Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register
This site may earn commission on affiliate links.
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.


screenshot-teslamotorsclub.com-2022.01.26-21_30_17.png


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).

 
Last edited:
I hear you made a “bad financial decision”! 🤣 What, you say I can’t believe everything I read in the internet?!
I actually wish I ran into more “naysayers” when I did my FSDj research. Instead I saw a slew of shill videos and posts and fell for it, ultimately wasting my money. I don’t necessarily regret it, but if I had to do it again, I would’ve done it differently.
 
I actually wish I ran into more “naysayers” when I did my FSDj research. Instead I saw a slew of shill videos and posts and fell for it, ultimately wasting my money. I don’t necessarily regret it, but if I had to do it again, I would’ve done it differently.
Agreed. Many “no interventions,” “mind blowing,” “100% improved,” yada yada yada videos and similar forum posts promoting FSD’s then current greatness and, either “canonizing” Elon, or outright declaring him ad a demigod….

Anyhoo, let’s hope v11.4 is something of an improvement…
 
Trying to find a definitive answer on how I can subscribe to and get FSD beta (city streets). We had a M3P last year and I subscribed and got it before a road trip which was great. We now have a MYLR and I recently subscribed, but never got the full FSD package. All I got was the EAP, and whatever else comes with that basic package. I have heard if your car's software version is to far ahead that is the reason you cannot get the FSD beta (city streets). I am currently on 2023.6.9 and there is a new update available 2023.12.5. If I keep denying the install for a while will I eventually be able to successfully subscribe to "FULL" FSD? How are we supposed to know which version and which version will allow the FSD beta? I find a lot of anecdotes and opinions but have not found any good source to know how to successfully get FSD beta. You would think Tesla would explain this somewhere.
 

Y’all might need to think this through and stop wasting my time, ’cause Omar can come back tomorrow. And the next day. And the next day, ya feel me?​


29202.jpg
For those on the other side, a "badge of courage" for speaking your mind (and the truth).

Note: The badge is color appropriate...

Enjoy! 🤣


Screenshot 2023-05-08 at 3.57.46 PM.png
 
Last edited:
Drop me a line where and when. Next weekend would be perfect - my Mon thru Thurs is pretty busy. Anywhere within 30 mins of the Seattle area is cool with me. I assume you all have GoPros etc to document and someone has put together a test plan and route?
Are you able to make it over to West Seattle? I have some great roads for you try (and fail) on! :D
 
Anyway, the whole idea of "end-to-end NN" is just another pie-in-the-sky.

Rules of the road can't all be inferred by NN - besides they change somewhat based on jurisdiction.

We all understand "end-to-end" to mean 1 gigantic neural network that takes camera data as inputs and outputs controls, but I wonder if Elon is just misusing it. Contextual, we know Elon was tweeting about replacing more of the control and planning code with neural nets:


So he might just mean perception, planning, and control each being primarily run by their respective neural networks. A network at one end, and a network on the other, but not really "end-to-end."
 
We all understand "end-to-end" to mean 1 gigantic neural network that takes camera data as inputs and outputs controls, but I wonder if Elon is just misusing it. Contextual, we know Elon was tweeting about replacing more of the control and planning code with neural nets:


So he might just mean perception, planning, and control each being primarily run by their respective neural networks. A network at one end, and a network on the other, but not really "end-to-end."
Agreed, Elon is more likely than not having a estranged relationship with the facts/truth here - just like his comments over the years that the Cybertruck is based on a structurally stressed skin exoskeleton, when in fact (as discussed by the Munro Live team) it is not…. But listening to him, it sure sounds cool and we should all line up and buy one or more… 🙄
 
Last edited:
So he might just mean perception, planning, and control each being primarily run by their respective neural networks. A network at one end, and a network on the other, but not really "end-to-end."
Interestingly they were earlier talking about Monte Carlo tree search to figure out the best path. They no longer talk about that. I guess whatever they are doing now has replaced it.

"Planner optimization using NN / Monte Carlo Tree Search (MCTS)"