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

How is FSD ever going to work with current camera layout?

This site may earn commission on affiliate links.
U better find that quote
5 minute google search and I found 3 examples of him bragging about how Tesla will acheive L5 in 2019 or 2020.

Jan. 27, 2021 quarterly earnings call "by the end of the year"

"I'm extremely confident that level 5 or essentially complete autonomy will happen and I think will happen very quickly," Musk said in the video. "I remain confident that we will have the basic functionality for level 5 autonomy complete this year."
 
Agreed! It’s the reason why Tesla is working on autopilot 4.0 hardware including 5mp cameras.


If we paid for FSD,...I hope we will be eligble to upgrade the cpu and cameras.... will older models have an issue? (legacy model x 2020) I hope not...I paid!

part of article..."There has also been no word on whether there will be an upgrade program, free or otherwise, for current FSD owners with HW3.0."

Rob
 
Last edited:
Go on YouTube and watch some drive videos of FSD Beta testers (TESLA EV owners). It is remarkable—not perfect, but it is working incredibly well on both highways and city streets. Remember: FSD does NOT need to be perfect; only better than a human driver and have notably fewer accidents/incidents. Not here there yet, but is is fascinating to see how far FSD has come in just handful of years.
That is the core of the issue: No one is saying that FSD is not _feasible_; it is just not ready for _prime/production_ and it seems it will not be for years. Tesla is behaving as if FSD is ready, which is not. Hence, all the issues we have.

Best approach would be to split the EV from AI/FSD. Then they can focus on what is important for mature product EV (volume, quality, service), what startups do FSD (beta testing, experimentation) and keep those separated and “graduate” only mature features. There are plenty of people who will help with the startup (e.g. FSD beta) and plenty of people who prefer to have a reliable, mature product. But those rarely mix.

This is a tried and true approach to innovation and it is beyond understanding why Tesla is mixing those.
 
If we paid for FSD,...I hope we will be eligble to upgrade the cpu and cameras.... will older models have an issue? (legacy model x 2020) I hope not...I paid!

part of article..."There has also been no word on whether there will be an upgrade program, free or otherwise, for current FSD owners with HW3.0."

Rob
It depends on when you bought the car/FSD.

When I purchased my car, the car configurator stated that purchasing FSD right now would include all necessary software and HARDWARE upgrades needed to run the final FSD product.

The language was later changed in the configurator to remove such promises. No, I don't know the exact date that the change in language was made.

Regardless, it would appear that Elon is on the hook to provide hardware upgrades for a large number of early FSD purchasers, myself included. Assuming, of course, that they're not able to get FSD version 1.0 running with promised features on the current hardware. :D
 
  • Like
Reactions: dbldwn02
That is the core of the issue: No one is saying that FSD is not _feasible_; it is just not ready for _prime/production_ and it seems it will not be for years. Tesla is behaving as if FSD is ready, which is not. Hence, all the issues we have.

Best approach would be to split the EV from AI/FSD. Then they can focus on what is important for mature product EV (volume, quality, service), what startups do FSD (beta testing, experimentation) and keep those separated and “graduate” only mature features. There are plenty of people who will help with the startup (e.g. FSD beta) and plenty of people who prefer to have a reliable, mature product. But those rarely mix.

This is a tried and true approach to innovation and it is beyond understanding why Tesla is mixing those.
From what I’m seeing on YouTube—considering how far FSD has come in just a handful of years—I don’t think L5 autonomy is far off. You do—fair enough. TESLA is not working to create the best EV’s. Everything that TESLA does is to advance the advent of the robotaxi fleet. Ultimately, TESLA will become an AI service provider, not a manufacturer of hardware available to the public. The reason that TESLA is working on a “million mile battery” is for its own long-term benefit, not for the benefit of TESLA owners (current or future).

Check out the ReThinkX Project. Founder Tony Seba employs a predictive model that imagines the mass-transformation of vehicle ownership, such that he asserts the rise of TaaS (Transport-as-a-Service), wherein individuals will spurn the high cost of vehicle ownership—especially ICE vehicles—opting instead to purchase a monthly subscription for a range of available travel miles to be provided by an autonomous robotaxi. Sadly, this plays right into WEF Director Clause Schwab’s sentiment that “You’ll own nothing and be happy,” which sentiment I wholeheartedly reject.
 
Last edited:
  • Funny
Reactions: 2101Guy
It depends on when you bought the car/FSD.

When I purchased my car, the car configurator stated that purchasing FSD right now would include all necessary software and HARDWARE upgrades needed to run the final FSD product.

The language was later changed in the configurator to remove such promises. No, I don't know the exact date that the change in language was made.

Regardless, it would appear that Elon is on the hook to provide hardware upgrades for a large number of early FSD purchasers, myself included. Assuming, of course, that they're not able to get FSD version 1.0 running with promised features on the current hardware. :D
Sad for Elon. How long has that been? 7yrs? 10yrs? Imagine the amortization on the amount of money paid for FSD by those early adopters. TESLA can afford it handsomely.
 
That is the core of the issue: No one is saying that FSD is not _feasible_; it is just not ready for _prime/production_ and it seems it will not be for years. Tesla is behaving as if FSD is ready, which is not. Hence, all the issues we have.
I got FSDb in the first round of Safety Score recipients. After thoroughly testing the initial version that was released to us (and several follow on versions), my thoughts very firmly aligned with yours.

Fast forward to today, however, and I'm not so sure. Due to personal events that I won't get into, I didn't drive my car for months. I missed a ton of releases. Two weeks ago, I started driving and using FSDb again. When I think back at how that initial version performed compared to what is currently installed in my car, I now find myself disagreeing with past me and current you. :)

Back then, I posted that I thought full on FSD was three to five years away. Current me says it's a lot closer that what past me thought. I'm thoroughly impressed with the capabilities of the current version, and IMO, we're now about a year away from obtaining 95% of driving conditions (at least here in the US).

But there are several things that must be accounted for when judging the accuracy of my guesses: 1. I've been married for 26 years. 2. I'm pretty sure past me was wrong. 3. I've been married for 26 years. That's all to say my guess of 95% in one year has a 95% chance of being completely wrong.

So why bother with this post? To point out the fact that NO ONE knows when it'll be done. Could be 2023, could be 2053. We just don't know. Any dart we throw at the calendar has no basis in fact.

Long story short: It's done when it's done.

Me? I'm just along for the historical journey that this will end up being in future history books (that is, if they still make books at that point).
 
Back then, I posted that I thought full on FSD was three to five years away. Current me says it's a lot closer that what past me thought. I'm thoroughly impressed with the capabilities of the current version, and IMO, we're now about a year away from obtaining 95% of driving conditions (at least here in the US).
Due to personal events that I won't get into, I didn't drive my car for months. I missed a ton of releases.
Quoting my own post... how dreadfully tacky! :D

So why would I breach forum etiquette in such an awful manner? Because I have a thought I'd like to express, and it's quite off topic from my earlier post.

Having gone so many months without driving my car came with a benefit: I didn't see the incremental progress of each FSDb release. I jumped forward in time, with a memory of how FSDb was performing at the time I stopped driving.

This gives me a bit of a unique perspective, having missed out on all those versions. And what a perspective it is!

I am utterly floored at how much better FSDb is now compared to how it was in April when I stopped driving my car. Early on in the beta, I was a bit (OK, very) dismayed by the lack of progress from version release to version release. But jumping forward in time like this has me excited again; the progress is real, and I find myself using FSDb all the time now (even with the wife or other passengers!), and not just by myself in very light traffic late at night.

I would rate its driving ability (in my area, at least) to now be on par with a 15 year old student driver, which is a massive improvement from the initial beta release I received.

I find myself excited again for FSDb... something I didn't think was going to happen again for years to come.
 
Last edited:
I got FSDb in the first round of Safety Score recipients. After thoroughly testing the initial version that was released to us (and several follow on versions), my thoughts very firmly aligned with yours.

Fast forward to today, however, and I'm not so sure. Due to personal events that I won't get into, I didn't drive my car for months. I missed a ton of releases. Two weeks ago, I started driving and using FSDb again. When I think back at how that initial version performed compared to what is currently installed in my car, I now find myself disagreeing with past me and current you. :)

Back then, I posted that I thought full on FSD was three to five years away. Current me says it's a lot closer that what past me thought. I'm thoroughly impressed with the capabilities of the current version, and IMO, we're now about a year away from obtaining 95% of driving conditions (at least here in the US).

But there are several things that must be accounted for when judging the accuracy of my guesses: 1. I've been married for 26 years. 2. I'm pretty sure past me was wrong. 3. I've been married for 26 years. That's all to say my guess of 95% in one year has a 95% chance of being completely wrong.

So why bother with this post? To point out the fact that NO ONE knows when it'll be done. Could be 2023, could be 2053. We just don't know. Any dart we throw at the calendar has no basis in fact.

Long story short: It's done when it's done.

Me? I'm just along for the historical journey that this will end up being in future history books (that is, if they still make books at that point).
Great post! LOL
Love the bit about being married 26yrs. Reminds me of the three ways to get fired from my former (Fortune 100) employer: Lying; stealing; and lying about stealing.

Back to FSD: I see FSDb performing exceptionally well in the YouTube videos published by FSDb testers. Where it seems to have trouble is old and arcane traffic patterns; new and arcane traffic patterns; and road construction, which we all can agree is often a dumpster fire of poor planning and even worse direction and execution—not to mention navigating AROUND an accident or road closure being directed by emergency services. I once got caught up after a bad accident on a two-lane mountain road outside of L.A. CHP was on-scene and directing traffic. The impatient cop who was trying to get me to do something—what, I did not know—kept pointing his flashlight directly at me, as if that would somehow imbue unto me powers of telepathy. My point is that if I didn’t know how to proceed past a wreck with human direction (and not even poorly planned road construction signage), then how is a fully autonomous EV going to know how to proceed? It seems evident to me that greater and newer standardization of road construction and incident response protocols are soon going to be needed. This could even be managed via over-the-air updates in real-time that provide relevant advisories to autonomous EV’s traveling to, in, or through the affected geography. For example, such instructions (in code) could include:
“Auto incident at X (latitude) and Y (longitude) coordinates. Number 1 and number 2 lanes shut down. Merge right to avoid traffic and incident debris. Proceed with caution.”
Autonomous EV’s could begin merging 1/2 mile out and reduce speed by 20mph in order to avoid the incident.
 
  • Like
Reactions: heapmaster
Quoting my own post... how dreadfully tacky! :D

So why would I breach forum etiquette in such an awful manner? Because I have a thought I'd like to express, and it's quite off topic from my earlier post.

Having gone so many months without driving my car came with a benefit: I didn't see the incremental progress of each FSDb release. I jumped forward in time, with a memory of how FSDb was performing at the time I stopped driving.

This gives me a bit of a unique perspective, having missed out on all those versions. And what a perspective it is!

I am utterly floored at how much better FSDb is now compared to how it was in April when I stopped driving my car. Early on in the beta, I was a bit (OK, very) dismayed by the lack of progress from version release to version release. But jumping forward in time like this has me excited again; the progress is real, and I find myself using FSDb all the time now (even with the wife or other passengers!), and not just by myself in very light traffic late at night.

I would rate its driving ability (in my area, at least) to now be on par with a 15 year old student driver, which is a massive improvement from the initial beta release I received.

I find myself excited again for FSDb... something I didn't think was going to happen again for years to come.
Which begs the question: Are we now experiencing a sort of “Moore’s Law”, in terms of FSD development and advancement over time?
 
Quoting my own post... how dreadfully tacky! :D

So why would I breach forum etiquette in such an awful manner? Because I have a thought I'd like to express, and it's quite off topic from my earlier post.

Having gone so many months without driving my car came with a benefit: I didn't see the incremental progress of each FSDb release. I jumped forward in time, with a memory of how FSDb was performing at the time I stopped driving.

This gives me a bit of a unique perspective, having missed out on all those versions. And what a perspective it is!

I am utterly floored at how much better FSDb is now compared to how it was in April when I stopped driving my car. Early on in the beta, I was a bit (OK, very) dismayed by the lack of progress from version release to version release. But jumping forward in time like this has me excited again; the progress is real, and I find myself using FSDb all the time now (even with the wife or other passengers!), and not just by myself in very light traffic late at night.

I would rate its driving ability (in my area, at least) to now be on par with a 15 year old student driver, which is a massive improvement from the initial beta release I received.

I find myself excited again for FSDb... something I didn't think was going to happen again for years to come.
Quite interesting! Comparing FSDb to 15yr old definitely gives me hope.

But my point was different. I really do hope that they achieve FSDb. When? It is anyone’s guess at this point (assuming that the path they follow could lead to it). That is exactly the reason why we have startup ventures - to explore new things, to test, and (hopefully) be successful. But we cannot rely on them for crucial things.

On the other hand, we have tools that may not be bleeding edge but they must be reliable. Mixing those two things inevitably lead to complexity and problems down the road.

@THEbuz had an interesting point about the future of transportation. Honestly, I had similar thought when V11 came out - how awful it was for the human driver and how suddenly everything made sense if one took the human out of the picture. Even if the auto wipers do not work for a human, who cares if a human is not driving? One would care even less if that is a rentable asset rather than their own car.

That may be the real strategy for Tesla (and would explain a lot). However, there are some serious issues:
- FSD requires change in the mindset of the whole population, not just people who decide to participate. So, even if the technology advances with light speed (some serious doubts there) it will take a long time for the mass adoption which Tesla needs for the model to work
- The shared economy is a bit of a fad. Yes, it has its place but it is not as widespread as proponents want us to believe. People like to own things (just look at the real estate market) and part of the shared economy craziness could be explained by the fact that Gen X simply cannot afford to own the things they want.

So, those two fundamental premises are highly uncertain, especially at the scale Tesla will need them if they really pursue that strategy.

On the other hand, it is quite possible that we simply deal with a runaway ego, which would also explain a lot :)
 
  • Like
Reactions: THEbuz
Quite interesting! Comparing FSDb to 15yr old definitely gives me hope.

But my point was different. I really do hope that they achieve FSDb. When? It is anyone’s guess at this point (assuming that the path they follow could lead to it). That is exactly the reason why we have startup ventures - to explore new things, to test, and (hopefully) be successful. But we cannot rely on them for crucial things.

On the other hand, we have tools that may not be bleeding edge but they must be reliable. Mixing those two things inevitably lead to complexity and problems down the road.

@THEbuz had an interesting point about the future of transportation. Honestly, I had similar thought when V11 came out - how awful it was for the human driver and how suddenly everything made sense if one took the human out of the picture. Even if the auto wipers do not work for a human, who cares if a human is not driving? One would care even less if that is a rentable asset rather than their own car.

That may be the real strategy for Tesla (and would explain a lot). However, there are some serious issues:
- FSD requires change in the mindset of the whole population, not just people who decide to participate. So, even if the technology advances with light speed (some serious doubts there) it will take a long time for the mass adoption which Tesla needs for the model to work
- The shared economy is a bit of a fad. Yes, it has its place but it is not as widespread as proponents want us to believe. People like to own things (just look at the real estate market) and part of the shared economy craziness could be explained by the fact that Gen X simply cannot afford to own the things they want.

So, those two fundamental premises are highly uncertain, especially at the scale Tesla will need them if they really pursue that strategy.

On the other hand, it is quite possible that we simply deal with a runaway ego, which would also explain a lot :)
@Boza, consider that Uber is also working to develop their own autonomous robo-taxi fleet; TESLA is not alone in this. History will refer to this season in time as the “Autonomy Wars”, documenting the many tech firms and automakers that have worked to achieve L5 autonomy during this time. Also, there is no denying that the advent of minimalist design and lifestyle adaptation is on the rise, along with hyper-utilitarianism. With the current model of vehicle ownership, vehicle utility is around 4% - 7% with vehicles being parked/garaged the rest of the time. This is gross waste that comes at a cost. (Personally, the idea of sharing a “borrowed” AEV with thousands of other riders is not at all attractive. I like owning my vehicle/s and storing a bit of my own crap in my vehicle/s.)

As to mass adoption, I believe the numbers will speak for themselves. Most people are frightened to death at the prospect of having “Hal” driving them around, but that is merely a stigma. AEV’s don’t have to be perfect. They only need to be safer than human drivers. When the statistical facts about AEV safety surpassing human driver safety are publicized—especially by the media—public perception will change in favor of AEV’s and mass adoption will have been achieved. (Not saying that autonomous vehicles are now safer than human drivers. They are not, but they eventually will be.)

Moreover, highway speeds will increase as the size of the AEV fleet increases. First, we will see an “AEV Lane”—a dedicated lane on the highway for AEV’s only. Following distances and incidence of unnecessary braking will decrease, leading to hyper-efficiency of lane usage, which leads to the aforementioned increased highway speeds. In city driving, we will begin to see stop signs and traffic lights disappear, as a growing AEV fleet will be able to operate amidst itself seemingly telepathically.

In short, it will become increasingly expensive for human drivers to own, operate, and maintain their own ICE vehicle(s) while simultaneously becoming increasingly difficult and dangerous to operate on the roadways due to the advent of the aforementioned changes in autonomy and vehicle operation. The value proposition of TaaS will become the obvious choice.
 
AEV lane is actually a great idea because it will simplify the FSD problem tremendously. If we could upgrade the existing fleet overnight we could solve the problem much sooner by completely removing the human from the system - but that is another example of unrealistic thinking.

“Also, there is no denying that the advent of minimalist design and lifestyle adaptation is on the rise, along with hyper-utilitarianism.”

Owning a home is even bigger “waste” but people still prefer it. You are not alone in your sentiment about your car - just poll this forum. Hyper-utilitarianism is a niche phenomenon.

There is a logic behind what you describe as near future but human societies are not so logical and there are so many dependencies that it will take a long time to be resolved. For example, the current highway speeds were established in the early 1970s. Our cars are much safer now - why those limits haven’t increased?
 
  • Like
Reactions: THEbuz
Hyper-utilitarianism is a niche phenomenon.
The same can be said for EVs. Trends matter more than status quo if you're trying to surmise what the future looks like.


For example, the current highway speeds were established in the early 1970s. Our cars are much safer now - why those limits haven’t increased?
not sure this is true. the interstate speed limits I remember from my childhood (55mph) are all now 70+mph.
 
The same can be said for EVs. Trends matter more than status quo if you're trying to surmise what the future looks like.



not sure this is true. the interstate speed limits I remember from my childhood (55mph) are all now 70+mph.
Here in Idaho, our 75mph Interstate speed limit was raised to 80mph about 8-10yrs ago. People routinely drive 85+ mph. I, myself, set the cruise control at 87mph and almost never see a cop. (I am referring to stretches of the Interstate that are outside of the populated area—on either end of the valley where most people live.)
 
...not sure this is true. the interstate speed limits I remember from my childhood (55mph) are all now 70+mph.
What happened was that a 55 mph US national speed limit was imposed in 1974, as a measure to deal with a major gasoline price and availability squeeze stemming from the OPEC oil embargo. Prior to that, speed limits were set locally and we're much higher on many roads (including of course many undivided rural highways where the passing lane is also the opposite oncoming lane).

I got my license in 1976 and expected/hoped for the opportunity to drive faster without resorting to CB radio speed-trap reports, or monitoring the warning high beam flash of helpful opposite-lane drivers. :) For better or worse, this never came for me until I was already an adult and a (somewhat) responsible father.

After the energy crisis subsided, there was an ongoing national dispute about raising or or abolishing the 55 mph limit. It became not an energy-focused issue, but a political and regulatory control debate ostensibly in the name of safety.

Proponents of keeping 55 mph cited statistics of reduced fatalities after 1974, and predicted a return to dark days of highway mayhem. Opponents pointed out the disparate effect on highway travel in the less populated regions outside the Eastern seaboard and other metropolitan areas.

I believe it was 1995 (?) that Congress finally abolished the 55 mph limit. The predicted "carnage on the highways" never materialized, somewhat because of evolved automotive engineering but mostly because of the vastly increased usage of mandatory shoulder restraints and airbags.
 
My MY with "FSD" turns off AP with "bad weather detected" when I spray windshield washer on the windshield... If I just used the spray button (which runs through the computer) and right after that it sees "rain" on the windshield (detected by the same computer) and it can't connect the dots... it doesn't take a lot of intelligence, artificial or otherwise, to figure out what just happened, but my car doesn't have it, so no, I don't think it's ready.

I have been driving this "FSD" MY for 15 months now. Summon worked exactly zero times. The only capability of Summon seems to be to display various error messages and stop, in any scenario I tried. Autopark worked ONCE. I was ecstatic... it actually found a parking spot and parked itself! Yes, once in 15 months. I really don't understand what I got for my money when I paid for FSD vs EAP. Don't get me wrong, I love the car, but if I would buy one today, I would absolutely not pay for FSD, and I see it as a waste of money - if you are in Europe, the only real life difference between EAP and FSD at the moment is a line of text on the specs screen.
 
Last edited:
I would absolutely not pay for FSD, and I see it as a waste of money - if you are in Europe, the only real life difference between EAP and FSD at the moment is a line of text on the specs screen.
Pretty much the same in the US although we do get Navigate on Autopilot which has been known to work on occasion. I will admit AutoPark is pretty solid here nowadays after years of being more of a joke.
 
From what I’m seeing on YouTube—considering how far FSD has come in just a handful of years—I don’t think L5 autonomy is far off. You do—fair enough.
To make sure I understand correctly, you believe L5 responsibility isn't far off?

Personally, I think it'll eventually drive extremely well in ideal conditions, and I think if we're lucky, we might get L3 responsibility on highways.

I have a difficult time believing it'll drive so well that Musk would be willing to let Tesla take on the financial responsibility for all of FSD's actions, though. There are just too many edge cases.

Because don't forget, that's what L5 is. It's not how well it drives. It's that they're driving and are financial responsibility for the car's actions, not you, not the owner of the vehicle, and not anyone sitting in the vehicle.
 
  • Like
Reactions: Boza