FSD Beta v11.x

The next big milestone for FSD is 11. It is a significant upgrade and fundamental changes to several parts of the FSD stack including totally new way to train the perception NN.

From AI day and Lex Fridman interview we have a good sense of what might be included.

- Object permanence both temporal and spatial
- Moving from “bag of points” to objects in NN
- Creating a 3D vector representation of the environment all in NN
- Planner optimization using NN / Monte Carlo Tree Search (MCTS)
- Change from processed images to “photon count” / raw image
- Change from single image perception to surround video
- Merging of city, highway and parking lot stacks a.k.a. Single Stack

Lex Fridman Interview of Elon. Starting with FSD related topics.

Here is a detailed explanation of Beta 11 in "layman's language" by James Douma, interview done after Lex Podcast.

Here is the AI Day explanation by in 4 parts.

Deep Understanding Tesla FSD Part 1: HydraNet

From Theory to Reality, Analyze the Evolution of Tesla Full Self-Driving

saneryee-studio.medium.com

Here is a useful blog post asking a few questions to Tesla about AI day. The useful part comes in comparison of Tesla's methods with Waymo and others (detailed papers linked).

Twelve Questions about Tesla Autopilot Technologies Exposed at 2021 AI Day

On Aug. 19 2021, Telsa held an AI day , exposing more technologies about Autopilot , from perception, planning, data labeling, simulation…

yuhuang-63908.medium.com

 

[Supcom]

I have no need for FSD.

Nobody said that you did. That is strictly up to you. But you framed your argument as if nobody could find value in vehicle autonomy unless it generated cash income.

There are plenty of other possible use cases that people other than you might find valuable.

For some, an autonomous vehicle dropping them off and picking them up at a restaurant, concert or sporting event where parking is inconvenient, or expensive might have value - especially if they have limited mobility and cannot walk long distances.

Some people have busy schedules and might like to send their car to pick up groceries, restaurant meals, or even their package deliveries.

Some people may find value in having the car drive them to the doctor when they are ill and feeling poorly.

Perhaps apps could be made for the car to act as a tour guide, driving you around with audio and video presentations on points of interest along a route.

I expect that autonomous racing cars will become a thing one day. At some point, it might be possible for individuals to customize their cars using an AV development kit to make them suitable for racing.

 

[MARKM3]

Show me a Corvette that sold for 15k more with the painted engine package or an F150 that sold for 10k more for the luxury/technology package. These upgrades rarely translate to much in resale.

Your X has already lost 30%+ in value.

Again, when you are talking 100k on a car you aren't working with much logic in "basic economics".

So there is more likely an attachment rate on the 50K+ cars, but I agree with you it doesn't with the low cost cars.

Edit: None of these rules applied during the pandemic. I sold my first Tesla with FSD for more than I paid for it. Same with my F150.

It's like adding a swimming pool to your backyard. Many say it doesn't really increase the value of your house at all and can even limit buyers, but you enjoy it and you may find someone who will pay more because it's there.

Basic economics apply to all purchases, supply/demand is elastic and to price. If they don’t apply, why don’t you Grubhub your dinners every night? You’d have so much more time to do other stuff?

FSD is literally the same thing. We don’t Grubhub food every night because of our financial constraints and other various factors. Just like I won’t be buying FSD because we just wouldn’t use it enough to justify cost.

 

[EVNow]

Why would I want to make my commute $10 more expensive every day when I can give that $10 or $200 to my local Methodist church?

For the same reason you didn't buy a $50k car and give $50k to charity (like I literally did) ?

You know best what works for you - but you can't generalize based on that.

We'll know for sure when we get there but I think people always prefer convenience. So I think it's likely people will buy cheaper car with FSD than an expensive one without - I mean have you seen anyone pay a lot more for less convenience?

 

[aronth5]

No - sharing a vehicle within a family is markedly different from using it commercially. I could drive to work, then send the car home so it was still available. likewise, if my kid needs to be picked up after school and I'm still at work the car could do it.

That would be interesting where I live. To pick up your kids at the elementary school you have to get in a special line that snakes around the building and you need an approved number specific to your family that you display for the school employee to see. Just trying to figure out where the line is since it varies based on when you show up is another challenge.

 

[EVNow]

Just trying to figure out where the line is since it varies based on when you show up is another challenge.

By the time AVs are common - they'll have solutions for this, since it's an important use case.

 

[Yelobird]

Basic economics apply to all purchases, supply/demand is elastic and to price. If they don’t apply, why don’t you Grubhub your dinners every night? You’d have so much more time to do other stuff?

FSD is literally the same thing. We don’t Grubhub food every night because of our financial constraints and other various factors. Just like I won’t be buying FSD because we just wouldn’t use it enough to justify cost.

Just an observation but for someone with absolutely No interest in FSD you certainly spend a LOT of hours posting about it lol. I’m not a fan of the Model X nor do I see a rational justification for the double price vs a MY yet I don’t believe I used one minute of my life to droll on in the forum why my lack of interest must be right. Lol. End of day, everyone has a different interest and justification. If you feel obligated to continuously sell your opinion you may actually have an interest….

 

[Knightshade]

And knightshade, I feel we are in the minority of American's with our daily commute. We have people at my work that commute 30 minutes, 1 hour, 45 minutes, 30 minutes, etc.

You live in Raleigh and have people with 1 hour commutes?

FWIW the average commute in Raleigh, pre-covid, was under 30 minutes. Should be even less now with a significantly higher % of people working from home.

Very few live close to work, so now you have an even LONGER empty leg where now not only are you paying $10 per day for FSD, but now you're paying an additional $5 in electricity per day

Of course you're saving significantly more than that not having insurance or a car payment on a second car. And again quite a lot of people have both shorter drives than you suggest, and ones where both people could get in the car from home and one is just dropped off a bit later than the other so no it won't be 3x the miles nor any long empty leg for most either.

Oh, and if an extra 20 minutes is $5 in electricity you need to find better electricity. My bill (on the west side of the triangle rather than the east) was about $2 per 300 miles of driving via ToU billing.

 

[MARKM3]

You live in Raleigh and have people with 1 hour commutes?

FWIW the average commute in Raleigh, pre-covid, was under 30 minutes. Should be even less now with a significantly higher % of people working from home.





Of course you're saving significantly more than that not having insurance or a car payment on a second car. And again quite a lot of people have both shorter drives than you suggest, and ones where both people could get in the car from home and one is just dropped off a bit later than the other so no it won't be 3x the miles nor any long empty leg for most either.

Oh, and if an extra 20 minutes is $5 in electricity you need to find better electricity. My bill (on the west side of the triangle rather than the east) was about $2 per 300 miles of driving via ToU billing.

Don’t know how you’re only paying what, .03c/kw ? But I pay $57 on average per month in electricity for my Model X. I did the math for my wife’s and I commute and mileage would easily triple with all the empty legs, so would go up to $180 per month.

So a $120 increase in electricity and $200 FSD bill is $320 more per month. That is easily the payment for a car for my wife.

The economics for me don’t make sense.

 

[Twiglett]

You live in Raleigh and have people with 1 hour commutes?

FWIW the average commute in Raleigh, pre-covid, was under 30 minutes. Should be even less now with a significantly higher % of people working from home.

not the same location I know, but in Austin my pre-covid commute was 35 minutes, now its 55mins on a good day. It's like everyone forgot how to drive and are now aggressive/self centered/doing their own thing instead of trying to get to work.

 

[MARKM3]

Just an observation but for someone with absolutely No interest in FSD you certainly spend a LOT of hours posting about it lol. I’m not a fan of the Model X nor do I see a rational justification for the double price vs a MY yet I don’t believe I used one minute of my life to droll on in the forum why my lack of interest must be right. Lol. End of day, everyone has a different interest and justification. If you feel obligated to continuously sell your opinion you may actually have an interest….

Because I was once excited about it, and I think once everyone goes down the rabbit hole of real world practicality most will come to the same conclusion - unless the price is heavily discounted, it's just not really worth it. I'd like to see the price for non commercial use around $2-3k per car.

 

[SSDTester#1]

Absolutely and not even a question. Try asking someone who can’t physically drive what they would pay/do for freedom to venture out freely.

Use a rideshare app?

 

[MARKM3]

Use a rideshare app?

My grandparents don't even have a smartphone, let alone would have any idea how to use FSD in a tesla . So that use case is just null and void. It's so funny what people come up with without actually going through what solutions we have today for those same problems at a much lower cost than $15,000.

A rideshare app is a much better approach for them, or hell, call a taxi

 

[SSDTester#1]

For the same reason you didn't buy a $50k car and give $50k to charity (like I literally did) ?

You know best what works for you - but you can't generalize based on that.

We'll know for sure when we get there but I think people always prefer convenience. So I think it's likely people will buy cheaper car with FSD than an expensive one without - I mean have you seen anyone pay a lot more for less convenience?

The junk is not convenient. Not today. Maybe one day, but not today. It requires more effort and supervision than manual driving.

 

[uscbucsfan]

Basic economics apply to all purchases, supply/demand is elastic and to price. If they don’t apply, why don’t you Grubhub your dinners every night? You’d have so much more time to do other stuff?

FSD is literally the same thing. We don’t Grubhub food every night because of our financial constraints and other various factors. Just like I won’t be buying FSD because we just wouldn’t use it enough to justify cost.

The Model S/X have relatively low demand compared to other cars and other Teslas, but we aren't talking about the general populace in this specific conversation. We are talking about the ones willing to spend more a higher amount on a car.

You thought it was beneficial to spend 100K on a car. I did, too, but I also found FSD beneficial as have 400k others in NA. Most of those aren't looking for robotaxi or a financial return on their value, but they do want actual FSD.

Again, you make sense when dealing in lower priced cars. I agreed with your initial comment that unless FSD could be finished the extreme low end sales would be almost 0 and the take rate overall would stay flat. That return on value with 15k/200 per month is debatable depending on the individual. Obviously, most here would say a car that can actually drive itself without intervention, even without being able to make additional money, would provide that value.

There's a reason why the most expensive upgrades are typically purchased and offered on higher priced cars, regardless of them adding actual monetary value to the car. Value is relative. Comfort, safety, ease of use are all things of value that extend beyond adding monetary value.

edit: It's .05c/kwh in SC.

 

[MARKM3]

The Model S/X have relatively low demand compared to other cars and other Teslas, but we aren't talking about the general populace in this specific conversation. We are talking about the ones willing to spend more a higher amount on a car.

You thought it was beneficial to spend 100K on a car. I did, too, but I also found FSD beneficial as have 400k others in NA. Most of those aren't looking for robotaxi or a financial return on their value, but they do want actual FSD.

Again, you make sense when dealing in lower priced cars. I agreed with your initial comment that unless FSD could be finished the extreme low end sales would be almost 0 and the take rate overall would stay flat. That return on value with 15k/200 per month is debatable depending on the individual. Obviously, most here would say a car that can actually drive itself without intervention, even without being able to make additional money, would provide that value.

There's a reason why the most expensive upgrades are typically purchased and offered on higher priced cars, regardless of them adding actual monetary value to the car. Value is relative. Comfort, safety, ease of use are all things of value that extend beyond adding monetary value.

edit: It's .05c/kwh in SC.

I think that is a great summarization. We all have our own value propositions to justify capital expenditures, and at the end of the day we make sense out of them

My only point was to get us thinking about how practical FSD will really be to each individual. Sharing my experience with my wife I think helps.

 

[sleepydoc]

If the justification was solely based on You and Now sure. How about if you or a family member had a disability? How about several years from now when you get older and your freedom to Safely do simple travel pleasures are in question or possibly dangerous? Your parents, others? What makes sense to you and your use case now is irrelevant to the world demographic or use case. I would Gladly pay 200$ per month to allow my disabled grandchild (true) the freedom to enjoy a life outside of a house. Same for my aging parents. Some plans don’t require economic sense to be valuable to others.

Forget that. In a previous job I had a 30-40 minute commute each way. How much would it be worth for me to have an hour a day where I could read, shop online, answer emails, etc instead of driving?

edit: here’s another use case - I go to a bar and have several beverages. How much is it worth to have a car that can safely drive me home without getting a DWI or worse, killing someone?

 

[sleepydoc]

1. Tesla never had FSD on the Interstate, so it's relevant they combine them.
2. Mercedes doesn't have ADAS on city streets.

Do you think Waymo or Cruise will separate them when they integrate interstate driving which is imminent according to Diplomat?

Why would they? It's relevant to total miles and miles without disengagement. If it's not a different stack, it's using the same logic and still dealing with complexities of each. It made sense to separate them when the highway/interstate was using AP instead of FSD, but not anymore.

It's just a dumb point to harp on.

Ok. so now you’re contradiciting yourself and arguing with points i didn’t make. I’ll just bow out and let you argue with yourself. It’ll be more entertaining.

 

[Knightshade]

not the same location I know, but in Austin my pre-covid commute was 35 minutes, now its 55mins on a good day. It's like everyone forgot how to drive and are now aggressive/self centered/doing their own thing instead of trying to get to work.

FWIW, I just checked, latest post-covid data is still under 30 minutes average for Raleigh.

Don’t know how you’re only paying what, .03c/kw ?

Slightly less--- it's ToU- rate is 2.79c per kwh overnight when you're likely to actually charge the car.

There's a 4-5 hour "peak" period that's like 26 cents or something, usually during hours you'd be at work anyway.

Rest of the time, and 24/7 weekends and holidays, it's just over 6 cents per kwh.

https://pemc.coop/wp-content/uploads/2023/04/Piedmont_Rates_2023-05-01-1.pdf

Page 27 for the plug-in EV ToU rates.



My electric bill went down compared to before I had an EV thanks to the ToU rate available to EV owners.


Anyway as others point out- a longer drive ALSO means you now have all that time to be doing something that is NOT driving that you'd otherwise waste on driving.

 

[sleepydoc]

ps : I don't expect FSD to be priced at $200 if and when it is actually available. It would definitely be more.

If and when it becomes available the price will be determined by supply and demand. Well, just demand.

 

[MARKM3]

FWIW, I just checked, latest post-covid data is still under 30 minutes average for Raleigh.






Slightly less--- it's ToU- rate is 2.79c per kwh overnight when you're likely to actually charge the car.

There's a 4-5 hour "peak" period that's like 26 cents or something, usually during hours you'd be at work anyway.

Rest of the time, and 24/7 weekends and holidays, it's just over 6 cents per kwh.

https://pemc.coop/wp-content/uploads/2023/04/Piedmont_Rates_2023-05-01-1.pdf

Page 27 for the plug-in EV ToU rates.



My electric bill went down compared to before I had an EV thanks to the ToU rate available to EV owners.


Anyway as others point out- a longer drive ALSO means you now have all that time to be doing something that is NOT driving that you'd otherwise waste on driving.

Understood, I'm on no such plan nor do I know if I have the option to be on that plan. Also, I no longer live in Raleigh - I live further east closer to the beach. Will have to check into the TOU plan.