Tesla.com - "Transitioning to Tesla Vision"

[gearchruncher]

What’s in that for Tesla?

What's in keeping their cars safe for their customers, keeping customers happy, and avoiding negative news cycles instead of installing radar for a few weeks until the software is ready?

You are a fast learner!

 

[qdeathstar]

A

What's in keeping their cars safe for their customers, keeping customers happy, and avoiding negative news cycles instead of installing radar for a few weeks until the software is ready?

You are a fast learner!

Well I mean if the choices are delivering fewer cars and delivering more cars that are just a little less safe than they were, but still overall very safe.... I know which choice I would make as a company.

safety third

 

[swheat6345]

Interesting article Tesla Model Ys without radar equip several changes to improve Autopilot performance

 

[RDoc]

Interesting article Tesla Model Ys without radar equip several changes to improve Autopilot performance

The video is more of an ad than a review.

 

[swheat6345]

The video is more of an ad than a review.

I do not watch the videos. I just read the article. There are some interesting points and I had to look at my neighbors 2020 MY vs MY 2021 MY and there are subtle differences which the article eludes too. I am not sure but the articles says the video does to but I did not watch the video.

 

[powertoold]

Vision-only only cars better than radar cars? Impossible!!

 

[JHCCAZ]

I do not watch the videos. I just read the article. There are some interesting points and I had to look at my neighbors 2020 MY vs MY 2021 MY and there are subtle differences which the article eludes too. I am not sure but the articles says the video does to but I did not watch the video.

I'd say the video is actually a little less ad-like than the article.

BTW, both talk about new wipers when they really mean new washers - but this is interesting and seems like a real improvement that I hadn't known about. Let's give credit for that while we're grumping about the lumbar scandal.

There's a little much glossing-over of potential down-sides, particularly regarding the Auto High Beams. I'll reserve final judgment for when I try it myself, but I tend to believe the reports of poor behavior.

To be fair, I think that this and some other AP gripes are made in the context of designed-for-highway AP being used in non-highway situations. I'm still very excited about City Streets release (yes I'm ordering FSD largely in hopes of getting that specific capabiliy), and I'm thinking that the headlights will be controlled more sensibly in urban/suburban street driving.

 

[gearchruncher]

Well I mean if the choices are delivering fewer cars and delivering more cars that are just a little less safe than they were, but still overall very safe.... I know which choice I would make as a company.

This situation only exists when there is a parts shortage. This would be the tough decision I eluded to.
The argument made is that there is no shortage of parts, only of money. So Tesla could have put radar in, but instead decided to save $75 and reduce safety.

 

[qdeathstar]

This situation only exists when there is a parts shortage. This would be the tough decision I eluded to.
The argument made is that there is no shortage of parts, only of money. So Tesla could have put radar in, but instead decided to save $75 and reduce safety.

Either way.... a shortage is a shortage....

 

[gearchruncher]

Either way.... a shortage is a shortage....

Tesla makes 38,000 3/Y vehicles a month. As we know, they tend to make North America cars in the last weeks of the quarter. They claim confidence that they will hit parity in just a few weeks, so they should easily not need to install radar next quarter, without any degradation experienced by customers.

38,000 radar units @ $75 each is $2.9M.

Elon's 2021 Compensation is $30,000M, based on the idea that he has made a very healthy company.

If Tesla is short $3M to the point they are willing to dilute their brand, while having a market cap of more than Toyota, Volkswagen, GM, Ford, Fiat Chrysler, Nissan and Daimler combined, there is a completely different story to be discussed here.

Or, you know, they couldn't get parts.

 

[qdeathstar]

Tesla makes 38,000 3/Y vehicles a month. As we know, they tend to make North America cars in the last weeks of the quarter. They claim confidence that they will hit parity in just a few weeks, so they should easily not need to install radar next quarter, without any degradation experienced by customers.

Yea, but they make a lot of claims...

If Tesla is short $3M to the point they are willing to dilute their brand,

I don’t think as many people have a missing radar on their radar as you think.

Most people who are wanting to buy a Tesla think the car already drives itself. They believe Elon when hey says pure vision is parity. They aren’t a bunch of nerds examining each frame of video like we are. Just saying.

Or, you know, they couldn't get parts.

Totally, either way.

 

[gearchruncher]

I don’t think as many people have a missing radar on their radar as you think.

Most people who are wanting to buy a Tesla think the car already drives itself. They believe Elon when they say pure vision is parity. They aren’t a bunch of nerds examining each frame of video like we are. Just saying.

If it's less than 38,000 vehicles, it's even more ridiculous that they removed it to save money. The fewer cars this impacts, the more insane this looks if it was a pure financial decision.

The brand dilution I am talking about is that even local news outlets are running stories saying that Tesla loses safety ratings after they remove radar. A news cycle any $600B company would happily pay $2M to avoid.

 

[rxlawdude]

Yea, but they make a lot of claims...



I don’t think as many people have a missing radar on their radar as you think.

Most people who are wanting to buy a Tesla think the car already drives itself. They believe Elon when hey says pure vision is parity. They aren’t a bunch of nerds examining each frame of video like we are. Just saying.

There's an ocean between being a frame-dissecting geek and one doing due diligence.

The latter would understand that at this time, the car can drive itself on divided highways, with driver supervision.

 

[qdeathstar]

I hear you both, but you are pointing angrily at the glare of headlights from a Mack Truck as it bears down on you, yelling at the top of your lungs “BUT ITS SLOWING DOWN!!!” as a deafening wind swirls round and through you.

maybe if it had radar it would a seen you....

 

[stopcrazypp]

Tesla is historically a company that focuses on safety first, autonomy second, cost third.

If removing radar does not make the car less safe, then there is minimal reason to not remove it on the S/X. Elon has been saying vision only is the future for Tesla, so who doesn't want a car that is most aligned with the future? If that's true, there is no reason to leave radar on the S/X. It's all wasted money, and it's as simple as not installing them on the production line. Software is only a few weeks away.

We are arguing in circles. I addressed all the points about why it's still on the S/X in previous posts. Removing them in S/X will also remove check marks for those models, and Tesla doesn't need this as part of the Plaid news. Model 3/Y is not the middle of a major refresh, so it matters much less for them. And given the relative cost to the S/X is low (vs margins), plus given the lower volume, the costs of keeping it for now isn't going to be as high. Again we don't know software is only a few weeks away. Tesla only officially said they will "start" rolling out OTAs to bring things to parity in the coming weeks, they have no solid timetable for when they will achieve it.

But the reverse is that they know the removal of this is less safe and Tesla just picked $75 over max safety, which is not a good look for them, or a something to make you feel confident about the company's decisions long term.

Again a false dilemma and non sequitur. You are making multiple assertions here: #1 it is less safe. #2 is $75. #3 they are making a choice between the two.
Again, I don't care which one is the "good look" for them. I'm talking about which one is the more likely cause given the facts we know so far and the cause may not be a "good look". Again, their official "good look" explanation is that they are removing it from Model 3/Y as it's the highest volume (better for testing) and this is part of their progressive rollout of Tesla Vision.

For many of us, this very much looks like a difficult decision Tesla had to make in a short time period - and the logical reason for that is supply chain. The story that they did all of this without any pressure except saving a few dollars seems to require them making quite a few very odd and worrysome decisions.

They just went and cut out the passenger lumbar button, which costs far less than a radar unit. It's fairly obvious they are in cost saving mode. If they feel like they'll be eventually removing radar anyways, why not remove it now when they are under price pressure? Tesla have already raised the base price of the Model 3 $2000 more than when I bought it at the end of last year.

The supply chain decision is in one way, a much better look for the company, because the other path makes it look like they are prioritizing the safety of their richest customers over the safety of the majority of their customers (and where they actually make their money due to volume). Those of us that believe it are in a sense giving the company the benefit of the doubt over the other worse possibilities. The actual shaking my head moment is that Tesla didn't just acknowledge the real reason for this change, and instead had to post some nonsensical excuse for why it's still on the S/X, while also trying to sell this as a good thing for 3/Y owners. Just tell the truth.

Again "better look" is irrelevant to the discussion. We aren't discussing which reason would make Tesla "look better". We are discussing which reason is more likely.

 

[stopcrazypp]

I'm going to insert a tecnical/logic comment, here not taking sides in the whole Elon is/is-not whatever. I tend to agree with much of your other viewpoint, so please take the following as constructive.

I just want to note that much of your recent commentary has been focused on whether there a "chip" shortage vs. a "radar" shortage. I would say that this is a distinction without a difference.

Radar is an electronic technology, a specialized form of radio, for which the system description cannot be separated into the "radar" vs. the "chip" part (noting that basically, "chip" = "semiconductor" = "most all electronic content" in our day). Radar is the whole module, the superset, and chips are inseparably at the heart of it. These include "controller" chips as well as transmit power amplifiers, low-noise receive amplifiers, tuning and modulation/demodulation sections and so on. Probably in these cost-sensitive automotive radars, the various functions are integrated into very few custom chips. Sure there are non-semiconductor elements like antenna/waveguide (probably integrated into the circuit board design), coils, resistors and capacitors - and yes these (especially caps) have sometimes been subject to shortages, but that isn't what I'm hearing today.

So think it's most sensible that a radar-module shortage would be driven by the known and overarching shortage of semiconductor chips. This isn't pedantry; I'm only posting it because of the repeatedly-stated concern about whether it's a radar shortage or a chip shortage - I think it's not either/or here.

Note, in this discussion, I'm talking about about the "radar unit" or "radar module" which includes all the components that you talk about above. I only refer it to "radar" as it's redundant to add "unit" or "module" to it. I see no evidence of "radar unit" or "radar module" shortages.

What Elon mentioned about writing firmware for replacement microcontrollers would not be applicable to the radar situation given they aren't replacing the microcontrollers inside the radar module (it's sold as a whole unit, so it's not like Continental would ship them ones without the chips), rather they are removing the whole radar module from the car (no replacement). So Elon is obviously talking about some other part in his comments.

I see however evidence of shortages of various control modules, infotainment and navigation systems, ADAS control modules (like ESP control modules, Mercedes Pre-safe). That's what I'm talking about.

 

[DanCar]

... I see no evidence of "radar unit" or "radar module" shortages. ...

CEO of largest U.S. chip foundry explains why semiconductor shortage could last through 2022

Currently, all of GlobalFoundries' manufacturing capacity is booked — and it expects semiconductor supply to lag behind demand until 2022 or later.

www.cnbc.com

... The shortages are for other parts cars need, like radar chips, which don’t necessarily require the most advanced manufacturing available at the time. ...

 

[stopcrazypp]

CEO of largest U.S. chip foundry explains why semiconductor shortage could last through 2022

Currently, all of GlobalFoundries' manufacturing capacity is booked — and it expects semiconductor supply to lag behind demand until 2022 or later.

www.cnbc.com

Thanks, that the only evidence so far I see that even points to a possibility of radar units being in short supply. From what I can find GlobalFoundries is a supplier to Bosch.
GLOBALFOUNDRIES 22FDX RF Solution Provides the Basis for Next-Gen mmWave Automotive Radar | GLOBALFOUNDRIES

 

[JHCCAZ]

Note, in this discussion, I'm talking about about the "radar unit" or "radar module" which includes all the components that you talk about above. I only refer it to "radar" as it's redundant to add "unit" or "module" to it. I see no evidence of "radar unit" or "radar module" shortages.

What Elon mentioned about writing firmware for replacement microcontrollers would not be applicable to the radar situation given they aren't replacing the microcontrollers inside the radar module (it's sold as a whole unit, so it's not like Continental would ship them ones without the chips), rather they are removing the whole radar module from the car (no replacement). So Elon is obviously talking about some other part in his comments.

I see however evidence of shortages of various control modules, infotainment and navigation systems, ADAS control modules (like ESP control modules, Mercedes Pre-safe). That's what I'm talking about.

OK, understood, but part of what I wrote is that controllers are also in the radar units, not necessarily separated into a controller module. To focus on controllers (aka MCUs, basically processors + some hardware I/O):
I searched just a little - here is a rather long article that mentions radar albeit in passing, in the context that nearly everything in the car uses MCUs, including ADAS features. The quote is emphasizing that the shortage is everything, not just increased ADAS content - but yes including that. And I think it's a safe bet that the radar module itself includes catalog-MCU or integrated custom-MCUs.

"But most of what we’re seeing the constraints in right now is MCUs, and that’s a lot of the legacy automotive stuff. It’s engine controls, transmission controls, steering, airbags, radios. It does touch some of the ADAS stuff — radars, front-view cameras. But the reason we’re having this problem is because the MCUs are literally everywhere in the car. It’s not like I’m just going to order a car and I’ll take some of the more modern features off, and then I’ll be able to get it. No, if you want the car to run, it’s got to have several microcontrollers in it. So it’s not the case that ADAS is the thing that’s causing this. It’s really the trend of putting more electronics in cars.”

Automotive IC Shortage Drags On

I would add that there are plenty of articles about RF semiconductor shortages also, as a whole post-Covid surge of 5G cellphones and infrastructure is ramping. This very probably affects the radar RF signal chain as well.

It's also easy to find articles about particularly bad delinquency / cost / profit problems at Continental, though they make many other things besides radars. The gist is that Continental did a particularly poor job of managing the supply disruption. If you negotiated particularly aggressive pricing and JIT terms based on your volume, but then you panic in a downturn and break the contract by refusing delivery, don't be surprised that your supplier doesn't put you right back to the front of the line in the upturn. This kind of thing happened across the auto industry, but Continental is apparently a prime example.

So I think we agree that as outsiders we don't have exact knowledge, and again I'm not arguing with you about what exactly is the supply situation and how it feeds Tesla's actions. But a lot of signs point to the conclusion that this is a likely factor. I posted that it made sense given the BMW radar story.

Like you, I also wouldn't tie this directly to Elon's comments about control-firmware rewrites. But it's reasonable to conclude there is an assortment of problems, some buried in assemblies that Tesla has no direct ability to fix. And I guess, if you stretch the interpretation, perhaps he could have been including the big AP rewrite in his explanation, thus obviating radar and thus obviating its need for an associated missing MCU. But I don't actually subscribe to that - it's like deep Biblical or Talmudic analysis over some comments that Elon, not a deity, made to get through a meeting with some analysts, a meeting that he wanted to be over so he could go back to doing something interesting.

 

[Matias]

I am not an aerospace engineer (I work in a field with a bit of overlap) but I know enough to realize Elon really knows his stuff. I am also not a machine learning expert (I know very little) but I know enough to realize Elon does not know his stuff. Guess what, it is totally consistent that somebody can know how to land a rocket on a ship, but be totally wrong about another field.

Yes. being a rocket scientist doesn’t have anything to do with being a ML expert.