "coming soon" as well, and they might be awesome.

I'd change one point - "there are times it seems that Tesla takes a long time, then there are times that it seems like Tesla takes a really long time." When it comes to developing 100% in house autopilot, I am going to guess it's one of those two.

 

[AdamSM]

Yeah, I'll be sad if I get my Model X in 6 or 9 months and then AP2.0 hardware rolls out 3 or 6 months later. But there's no guarantee that will be the pattern--it could just as easily be a year or 18 months later for AP2.0 hardware, and another year for AP2.0 software that actually opens up meaningful new functionality that AP1.0 hardware can't support. Not getting the car I want now because there might be a better version later is a pattern that results in not buying a Tesla for 5 or 10 years, if ever.

If it's the faster timeline, then we'd have to figure out whether it's worth the very large expense of selling the old model and buying a new one. It would probably mean a few years delay to when we get the new model. But frankly, that's not the end of the world. And getting a Model X soon is a big deal for us.

Also, the collective action issue is worth thinking about. If everyone waits for the new technology, it may never arrive, because Tesla needs to sell their current cars to afford the R&D for their future cars.

 

[ecarfan]

@ohmman's points are persuasive. The only thing we know for sure is that Tesla will continue to improve their vehicles. We don't know when improvements will occur because Tesla does not operate on the traditional "model year" schedule. If you are considering buying a Tesla, make your purchase decision based on what you can buy right now. If you think you might wait to buy because "something better is coming" you will never buy. "Something better" is always coming. So what? Teslas are awesome cars right now! My "classic" S is an awesome car.

^ Yes. Shop the car you know, not the car you're extrapolating. If you need or even really, really want full autonomous driving, Tesla doesn't meet your needs, so just wait until it does. If not, they have a pretty great product available today.

You could easily make the same argument about waiting for another manufacturer's EV. Those are all "coming soon" as well, and they might be awesome.

I'd change one point - "there are times it seems that Tesla takes a long time, then there are times that it seems like Tesla takes a really long time." When it comes to developing 100% in house autopilot, I am going to guess it's one of those two.

 

[zer0cool]

The fact is that Elon's promises generally come later than he say they would. For example, model X took 2 more years, autopilot software took 1 more year than the hardware, and Falcon 9 took 3 tries. Although he's thinking v2 will be full auto driving, but first that might be 2 to 3 years away, and in fact I think there will likely need to be a v3, if not more, and that's 5-6 years away etc. At the end, every car improves, and the new car is always better than the previous car, for every car company... If you don't buy the current car, which is the best there is today, in hopes of a better car 2 or 3 years down the line, then you ll never actually have a car... This is true for Tesla, for Porsche, Mercedes, whatever.

E.g. people all knew like a year ago that the new Mercedes E class would be amazingly better compared with the current E, but did people just stop buying the current E? Not really... because that new E was still a couple years away.

 

[RockinWithKids]

The 8-camera suite and the new EyeQ4 chips is not going to be a retrofit.

I don't know, this is seriously making me think about putting off my Model X purchase. My wife could still drive her current SUV for another year or two. I already did the upgrade thing with my Model S when autopilot was announced; don't feel like doing it with Model X again once v2 comes out.

According to MobilEye... the production EyeQ4 chips will not be in car manufacturers hands until 2018.

 

[HoptownT]

Tesla time people. It warps space time to slow everything down. Your 3 to 4 months translates to few years from now. 2 years from now means about 3 to 4. You can have your model X with no regrets. Even if full autonomy is ready, your local legal system isn't ready.

 

[CarlK]

It's a shame that existing TMX owners and near future owners (including yours truly) will not have this new set of features already installed and up and running. The current Autopilot actually gives me the features I need today and probably for a little while longer. Summoning the car from another city isn't really something I need per say.

It's also a shame that people who waited and buy AP v2 cars would be screwed when AP v3 comes out. Buy the car if it best suits you needs and not if it doesn't. Out-smarting yourself never does one any good.

 

[ModelXBoy]

It's also a shame that people who waited and buy AP v2 cars would be screwed when AP v3 comes out. Buy the car if it best suits you needs and not if it doesn't. Out-smarting yourself never does one any good.

Total agreed! There is never a end to chase the next greatest technology. Elon does things base on what he thinks what is advance and base on how feasible their engineers can execute his vision. If it's possible, they will do it regardless how long their customers have been waiting for the previous generation of technology. They will piss some people regardless when to come out the new technology anyway, so the best way is to come out at anytime when the technology is ready.

On the other hand, i really hope they can integrate this v2 tech into the production model X since we are still waiting for the x to be delivered. Finger crossed : )