Will current autopilot cars be able to upgrade to 2.0?

[BluestarE3]

Although I think Tesla will not sell retrofits, it is worth noting:

- wk057 did retrofit an AP0 car to AP1. 0 to 1 is probably more technically dificult than 1 to 2.

-Maybe another source than Tesla will offer the retrofit. As autonomous driving begins to penetrate the market, I suspect there will be a small industry that provides retrofits/enhancements/hacks/etc.

The issue isn't whether or not something is technically possible, but whether or not it's feasible to do so.

 

[FlatSix911]

The Tesla retrofit is not going to happen ...

 

[idealsol]

I have until Tuesday to finalize my order. 2.0 is a concern, and I anxiously await tomorrow's announcement to see if it may have anything to do with 2.0

That being said, it is not a deal breaker either way for me. I am involved with technology and understand that things change and the next, best thing is always around the corner. Just how it is. If you wait for whatever, you'll wait forever

I have a close friend who literally took years to decide on a new car and a even a computer. He could never buy, knew everything about all possible changes, and was paralysed

I know the Tesla, or any car, isn't perfect. But where it excels, it's awesome. A game changer for me regardless of what AP I get even if it can't be upgraded

 

[Canuck]

Its not about lawsuit costs, but reputation.

Tesla won't due retrofits to benefit their reputation since their reputation will be based on their current vehicles in production. That's how reputations work - both good and bad. You're reputation is only as good as your current product regardless of your past good or bad reputation. Just ask Blackberry.

-Maybe another source than Tesla will offer the retrofit. As autonomous driving begins to penetrate the market, I suspect there will be a small industry that provides retrofits/enhancements/hacks/etc.

This will happen but not until people are out of warranty since doing anything as significant as upgrading AP will likely not help you should you have any problem that Tesla could even remotely blame on this work. And since not many are willing to risk losing their warranty over this, you won't find much of a market for it for another few years at least.

 

[wamochi]

this is why i decided to lease this time around (and from now on). i was 110% SURE i was going to keep our 2014 pre-AP car for 8-10 years. i was so very excited they had added parking sensors and folding mirrors. people who wanted those retrofitted on earlier VINs had to pay an arm and a leg. and that's just parking sensors in a bumper.

3 months later AP came out . no way to retrofit that. i don't think a retrofit for AP 2.0 is going to be available either, given the addition of multiple sensors, radars, etc.

p.s. if someone from tesla is listening, can AP 2.0 get the extra sensors for a 360 degree camera view?

 

[Roger_wilco]

Tesla won't due retrofits to benefit their reputation since their reputation will be based on their current vehicles in production. That's how reputations work - both good and bad. You're reputation is only as good as your current product regardless of your past good or bad reputation. Just ask Blackberry.
...

I think something mixed up here. My concerns about reputation due to failure of ap functionality was not about retrofitted cars but factory fitted "old" versions. Next versions will (almost) always outperform older ones. Currently many People are using ap1 cars very carefully but upon landing of ap2 and other brands and increased usage of autonomous driving, ap1 owners may increase their usage in a false safety feeling, like increasing speeds etc.. and from this point, increased rate of ap1 accidents becomes a source of rep. loss for Tesla.

Currently autonomous car makers dont want to share their "big data" with other companies but my guess, at some point, governments will take care of this and will put some minimal but common standards on methods that autonomous car brands apply. In the end, its governments responsibility to protect lives of the citizens. This may be a single big source for environmental map or realtime sharing of env.map data among brands, etc.

 

[bob_p]

Didn't Tesla raise the price of AP 1.0 by $500 recently.

Perhaps those cars have the AP 2.0 wiring harness and able to accept the new sensors and processing (Drive PX2?).

It's possible Tesla could be build AP 2.0-capable cars, even before they are ready to deploy the improvements.

If that happened, coupled with building 100Ds - that might be enough for us to order our next Model S.

 

[supratachophobia]

What do you think? I suspect we all know that more hardware will be required. Almost certainly more cameras, but perhaps lidar.
I know Tesla doesn't have a great record of allowing hardware upgrades, but this is a biggie. I imagine a serious backlash if we get closer to autonomous and the existing autopilot drivers are left with the original system. I would definitely sell my AP1 car and buy a new AP2 car. If others did the same it would flood the user market with guaranteed-resale value cars.

What's the community consensus on this?

My argument has always been that the control systems are already in place. For example, the car can already "drive" itself with no input from a human. Stop, go, turn left/right (lock to lock even) are things that don't require any extra hardware. The obvious question then becomes the following: wiring harnesses for new sensor locations, and location for new AP CPU. In the grand scheme of things, I would think that these are not insurmountable challenges and that they could be done on the customers dime without any additional cost to Tesla. I mean, they've already retrofitted AP 2.0 to TEFV's so they know how to take things apart, put them back together, where to run new wires, etc. So the engineering part that might be left is "how do we make this sexy behind the scenes".

I figure, you take the hardware cost, add the premium that AP just increased by, and throw in an extra $1000 for good measure and that gives you a retrofit cost. So a high side of 3k, assuming you need a new windshield in the S for different/2nd camera? I know that's super simplified, but it's the best I got.

 

[jcaspar]

I think there is NO way that Tesla will offer an upgrade to the new AP 2.0 hardware that is likely coming soon. A waste of their time and resources.

 

[flashflood]

Didn't Tesla raise the price of AP 1.0 by $500 recently.

Perhaps those cars have the AP 2.0 wiring harness and able to accept the new sensors and processing (Drive PX2?).

It's possible Tesla could be build AP 2.0-capable cars, even before they are ready to deploy the improvements.

If that happened, coupled with building 100Ds - that might be enough for us to order our next Model S.

One of the Tesla people I spoke to recently (over a dozen, so he/she can't get in trouble for me posting this) tipped his/her hand a bit. My best recollection of the conversation:

FF: I hear on TMC that you're coming out with AP 2.0 hardware, and some cars may already have it.
T: I read the forums too. Of course, as you know, we can't comment on future products.
FF: Of course. But I'm not asking about a future product. I'm just asking for the specs of a car I already bought.
T: [... crap ... wasn't expecting that ...] Well, people have also commented on the price change for AP.
FF: Yes, it's gone up by $500.
T: Well, let's just say that there's a reason for that, and your P100D is future-proof. That's all I can say.

This makes a lot of sense. It would not be sporting -- it would insane -- ludicrous, even -- to introduce a brand new top-of-the-line, super-expensive car and then obsolete it a month later. A fair inference would be that new P100Ds (and upgraded P90Ds) already have the new hardware, and that all recent cars (going back at least one quarter) can be retrofitted at reasonable cost.

One thing that hasn't changed, though -- the P100D still has the same ancient Tegra 3 MCU, or at least it sure feels like the Tegra 3. (A friend of mine who worked at nVidia told me a bunch of war stories about why Tegra performs so much worse than its formal specs would suggest.) The MCU in general, and the browser in particular, are just as sluggish as my 2014 P85. I think that's part of the reason the 8.0 UI changes are so annoying to some of us -- it's not just the logic of the UX, but also the laggy physical experience. It's totally counter to the brand: you don't expect such slow and dated tech in such a fast and futuristic car.

 

[Hengist]

My prediction, for what it's worth, is that all Model Xs and all Refresh Model S will be able to upgrade to AP 2.0 for a flat $8,000.

We know the X has the wiring for the second camera. We know the refresh S has a redesign of many key front end parts, not just the fascia. So it makes sense that the wiring and mount points for the two extra radars would be there.

 

[beegee]

I have a 2012 P85, and there are no retrofits for the non-autopilot cars, so I doubt there will be upgrades to version 2.0 either. I just enjoy what I have and will eventually upgrade to the latest hardware one day.

 

[Canuck]

One of the Tesla people I spoke to recently (over a dozen, so he/she can't get in trouble for me posting this) tipped his/her hand a bit.

Your first mistake since you assume Tesla people know what's coming down the line. They don't. That's been proven time and time again. Do you think they say to employees: "AP2.0 is coming but don't tell anyone" !?

This makes a lot of sense. It would not be sporting -- it would insane -- ludicrous, even -- to introduce a brand new top-of-the-line, super-expensive car and then obsolete it a month later.

Your second mistake. Just ask someone who bought a P85+.

T: Well, let's just say that there's a reason for that, and your P100D is future-proof. That's all I can say.

Your third and most glaring mistake. Nothing at all is "future-proof". It's actually quite funny to even suggest that.

 

[flashflood]

@Canuck, regarding my three mistakes:

(1) True, they don't broadcast to all employees what's in the pipeline. But some employees are more resourceful than others. If this particular employee occasionally has lunch with the right engineers or marketing people, he'll learn stuff. Even Apple is leaky.

(2) The P85+ had a pretty good run time-wise, but never became popular. Any successful business cancels products that don't sell.

(3) Indeed, nothing is future-proof. I took the comment to mean "this hardware won't be obsoleted in the next year or so". There will no doubt be ongoing improvements in software, and modest improvements in hardware, but those won't obsolete you -- unlike the introduction of Autopilot, which left pre-AP owners out in the cold as almost all of Tesla's focus and firmware effort shifted to AP. I took the comment to mean that there's no comparable obsoleting event on the horizon, not that none will ever happen.

 

[Canuck]

(1) True, they don't broadcast to all employees what's in the pipeline. But some employees are more resourceful than others. If this particular employee occasionally has lunch with the right engineers or marketing people, he'll learn stuff. Even Apple is leaky.

Engineers discuss confidential information over lunch and risk breaching non-disclosure agreements? That's news to me. Yes, Apple had leaks to the WSJ, but we know for a fact now that was a planned corporate strategy. It certainly was not by the engineers discussing things over lunch.

(2) The P85+ had a pretty good run time-wise, but never became popular. Any successful business cancels products that don't sell.

It wasn't dropped due to low popularity. It was dropped since the D produced even better results. The same thing is going to happen with AP2.0 producing better results than 1.0. Tesla doesn't consider "run time" in making improvements -- we know that if history is any example.

(3) Indeed, nothing is future-proof. I took the comment to mean "this hardware won't be obsoleted in the next year or so".

So now "future proof" has become one year? In other words, Tesla will hold back for one year before introducing something better, or will offer retrofits to vehicles made during the past year before introduction, so as not to obsolete prior buyers even though historically they have never done that? I doubt it.

But only time will tell who is right on this issue. For all we know, tomorrow's announcement won't even be AP2.0 but if it is, we should know tomorrow about retrofits since I'm certain Tesla will have to deal with that elephant in the room if they do announce new AP hardware.

 

[Saghost]

Tesla obviously had a plan to upgrade cars in the future when they shipped the X with the empty housing. However, that was before the Mobileye split, and possibly before they saw that split coming. Four months ago I was confident that we'd have an upgrade to next generation for the X and refresh S and possibly the rest.

After the split and the 8.0 announcement I'm not. It seems clear that Tesla's roadmap for Autopilot has changed substantially in the last 6 months - and presumably that makes the preparations they'd made for the next generation no longer relevant, so unless they are able to build a new architecture that fits with the preinstalled hardware and supports the future roadmap, upgrades from Tesla are probably out.

It'll likely be several months after the release of new hardware before the new cars can do anything the current ones can't, and Tesla still has a fair amount of refinement they can do with the hardware we have, in addition to the whole whitelist/radar map evolution they are doing. Given how much they are still learning, they might not have finalized the next generation design yet.

In their shoes, I'd wait until the last minute - until I'm vendor lead time away from the end of the mobileye contract - before finalizing the configuration and placing orders. That would give me the most chance to learn from version 8 and the whitelist and evolve my configuration in light of that information.

The big question mark is when the Mobileye contract ends. One obvious possibility is the end of the year, but it also could be a random day in October - or March.

 

[Canuck]

Tesla obviously had a plan to upgrade cars in the future when they shipped the X with the empty housing.

Is that obvious? Perhaps they knew of that obvious camera location so rather than build the housing differently from the start, and change it later, they built it so that it wouldn't need changing later, to save costs, without planning to do retrofits. That seems possible to me.

 

[flashflood]

Engineers discuss confidential information over lunch and risk breaching non-disclosure agreements? That's news to me.

Yes. In other news, politicians lie and there's no such thing as Santa Claus.

A colleague once (accurately) described an industry conference a place where competitors get together and discuss trade secret.

 

[Saghost]

Is that obvious? Perhaps they knew of that obvious camera location so rather than build the housing differently from the start, and change it later, they built it so that it wouldn't need changing later, to save costs, without planning to do retrofits. That seems possible to me.

But we also have video showing wiring in the overhead for the second camera - extra weight and cost.

Given that Tesla has complete confidence in their configuration control (20 engineering changes a week, remember) why would they add the wiring unless they planned to use it later?

They also have a history of giving retrofits and of preinstalling wiring before the hardware arrives - look at the power folding mirror saga.

 

[Canuck]

Yes. In other news, politicians lie and there's no such thing as Santa Claus. A colleague once (accurately) described an industry conference a place where competitors get together and discuss trade secret.

So first you use Apple as an example, and when I point you to the source of the leaks, you fall back on politicians, Santa Clause and a colleague?

But we also have video showing wiring in the overhead for the second camera - extra weight and cost.

Perhaps but as others here have pointed out, wiring is cheap and the weight is nominal. Retooling the assembly line and making the changes later, if they planned to bring it in relatively soon, could cost more. But I do agree that the split with MobilEye definitely caused a change of plans. I'm just not so certain that because the wiring and housing is there, it's obvious that retrofits were planned. I agree it's a possibility, it's just not obvious to me, especially with how crowded and overbooked most service centres are, and how Tesla historically likes to give people reasons to constantly upgrade and buy new vehicles.