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

Tesla replacing ultrasonic sensors with Tesla Vision

This site may earn commission on affiliate links.
Andrej Karpathy speaking about this is what gave me confidence in the Tesla Vision and removal of uss.

Needing or being forced to remove USS vs being able to remove them one day when the time is right isn't quite the same.

He also seems to have very high praise for tesla and his work there.

His administrative work?

he became an administrator and director

If Tesla were following his (AK's) advice / guidance (presumably dropping USS, radar and any other superfluous sensors), and everything was going along just fine, then why would he leave? Surely either they weren't following the path he wanted, or he didn't have the role he wanted..... in which case, if everything else is roses, why wouldn't he be given the position he wanted? You'd think all things equal Tesla would want to keep him on board.

Imo there must have been something that looked / felt more like a stick than a 🥕!

If Tesla keep USS active on existing fitted cars, then presumably they must still serve a purpose, otherwise who would care? Removing something like USS before there is a working replacement doesn't seem like something you'd do by design or choice.
 
Last edited:
Needing to remove USS vs being able to remove isn't quite the same.



His administrative work?



If Tesla were following his (AK's) as device / guidance (presumably dropping USS, radar and any other superfluous sensors), and everything was going along just fine, then why would he leave? Surely either they weren't following the path he wanted, or he didn't have the role he wanted..... in which case, if everything else is roses, why wouldn't he be given the position he wanted?

Imo there must have been something that looked / felt more like a stick than a 🥕!

If Tesla keep USS active on existing fitted cars, then presumably they will still serve a purpose, otherwise who would care? Removing something like USS before there is a working replacement doesn't seem like something you'd do by design or choice.
sounds like questions and assumptions you can ask him.
 
People seem to give Elon & Tesla so much latitude and benefit of the doubt, and I'm not entirely sure what it's based on?

Eventually this stuff comes to market? Well, great.

I get strong "you don't need a charger so we're not supplying one" Apple vibes from what Tesla does lately, with consumers (fanboys) supportive of having things taken away from them, making excuses for massive profit making entities. At least with Apple you can actually buy another charger to replace the one you don't get anymore (even though the price hasn't reduced to reflect it). You can't even get ultrasonics on new cars even if you wanted to.
 
People seem to give Elon & Tesla so much latitude and benefit of the doubt, and I'm not entirely sure what it's based on?

Eventually this stuff comes to market? Well, great.

I get strong "you don't need a charger so we're not supplying one" Apple vibes from what Tesla does lately, with consumers (fanboys) supportive of having things taken away from them, making excuses for massive profit making entities. At least with Apple you can actually buy another charger to replace the one you don't get anymore (even though the price hasn't reduced to reflect it). You can't even get ultrasonics on new cars even if you wanted to.
The thinking for me is this:
-TV will give you all the features of USS, with better reliability and function.
-It will improve FSD and driving experience
-This will come, according to Tesla, with a major software update (which I've read is coming this month, though I have nothing factual)
-It is unfortunate that Tesla did not make this a seamless transition before taking away the USS,
this gap in features is what is making people upset.

The 'benefit of the doubt' is not accurate. "No choice but to wait for what they have officially stated" is more accurate.

As for chargers, VW and others have done the same thing. Chargers were included and, without notice, stopped supplying them with cars.
Tesla honored free chargers to those that ordered before that cutoff was published. VW did not.
 
  • Like
Reactions: Yelobird
If Tesla were following his (AK's) advice / guidance (presumably dropping USS, radar and any other superfluous sensors), and everything was going along just fine, then why would he leave? Surely either they weren't following the path he wanted, or he didn't have the role he wanted..... in which case, if everything else is roses, why wouldn't he be given the position he wanted? You'd think all things equal Tesla would want to keep him on board.
Probably lots of reasons. One might be FSD transitioning from an academic-like research project to a production project needing to satisfy hundreds of thousands of drivers. It's all good fun doing research and teaching but he isn't wired for the stressors of FSD production. He can always come back as a part time research consultant.
 
  • Like
Reactions: Eto Demerzel
The thinking for me is this:
-TV will give you all the features of USS, with better reliability and function.
-It will improve FSD and driving experience
-This will come, according to Tesla, with a major software update (which I've read is coming this month, though I have nothing factual)
-It is unfortunate that Tesla did not make this a seamless transition before taking away the USS,
this gap in features is what is making people upset.

The 'benefit of the doubt' is not accurate. "No choice but to wait for what they have officially stated" is more accurate.

As for chargers, VW and others have done the same thing. Chargers were included and, without notice, stopped supplying them with cars.
Tesla honored free chargers to those that ordered before that cutoff was published. VW did not.
If Tesla were known for delivering things when they said, and they performed like they promised, then I doubt anyone would have a problem. The sad reality is Tesla have never delivered when they said, and the initial versions have never performed like they promised. Hence why people are unhappy.
 
If Tesla were known for delivering things when they said, and they performed like they promised, then I doubt anyone would have a problem. The sad reality is Tesla have never delivered when they said, and the initial versions have never performed like they promised. Hence why people are unhappy.
Doesn't contradict anything I've stated. I'm not happy about missing features either.
 
  • Like
Reactions: jeewee3000
-TV will give you all the features of USS, with better reliability and function.
Aside from reducing production part count I'm not sure how one comes to that conclusion especially when the cameras aren't covering important bumper space around the car.

Ultrasonic sensors are simple, reliable, inexpensive, provide good resolution at short distances, consumes low energy, no need for NN, no need to add auto labels or store occupancies into NV memory, blends-in well with the vehicle, requires minimal processing power and memory.

Most might not feel comfortable with anything less than five 9's reliability in an around the car which has been an impossibility with vision only. And that's all I have to say about that. :)
 
Aside from reducing production part count I'm not sure how one comes to that conclusion especially when the cameras aren't covering important bumper space around the car.

Ultrasonic sensors are simple, reliable, inexpensive, provide good resolution at short distances, consumes low energy, no need for NN, no need to label/identify/store occupancies into NV memory, blends-in well with the vehicle, requires minimal processing power and memory.

Most might not feel comfortable with anything less than five 9's reliability in an around the car which has been an impossibility with vision only. And that's all I have to say about that. :)
I felt the same. Simple cost cutting.
It is a large cost savings, and i'm sure factored into the decision.

The reason I feel differently is hearing Andrej Karpathy speaking about this subject. It's a different mindset and different way of thinking of how to tackle the 3d world with sensors. At least listening to his logic, it's a superior way.

Lets see how it is implemented. Hopefully it's at least as useful as uss, but seems far more plausible that it will offer much more than uss.
At least, that's my hope based on Karpathy's reasoning behind it.
 
plausible that it will offer much more than uss.
At least, that's my hope based on Karpathy's reasoning behind it.

That alone doesn't really justify removal of USS, certainly not before you have a fully developed alternative.

That said, AK is a smart cookie and I also enjoyed hearing is passion and confidence that a VO / TV solution will deliver.
 
  • Like
Reactions: Eto Demerzel
Aside from reducing production part count I'm not sure how one comes to that conclusion especially when the cameras aren't covering important bumper space around the car.

Ultrasonic sensors are simple, reliable, inexpensive, provide good resolution at short distances, consumes low energy, no need for NN, no need to add auto labels or store occupancies into NV memory, blends-in well with the vehicle, requires minimal processing power and memory.

Most might not feel comfortable with anything less than five 9's reliability in an around the car which has been an impossibility with vision only. And that's all I have to say about that. :)
It's been done to death but I've heard of a Tesla customer being asked to supply the sensors from their damaged bumper when a new one has been sourced, and other manufacturers advising of delays with ultrasonics.

In my opinion this is - at best - 90% because of parts shortages and Tesla not wanting to stop the production line and 10% about a future goal of Tesla Vision replacing USS.

If it were another manufacturer deciding to supply cars without ultrasonics they'd get ridiculed, but as it's Tesla people just give them the benefit of the doubt in perpetuity that some day this and other features will work. Maybe it will, maybe it'll take months or years, maybe whatever gets done with Tesla Vision in terms of simulating USS will never be as accurate as USS was, which when we're talking about people relying on it is worrisome. I have strong doubts that "occupancy network" will ever work as reliably as USS does.

To be fair to Tesla, however, if I were in their position I'd probably do the same in terms of removing these parts and increasing my margin per car. Why wouldn't they? It doesn't seem to affect their sales, people are still buying the cars even with them subtracting features (passenger lumbar, data in USB ports, USS, maybe other things I've forgotten). Until such time as it actually does have an impact I expect the value engineering to continue.
 
Aside from reducing production part count I'm not sure how one comes to that conclusion especially when the cameras aren't covering important bumper space around the car.

Ultrasonic sensors are simple, reliable, inexpensive, provide good resolution at short distances, consumes low energy, no need for NN, no need to add auto labels or store occupancies into NV memory, blends-in well with the vehicle, requires minimal processing power and memory.

Most might not feel comfortable with anything less than five 9's reliability in an around the car which has been an impossibility with vision only. And that's all I have to say about that. :)
Given how quickly the rear camera on my Y gets dirty during the winter I'm comfortable saying that Tesla Vision will NOT equal the performance of ultrasonic sensors in the rear and I challenge anyone to explain how they could.
 
Given how quickly the rear camera on my Y gets dirty during the winter I'm comfortable saying that Tesla Vision will NOT equal the performance of ultrasonic sensors in the rear and I challenge anyone to explain how they could.
Forget varsity stuff like slush and mud - it's practically useless backing up in just plain ole rain.
 
Given how quickly the rear camera on my Y gets dirty during the winter I'm comfortable saying that Tesla Vision will NOT equal the performance of ultrasonic sensors in the rear and I challenge anyone to explain how they could.
What dirt?

20221127_180356.jpg
 
It's been done to death but I've heard of a Tesla customer being asked to supply the sensors from their damaged bumper when a new one has been sourced, and other manufacturers advising of delays with ultrasonics.

In my opinion this is - at best - 90% because of parts shortages and Tesla not wanting to stop the production line and 10% about a future goal of Tesla Vision replacing USS.

If it were another manufacturer deciding to supply cars without ultrasonics they'd get ridiculed, but as it's Tesla people just give them the benefit of the doubt in perpetuity that some day this and other features will work. Maybe it will, maybe it'll take months or years, maybe whatever gets done with Tesla Vision in terms of simulating USS will never be as accurate as USS was, which when we're talking about people relying on it is worrisome. I have strong doubts that "occupancy network" will ever work as reliably as USS does.

To be fair to Tesla, however, if I were in their position I'd probably do the same in terms of removing these parts and increasing my margin per car. Why wouldn't they? It doesn't seem to affect their sales, people are still buying the cars even with them subtracting features (passenger lumbar, data in USB ports, USS, maybe other things I've forgotten). Until such time as it actually does have an impact I expect the value engineering to continue.
To be fair, I don't think anyone states Tesla is making the right moves here by removing USS before there is Tesla Vision software functionality ready.

Nobody likes it, especially those who just took delivery of a non-USS-Tesla vehicle (like me).

But of course we can just hope Tesla will restore functionality soon (i.e. within weeks, not months). I just hope HW3 can deliver on this, I'm still not convinced.
 
  • Like
Reactions: Durzel
-This will come, according to Tesla, with a major software update (which I've read is coming this month, though I have nothing factual)
May I ask where you've read this?

I've been scouring the internet for leaks and info on non-USS software updates, but the only nugget of info I found comes from hacker "green" his twitter account in which he stated in the beginning of November that the newest update contains a hint of Autopilot distances being transformed in virtual USS functionality. Might run in shadow mode, might not.

But I've never heard "this month". Would love if it were true.
 
  • Like
Reactions: Matias
May I ask where you've read this?

I've been scouring the internet for leaks and info on non-USS software updates, but the only nugget of info I found comes from hacker "green" his twitter account in which he stated in the beginning of November that the newest update contains a hint of Autopilot distances being transformed in virtual USS functionality. Might run in shadow mode, might not.

But I've never heard "this month". Would love if it were true.
Honestly can't remember. I read that a "major update" was coming in december and that parking sensor capability would return.
But it wasn't Tesla *that said December, and probably more rumor mill stuff. It stuck in my mind, so must have been more than just some forum gossip.

Sorry it's nothing concrete.
 
  • Helpful
Reactions: jeewee3000
dont expect it ANYTIME soon.
We are close to coming up on a year from when Elon promised to allow horn activate via airbag on Yokes via over the air update.

And yet..here we are....
No offence, but
a) people with yokes are a minority and shouldn't be complaining since they're driving the best cars on the road;
b) the occupancy network that is supposed to provide the data on which Tesla Vision parking functionality will depend on is actually there, they just need to improve it/test it before they roll it out to wide release.

The latest FSD beta uses these occupancy networks and can estimate the surroundings pretty nicely, so it seems most of the pieces of the puzzle are there.
And also: every month that goes by a gigantic amount of people take delivery of more non-USS Y's/3's. The amount of complaints Tesla will receive will rise exponentially. So they will prioritize Tesla Vision over people that want easier horn access.