Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Almost ready with FSD Beta V9

This site may earn commission on affiliate links.
The fact remains that Mr Musk's NN should have deep-learned by now to keep his transient thoughts, however cool, in RAM.

Comforting to know that the CTO/CEO is amazed at the Beta street level NOA actually working,

... and is debating ripping out hardware and commenting out half the C++ code with the gardeners.

Got nobody to talk to? Elon, you're scaring the horses!
 
Last edited:
  • Like
  • Funny
Reactions: APotatoGod and goRt
Am starting to come around to vison-only for City NoA. Initially my reservations were the same as others - radar confirms distances, radar can spot & react to sudden braking ahead before the truck in front of you can react, it can tell the difference between a tunnel and a picture of a tunnel, etc...

But if radar continues to be used for safety features only, such as AEB, then vision + maps makes more sense.

I still have a concern about latency tho. The time-to-warn of the current system is too long. i.e. collision warnings and obstacle-aware accelleration always seem to kick in after the threat has passed.
 
  • Like
Reactions: APotatoGod
I still have a concern about latency tho. The time-to-warn of the current system is too long. i.e. collision warnings and obstacle-aware accelleration always seem to kick in after the threat has passed.
I haven’t noticed this. I have my collision warning set to “Early” and it’s always activated at what I feel is the appropriate time. Not too early to gove excessive false warnings, not too late to require extreme braking.
 
Seeing his reply about when the sensors disagree has always been my concern.

If you can get vision to lidar type measurements, and make it consistent, then you are close. (It appears they have) now if stiching the video goes well, I don't see how they can't get there fairly quickly. The only concern I still have is the camera placement and whether that's good enough for busy intersections that span multiple lanes like @Chazman92 Has tested.
 
Differences: only one of them has been taking money from consumers for almost 5 years for a product that can only be used on one device that may not even be functional anymore if the product is ever finished.
I didn’t buy autopilot when I first purchased my 2017 S new. It wasn’t until the 5k fire sale that I bought all of AP and FSD. For that price, I think it’s a good deal. But wouldn’t pay 10k for it today. A person can only blame themselves if they bought something on a future estimated plan that has yet to pay off. It’s a gamble. One I wasn’t willing to take.

Also the FSD purchase today does give a few more options than just what comes standard. It isn’t like the 10k is purely for future functionality.
 
I haven’t noticed this. I have my collision warning set to “Early” and it’s always activated at what I feel is the appropriate time. Not too early to gove excessive false warnings, not too late to require extreme braking.
Same here. Early seems fine to me. And that’s even with HW 2.0. Hopefully will be getting HW 3 in a couple weeks now that my local service center finally opened up instead of driving 5 hours. Woohoo!
 
Someone studying the Tesla firmware confirms the Neural Network (NN) is processing 15 frames at a time and is now outputting velocity and direction.
Tristan said:
... From the binaries we can see that they've added velocity and acceleration outputs. These predictions in addition to the existing xyz outputs give much of the same information that radar traditionally provides (distance + velocity + acceleration). ...
 
If all cars are autonomous in the future and are connected like a hive mind eg. the Borg, then radar is obsolete as the car will know what all the other cars around it are doing and seeing.
 
temporal persistence"...seeing the trajectory of hazards over time, predicting where things will go, etc. The code does account for that, although indeed they have improvements yet to make.

If all cars are autonomous in the future and are connected like a hive mind eg. the Borg, then radar is obsolete as the car will know what all the other cars around it are doing and seeing.
But not bikes, farm equipment and other non-connected vehicles.
 
It's a shame Elon didn't get into the space industry, could have thousands of Low Earth Orbit satellites giving live images from space, a Bird's eye view to assist FSD...
He could also use the low latency broadband Internet of Starlink to have thousands of far-flung unemployed foreigners remote driving our cars on subscription instead of FSD.

/s
 
Last edited:
I honestly bet that vision-only can't tell these aren't tunnels, or at best gets confused most of the time. Radar & Lidar would say "Wall" from any distance or approach angle.

StreetArt_Painted_Tunnel_02.jpg


StreetArt_Tunnel_To_Nowhere.jpg

No different than a sign post that is intended as a misleading art. Dangerous and should be banned.
 
  • Like
Reactions: qdeathstar
but better cars and manufacturing processes have been available in, for example, Japan for decades, Musky trying to reinvent the manufacturing process has been a disaster e.g. not having enough staff to move the cars when M3 went into production, not bothering to wait for paint to dry...
Unlike Toyota that says EVs can't be made and the future is H2. Got it.