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:
In an NHTSA recall report about safety, why would a collision of any type be called a "warranty claim"? Seems odd.

It would be if the collision was caused by the defect in the recall.

A better question is why would the warranty claims be in the NHTSA recall report otherwise? What would be an actual warranty claim about software where nothing was damaged?



And if there was a collision related to these defects, wouldn't at least one of them have some injury involved, even if very minor?


Probably not? FSDb tends to operate at relatively low speeds, and Teslas are phenomenally safe cars.

Further- most of the accidents we know for a fact happen on fsdb are things like curbing your wheel or hitting a lane/parking pole.

Which would have no injuries, but damage to the vehicle.
 
Is TeslaBull an influencer? Where is WholeMars and Chuck Cook? They are almost always in the first batch. There must be others since not likely sent to one person. Hope we see some other videos today. So far don't see anything on YouTube.

That Blue line is some American SuperSized. Looks like a "Get the hell out of my way because I own this LANE". 🤣
 
Is TeslaBull an influencer? Where is WholeMars and Chuck Cook? They are almost always in the first batch. There must be others since not likely sent to one person. Hope we see some other videos today. So far don't see anything on YouTube.
I have noticed a pattern of seeing five or six cars on TeslaFi get the "influencer batch" installation. But so far TeslaFi only detects one car, and that one was on February 14.
 
From the highway to city streets video, looks like signalling to exit and taking forks is gone.

The visualization is also looking more and more like what Tesla showed us during Autonomy Day in 2019 (the test drive).

Yes, a one time scenario or?

Off ramp deceleration still appears rough with throttle-on to hard-braking and no noticeable regen. In general it looks like the AI team's prototypical robotic throttle/brake action hasn't been addressed. One would hope NN controls would add some smoothing.
 
  • Like
Reactions: AlanSubie4Life
The visualization is also looking more and more like what Tesla showed us during Autonomy Day in 2019 (the test drive).

Yeah red curb road edges are now a different shade of gray with the wider predicted path and lane visualization. I wonder if the blue line width helps avoid some jitter of the closer to raw output from before?

There's also a new blue highlight for relevant traffic lights instead of making them slightly more opaque. You can see it at the very end of the video along with the red stop line, which I suppose could also be a reason for a wider blue line to match the red.
 
Less than human like anticipation for cut-in at 1:00. The cut-in vehicle begins the lane change slightly before initiating the turn signals but ego is still late to acknowledge via lifting throttle. End result is at least 1sec delayed response with excess required ego braking. I should add at times this videos seems sped up and if so the mentioned 1 sec delayed response is even longer.

 
  • Like
Reactions: AlanSubie4Life
Less than human like anticipation for cut-in at 1:00. The cut-in vehicle begins the lane change slightly before initiating the turn signals but ego is still late to acknowledge via lifting throttle. End result is at least 1sec delayed response with excess required ego braking. I should add at times this videos seems sped up and if so the mentioned 1 sec delayed response is even longer.

Also missed the turning lane entrance at the end of the video…
 
Yeah red curb road edges are now a different shade of gray with the wider predicted path and lane visualization. I wonder if the blue line width helps avoid some jitter of the closer to raw output from before?

There's also a new blue highlight for relevant traffic lights instead of making them slightly more opaque. You can see it at the very end of the video along with the red stop line, which I suppose could also be a reason for a wider blue line to match the red.

The prior subtlety of making the relevant light slightly brighter probably sounded good on paper but the reality is that normal general public need things a lot more obvious. Nerds or obsessed tech folks would look for these fine differences. I recall when I first noticed my turning lane stoplight being brighter and had to stare at it a minute to convince myself that it was intentional and not a glitch.

v11 blue stoplights.jpg