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:
Good points. I always assumed that the vision models in FSD Beta were relatively camera-placement agnostic, but the fact that Tesla cannot just downscale HW4 cameras and run FSD Beta means the system as it stands isn't that flexible.
Why would you think HW4 doesn't support FSD Beta? It seems to run on it just as well/poorly as HW3.
 
Yes. Doesn't it seem weird that AP/NoA run just fine on HW4, with both the new cameras and HW4 FSD chip, but FSD beta doesn't? So obviously they can make them work, but it seems like they have chosen not to at this time. (Which isn't totally true, as there are some HW4 vehicles with FSDb.)
11.4.X runs fine on HW4. 11.3.6 doesn't support HW4.
 
Just a thought that occurred.

There's been some chaff out there about Tesla hawking its FSD software to other auto vendors. Not that any of them have reportedly bit, but, still.

The implication is that a Working FSD, at least a working ADAS, non Robo Taxi package, might be worth serious money to Tesla and a Leg Up to other auto manufacturers, making for a possible deal. So, Tesla might be very interested in getting a working FSD/ADAS out the door.

If this is true, let's take it a bit further and look at what it might mean for the software development process over at Tesla.

First off: It's one thing to keep on plotching continual less buggy releases upon the we-volunteered-more-or-less types that are currently running around with FSD-b (excluding @jebinc, of course), it's another if one is sending something like this to, say Ford. Ford's got lawyers, lots. And they would likely not take well to be jerked around, even if Tesla is doing their best.

So, feature sets or no feature sets, a working ADAS from Tesla sans bugs would be a very, very high priority. Higher than an ADAS with bugs, which is what we in the FSD-b party are playing with.

Conclusion: Tesla might be madly stiffing all known bugs, right, front, and center, to get to a reasonably salable product. And that might explain why it's been a long, long time since the last FSD-b general release.

Final note. I've read some recent reviews of Ford's Blue Drive and, I think, GM's equivalent. Compared to Tesla's offerings, they appear pretty blame pitiful.
 
Just a thought that occurred.

There's been some chaff out there about Tesla hawking its FSD software to other auto vendors. Not that any of them have reportedly bit, but, still.

The implication is that a Working FSD, at least a working ADAS, non Robo Taxi package, might be worth serious money to Tesla and a Leg Up to other auto manufacturers, making for a possible deal. So, Tesla might be very interested in getting a working FSD/ADAS out the door.

If this is true, let's take it a bit further and look at what it might mean for the software development process over at Tesla.

First off: It's one thing to keep on plotching continual less buggy releases upon the we-volunteered-more-or-less types that are currently running around with FSD-b (excluding @jebinc, of course), it's another if one is sending something like this to, say Ford. Ford's got lawyers, lots. And they would likely not take well to be jerked around, even if Tesla is doing their best.

So, feature sets or no feature sets, a working ADAS from Tesla sans bugs would be a very, very high priority. Higher than an ADAS with bugs, which is what we in the FSD-b party are playing with.

Conclusion: Tesla might be madly stiffing all known bugs, right, front, and center, to get to a reasonably salable product. And that might explain why it's been a long, long time since the last FSD-b general release.

Final note. I've read some recent reviews of Ford's Blue Drive and, I think, GM's equivalent. Compared to Tesla's offerings, they appear pretty blame pitiful.
If your conclusion is correct, Tesla has a long road ahead to completion (bug elimination).
 
  • Funny
  • Like
Reactions: Pdubs and kabin
Yes. Doesn't it seem weird that AP/NoA run just fine on HW4, with both the new cameras and HW4 FSD chip, but FSD beta doesn't? So obviously they can make them work, but it seems like they have chosen not to at this time. (Which isn't totally true, as there are some HW4 vehicles with FSDb.)
Got my new Model Y with HW4 this week and AP/NoA doesn't work as well as FSD did on my replaced MY with HW3. Lots of subtle regressions that I noticed on my first drives both on highway and secondary roads.
 
Based on the context of the post he replied to, I don't think he's really getting it. Also there have been no other posts regarding FSD by him since then but he's made other posts. False alarm in my opinion
Obviously this was not true since Elon stated "...within A week...". Clearly this week is not A week and A week is "two weeks" away. How could this be misread since Elon's XeyTweet was concise?