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:
Speaking of following distances.. I happen to have a ringer in my corner. She’s an Ergonomics Engineer, with both a Bachelor’s and Master’s to prove it. For those not in the know, those in that field have to contend with a human in the control loop. I don’t do humans, but control theory and practice is in my wheelhouse.

It is indeed possible to build a vehicle that can’t be controlled by a human. And there were lots of aviation smoking ruins scattered about during the run-up and early stages of WWII because human reaction times and control wasn’t thoroughly considered in the design and build of airplanes. Which is a large reason the field was invented out of an alliance of medical doctors, mechanical engineers, and psychiatrists.

Processing time in the brain may be variable, given training and neural nets of the flesh ware kind. But propagation time up and down nerves doesn’t really vary. There’s reasons why there’s grey matter in one’s spinal cord; without it, one would be hard-pressed to keep one’s balance or ride a bicycle, because the additional delay to and from the actual brain would have one tripping over cracks, a lot. Kinesthesia, anyone?

So, say one is following the next car up at a one-car distance at 65, as a couple of people have proposed? One can keep the distance steady since one is in active control with one’s foot on the accelerator, there’s only a 0.1 second delay from brain to leg muscles, and with some fairly advanced time delay processing (which we kind of have, built in), one can keep the distance steady and interlopers out.

But! Five cars up, where one can’t see them, somebody moderately brakes. Next person back has a .1 second delay to see and recognize the problem and, let’s be generous, .5 seconds before they touch their brakes; during that .6 second interval, the back-one car isn’t slowing down, so they have to brake harder. The next schmuck has the same delay and has to brake even harder than that, and so on. If our hero at the end of this chain is lucky, the high-mounted brake lights will give him some early warning. But if he’s following with a .6 second delay from the car in front-well, lotsa luck, sucker. I don’t care how superhuman your grey matter is: it takes time to get signals down the nerves to the muscles in the body so one gets off the gas and on the brakes and, at that following distance, You Lose.

Race car drivers push these limits. They crash a lot, too. And, funnily enough, they organizers of the fastest races out there limit the maximum speeds: too many dead bodies because the required reaction times are faster than human physiology allows.

Aanndd: now the 3-second rule. Notice: that’s a time, not a distance. It gives a slightly inattentive human time to react, smoothly, without having to resort to panic stops. Moreover, it means that that a chain of cars following that rule doesn’t go unstable, period, even if #1 in the chain panic stops for an object falling off a truck.

This isn’t a hypothetical. Personally, I was in fast, heavy traffic on I-95 in CT one day, on TACC, when the car went into just-short-of-screeching-the-wheels braking as the cars in front stopped short, coming to a halt with a few feet of clearance. So did the two cars behind me. But three cars back there was this pick-up truck with a somewhat aggressive driver who had been swapping lanes and such, trying to get ahead. Him.. he couldn’t stop in time, went left to avoid a direct collision, and sideswiped both the center Jersey barrier and the car in front of him.

In my experience, I tend to use the three second rule. On long trips, people who dive into the gap clearly feel uncomfortable and dive back out again. In heavy traffic, there’s always a maniac. Just let them be them. As my Grandmother used to say, ‘Just because someone jumps off the Brooklyn Bridge, it doesn't mean that you have to."
 
Must be a mass rollout as I just got my download and install notice - no YB’er or Charlatan @$$ Kiss here, either. 😳

View attachment 923493

Okay… I was going to talk about my update, or lack thereof, but we first have to discuss this screen shot. Is this an iPhone 4 or is your text set to 300% zoom? 😅

For other OGs who didn’t get it, don’t feel too bad, I also didn’t get it. BUTTTTT as of March 26th the car was up to date. Still says that now, even after a restart 😅

IMG_0239.jpeg