Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Any chance for upgrading hardware from "2014 Autopilot" to "2016 Autonomous ready"?

This site may earn commission on affiliate links.
I really hope not. Mercedes Benz is already using the slogan "More sensors, more protection" as a shot at Tesla:

I don't see Tesla falling further behind when it comes to hardware, especially since other automakers will have more hardware on 2017 vehicles (released in 2016) than Telsa will have. I think the hardware is coming early 2016 but the software will come a lot later, just like auto pilot 1.0. The hardware is not expensive and this is Elon Musk's pet project.

Yeah, but more hardware isn't entirely necessary. MB isn't planning to do any more with their stuff than what Tesla already has... Unless I missed something. And MB is in need of an update since their previous hardware was from 2014, so 2 years later, they are due for an update. So Tesla isn't "first" in what they have done, but they did it better. I suspect that Tesla will do the same thing with the next update. They won't be the first (MB will likely beat them to it), but they will do it better.
 
Yeah, but more hardware isn't entirely necessary.

I think Tesla wants to be the first to do turns at intersections. That makes new hardware entirely necessary. MB uses this graphic also likely as a shot against Tesla, since Tesla's single front facing camera's known weakness is intersections:


mb5.jpg


I guess only time will tell who is right on the hardware roll out date. I doubt that there will be any announcement. It will likely be similar to 1.0 when the cars just started showing up with the hardware.

I do have a horse in this race since I don't want to wait until 2017 to upgrade, and I don't want to upgrade now only to find out auto pilot 2.0 rolls off the line on the car after mine. So there's no doubt I am bias in my views and I look for cues that support my position.
 
Counterpoint on the tea leaves for Tesla releasing it next year is that they very likely will need to do a lot of internal testing of it first. Note that it generally takes time to take a product (the 8 cameras) and incorporate it into your vehicle, and then program the car's responses to this environment. So, even if they are releasing it to Tesla next year, it will be some time before they actually get incorporated and then the supply ramps up to be added into production cars. They didn't start putting it in the Tesla until, what? 1 year after it first became an available thing?

Second, your clipping out of the commentary from the CEO makes it seem like it could be Tesla, but if you go and read the whole transcript I think you are misreading the comments. Let me give that whole section for all to read:



So you have the CEO describing a product (deep learning and reading of free space) that they confirm is what is in the Tesla and then state that additionally these capabilities will be available to two OEMs in 2016. This basically means that the 2016 release is *NOT* Tesla, because Tesla already had implemented this capability. Why would Tesla need to re-release this capability if it already exists, as stated by the CEO? This doesn't flow with how the speech was prepared, and would be reading something that wasn't even hinted at.

The next paragraph then goes back to Tesla, he states that they are using a single camera, but then states that the multi-camera launches could begin as early as next year. Now, you could either read that this "early as next year" is alluding to the aforementioned two OEMs (wording doesn't fit), or most likely this is alluding to an as of yet, un-solidified date by some *other manufacturer*(s). This could be Tesla... maybe not... but it isn't confirmed to be 2016, it is a lot less certain. There are known OEMs with the three camera setup in the front upcoming on cars, so this could be that?

Then they talk about 4 separate launches of the tri-camera only and then two launches of the 8 camera which, based on the next sentance, the whole launch process will go from 2016 - 2019. Tesla is very likely to be one of these, but which is it? the 3 camera setup? the 8 camera setup? I don't think anything is certain, and it could be that they are on track to implement this "as early as 2016" or in the "2016 to 2019 timeframe".

My personal bet is that it will come either at the end of 2017 as they get closer to the Model 3, they will push down the Model S hardware into the 3, and refresh the S and X with the hot new toys to ensure that those who want to keep with the latest and greatest will buy the high price Model S/X and those who can't won't care either way, since a 35k (plus 2,500 on an option) partially self-driving car is better than anyone in that price range is likely to have available to them.
MobileEye has multiple different development tracks with different OEMs, so they're referencing one of their other configurations (likely 3-camera front facing). Their cutting edge solution is 8 cameras which they think is Automous ready, so that's the finish line in terms of hardware. They have only 1 automaker who is really leaning in on this technology - that's Tesla. MBLY's CEO praised TSLA's team earlier on in the call, saying it was a delight to work with an OEM who is so willing to push the envelope. Tesla will get the 8-camera system first, don't think there's any question about that.
 
It's OK I'm biased too, I have a car on order for December... So if they release something now it will be a little depressing... Lol.

Well that explains our positions... lol. If I am right, I don't think you'll really miss much if 2.0 does come out soon. I bet the software for it is years away. The car does more now with 1.0 than people can handle (or at least the yahoos on Youtube are ruining it for everyone) that they are apparently dialing it back, or at least doing something about it, as stated in that other thread. And if I am wrong, if I don't mind driving my current car, in fact I really enjoy it, and I'll be happy for you. After all, I don't want to get in an argument with an evil scary looking chicken! One punch from those arms and I'm out!
 
Just so we have the numbers close to reality. We are probably at around 105k Model S's sold worldwide to date. 50k of those have AP. I haven't crunched the numbers exactly but there aren't 60k AP and there are definitely not 200k total.

I think we have 2 years of SW improvements with the current sensors. So my 3 year lease started in May 15 is going to work out pretty well.
I just looked up the numbers. They've shipped 89,925 MS units thru the end of Q3, of which 43,008 have AP if you assume AP starts at the beginning of Q4'14.

The question is whether Tesla offering an 8-camera upgrade for current AP owners is going to matter for future sales. They're targeting shipping 500k units annually once the Model 3 launches (altho Wall St analysts have estimates much lower, about half that). Lets say 5% of current gen AP hardware owners would upgrade just for autonomous hardware, that's 2,150 units or 0.4% of 500k. Tesla shouldn't be worried about it - they should try to make current owners happy if (assuming this upgrade even makes economic sense).
 
MobileEye has multiple different development tracks with different OEMs, so they're referencing one of their other configurations (likely 3-camera front facing). Their cutting edge solution is 8 cameras which they think is Automous ready, so that's the finish line in terms of hardware. They have only 1 automaker who is really leaning in on this technology - that's Tesla. MBLY's CEO praised TSLA's team earlier on in the call, saying it was a delight to work with an OEM who is so willing to push the envelope. Tesla will get the 8-camera system first, don't think there's any question about that.

Except they already said Nissan was the first to stick their cameras in a car (sure it was a concept... But still...) I am not denying that Tesla and MobileEye are the bestest of friends just saying that it isn't so clear cut that it is Tesla that they were referring to for 2016 and if it is then it likely isn't the 8 camera one, but rather 3 which would be disappointing... So I would rather Tesla go for 2017/18 and get the 8 cameras so they have a full 2 years of people mapping the roads for them (basically we paid Tesla to get the privilege to map the roads... It's pretty genious). And when they drop the 8 camera system it will essentially be a completed solution. If they go before that, great, but would it be better to sink all that extra hardware cost on something you won't see pay out for your customers until 3 years later or just wait and sell the lesser hardware and get a nice GM on it? There are quite a few reasons it makes sense for them to stick with what they have at the moment. I could totally be wrong. Just how I see it.

- - - Updated - - -

Well that explains our positions... lol. If I am right, I don't think you'll really miss much if 2.0 does come out soon. I bet the software for it is years away. The car does more now with 1.0 than people can handle (or at least the yahoos on Youtube are ruining it for everyone) that they are apparently dialing it back, or at least doing something about it, as stated in that other thread. And if I am wrong, if I don't mind driving my current car, in fact I really enjoy it, and I'll be happy for you. After all, I don't want to get in an argument with an evil scary looking chicken! One punch from those arms and I'm out!

Meh, as long as I get to keep highway driving I'm cool. Most of my driving is on the highway, and what isn't is technically well marked enough that I can get away with it there to. But really, it is that morning commute that I want it for, and possibly road trips, everything else is lesser driving that I would feel safer doing myself anyway.
 
So let me get this straight.

Model X Founder's and Signature series will have APv1, and then general production will have APv2?

Hey well, having your Model X this year is its own feature, in Tesla sense. Plus, its Branding will allow you to sell it for a higher amount as you get your upgraded Model X shipped to you.

Edit: the above is my wild speculation, not any type of informed knowledge. I based it upon the timelines quoted in this thread for APv2 and TMC for Model X deliveries by other posters. It might even be close to right, and you get your general production Model X in about a year, two weeks before APv2 is installed on all vehicles, and you get no upgrade path.

- - - Updated - - -

Since at some point APv2 HW will be tested enough to know where to put it, there might be a lead-up time where Tesla knows what body and cabling to install for APv2 but doesn't have the actual cameras and chips to install, so Tesla could build in a very small number of VINs as "APv2 HW ready but no APv2 HW installed". This would also be easy to market since it would imply that prior VINs are "not APv2 HW ready". Once the APv2 HW starts showing up at Tesla they can start preinstalling it. In fact, this whole thing might offer its own fork in the road: every Tesla after that could be made AP HW ready but without the AP HW installed, and make it an option (both factory and post-factory installed).

More ruminations.
 
The video I posted earlier has some interesting hardware attached to the MS and the test track seems to show capability that is much closer to semi-autonomous driving. What do you all think is up with the "attachments" to the MS in this photos?
Tesla3.JPG
Tesla2.JPG
Tesla1.JPG
 

Attachments

  • Tesla4.JPG
    Tesla4.JPG
    89.6 KB · Views: 114
Any one knows if there is any limitation coming from government regulation for autopilot? So far I heard only limitation coming from Tesla :)
If autopilot fails and creates dangerous situation on the road, who will cover expenses? Insurance company? Tesla? The first several years anyway all drivers using autopilot will be like Beta-testers.

The driver will be responsible, as it stated that the driver must keep hands on the wheel. Autopilot != Autonomous driving. Autopilot is a tool to ease your driving, it doesn't take over your responsibility to pay attention.

Once autonomous driving is out, things will be different, and I believe several manufacturers (Volvo and someone else?) stated that they cover any [at-fault?] accidents related to autonomous driving.
 
Came across this DIY install guide for front and back dashcams.. somewhat comparable to what might be required to upgrade MobileEye-powered cameras in the front and back. Side cameras maybe a little more tricky..

Tesla Model S BlackVue HD Dashcam Installation Guide - TESLARATI.com

Except knowing Tesla they will want it integrated cleanly into the windshield, which will need a new windshield and possibly new headliner (again). Rear camera would require new back windshield at a minimum. Side cameras (there are 4 of them) not sure where they get positioned, but would likely require either new windows or holes drilled into the side of the car. No, I would not expect a feasible upgrade path to the 8 camera setup. Nevermind that you will also need to add in some location for the multiple added computer chips to run these (which I don't think reside near the camera unit). I think they said the 8 camera setup is powered by 5 of their processors.

I would love it to be some kind of DIY/Aftermarket option since all the controlling capability is already there, but I am afraid this isn't likely to happen as the cost would still be greater than Tesla would be willing to allow.