FSD v12.x (end to end AI)

[aronth5]

Just another interesting low visibility intersection. 3 roads but only 1 Stop sign where the white line is.
If you don't know the intersection wouldn't you assume each road has a Stop sign? Nope and difficult to see from the other two roads.
FSD must know from the map data to proceed very cautiously because it doesn't stop for the other 2 roads.
Elon should try driving in New England.

 

[Daniel in SD]

Just another interesting low visibility intersection. 3 roads but only 1 Stop sign where the white line is.
If you don't know the intersection wouldn't you assume each road has a Stop sign? Nope and difficult to see from the other two roads.
FSD must know from the map data to proceed very cautiously because it doesn't stop for the other 2 roads.
Elon should try driving in New England.

View attachment 1053894

Intersections like that exist in San Diego too. It doesn’t look difficult to see that the other directions do not have a stop sign.

This one is actually a little trickier because there is a sign to the right but it’s a yield sign. No stop sign for cars turning left.

 

[zoomer0056]

First off: No offense, but you've got to change your mindset about Who's Driving. Without diving into the terminology weeds, you are.

You're definitely reading some alternative meaning into my post. An overly lengthy response saying things about FSD that one already knows if one is following FSD and it's development.

Ha, I hope you're not trying to resurrect a discussion on the definition of driving.

 

[LSDTester#1]

Speaking of cool kids... where is @PACEMD???
The King of off color humor and analogies.

@PACEMDis planting his fall harvest of dental floss.

 

[aronth5]

Intersections like that exist in San Diego too. It doesn’t look difficult to see that the other directions do not have a stop sign.
View attachment 1053903
This one is actually a little trickier because there is a sign to the right but it’s a yield sign. No stop sign for cars turning left.

At least you can tell there if there are Stop signs. The intersection I posted has too much foliage to determine if there are Stop signs or not. Complicated by lack of forward cameras.

 

[LSDTester#1]

As always "by this weekend" AGAIN. Starting to wonder if this is an Elon persona bot.

Told you!

 

[Daniel in SD]

At least you can tell there if there are Stop signs. The intersection I posted has too much foliage to determine if there are Stop signs or not. Complicated by lack of forward cameras.

I guess I would need to a similar intersection with stop signs that can’t be seen to understand what you’re talking about. There are also no stop lines in the other directions.
It also doesn’t look like any of the stop signs would be in the camera blind spot.

 

[gottagofast]

Anyone else more interested in ass than FSD 12.4

 

[uscbucsfan]

Anyone else more interested in ass than FSD 12.4

I am, but we have no timeline for that at this point.

 

[swedge]

I have no clue when I got 3.25 (first time I have actually checked my software version) but I did get the one month free trial of V12.3 right away, before anyone else I knew. Not sure if that means anything.

2024.3.25 has both old time "testers" as well as some others who opted in to FSD more recently. It was an FSD for anyone version, not a tester only version. Many of the non-testers on that version have already been updated to 2024.14.x, but some have not yet.

Tesla does not tell us who is or is not in the tester group. Looking at the entire history of updates, it appears that they stopped adding testers back when FSD was offered to averyone, in May of 2023 and with version 2023.12.10 (FSDβ 11.3.6). Prior to that, the FSD code was included only in updates of the tester cars, but after that point, all the main updates included the FSD code. However, releases which went only to testers continued. If you have been subscribed to TeslaFi all along, you can look up your own update history - if your first update which included FSD was after 5/22/2023, you are probably not in the tester group.

This matters because testers in this thread have a very different perspective on FSD than new FSD users. Testers have spent years noticing and reporting every glitch and foible in FSD, some of which persist in the current FSD 12.3.6 version. They tend to consider flaws to include any behavior which they could do better, or which they choose to intervene or disengage. Many reports relate to the driver or passenger comfort level, for example spending too long at stop signs.

It also matters because the testers have had updates to FSD 12.x since February, and 12.3.6 since April. They are ready to start reviewing something new. But it is likely that main branch folks will have to wait a while. So the tester discussions here differ vastly from the concerns of newer FSD and V12.3.6 users.

 

[LSDTester#1]

Anyone else more interested in ass than FSD 12.4

As long as I’m the DIC

 

[sleepydoc]

Anyone else more interested in ass than FSD 12.4

After doing @FSDtester#1 ’s colonoscopy I’m definitely more interested in 12.4

 

[OxBrew]

What does Elon's Tweet say? We both must be high because this can't be what he said and I MUST be reading incorrectly.

".....Two other versions are in earlier stages of testing: 12.5 and 12.6, which could be called v14 and v15. We are starting to get to the point where, once known bugs are fixed, it will take over a year of driving to get even one intervention."

Say WHAT???????

That is bound to disappoint Alan unless he drives ½ mile a year.

View attachment 1053815

Only 179 years to go. Needs to be able to drive about 3 lifetimes without an intervention to be better than 75th percentile drivers.

50th percentile is not good enough. You'd be asking half the drivers to get into a more dangerous car.

Actually might need to be better than 90% of drivers to be really safe and have a positive safety impact on aggregate accident rates.

IDK about exact numbers, but it's several lifetimes worth of driving to make any difference at scale.

Newsflash: it's not going to go a year without an intervention this year. Maybe not this decade.

lol

 

[FSDtester#1]

2024.3.25 has both old time "testers" as well as some others who opted in to FSD more recently. It was an FSD for anyone version, not a tester only version. Many of the non-testers on that version have already been updated to 2024.14.x, but some have not yet.

Tesla does not tell us who is or is not in the tester group. Looking at the entire history of updates, it appears that they stopped adding testers back when FSD was offered to averyone, in May of 2023 and with version 2023.12.10 (FSDβ 11.3.6). Prior to that, the FSD code was included only in updates of the tester cars, but after that point, all the main updates included the FSD code. However, releases which went only to testers continued. If you have been subscribed to TeslaFi all along, you can look up your own update history - if your first update which included FSD was after 5/22/2023, you are probably not in the tester group.

This matters because testers in this thread have a very different perspective on FSD than new FSD users. Testers have spent years noticing and reporting every glitch and foible in FSD, some of which persist in the current FSD 12.3.6 version. They tend to consider flaws to include any behavior which they could do better, or which they choose to intervene or disengage. Many reports relate to the driver or passenger comfort level, for example spending too long at stop signs.

It also matters because the testers have had updates to FSD 12.x since February, and 12.3.6 since April. They are ready to start reviewing something new. But it is likely that main branch folks will have to wait a while. So the tester discussions here differ vastly from the concerns of newer FSD and V12.3.6 users.

Well said, and you would know. You have been around since the beginning!

 

[FSDtester#1]

Anyone else more interested in ass than FSD 12.4

After doing @FSDtester#1 ’s colonoscopy I’m definitely more interested in 12.4

By the way Doc, I forgot to mention. There is this new medical ointment most use today for my procedure, it's called lube. Maybe get some for your office.
I haven't sat down since last week

 

[Electroman]

I don't have FSD, only tried free FSDS for a month. But I am curious here. When FSD messes up like this, is there risk of an accident? Does FSD act impatient, get increasingly aggressive and try to make the exit? Or remain safe and patient, miss the exit and reroute?

Nope. I have not seen where it does dangerous stupid things to fix an earlier mistake. It feels no shame in missing exit after exit.

 

[FSDtester#1]

Only 179 years to go. Needs to be able to drive about 3 lifetimes without an intervention to be better than 75th percentile drivers.

50th percentile is not good enough. You'd be asking half the drivers to get into a more dangerous car.

Actually might need to be better than 90% of drivers to be really safe and have a positive safety impact on aggregate accident rates.

IDK about exact numbers, but it's several lifetimes worth of driving to make any difference at scale.

Newsflash: it's not going to go a year without an intervention this year. Maybe not this decade.

lol

You and Alan should become roommates. Get a TV station to pick up the reality show. It would be absolutely riveting television . "The House Husband's of San Diego "....

 

[JulienW]

By the way Doc, I forgot to mention. There is this new medical ointment most use today for my procedure, it's called lube. Maybe get some for your office.
I haven't sat down since last week

Damn't Doc I can't get FSD to engage standing up.

 

[Dewg]

Only 179 years to go. Needs to be able to drive about 3 lifetimes without an intervention to be better than 75th percentile drivers.

50th percentile is not good enough. You'd be asking half the drivers to get into a more dangerous car.

Actually might need to be better than 90% of drivers to be really safe and have a positive safety impact on aggregate accident rates.

IDK about exact numbers, but it's several lifetimes worth of driving to make any difference at scale.

Newsflash: it's not going to go a year without an intervention this year. Maybe not this decade.

lol

You're going to need to back up those numbers.

 

[Daniel in SD]

Only 179 years to go. Needs to be able to drive about 3 lifetimes without an intervention to be better than 75th percentile drivers.

50th percentile is not good enough. You'd be asking half the drivers to get into a more dangerous car.

Actually might need to be better than 90% of drivers to be really safe and have a positive safety impact on aggregate accident rates.

IDK about exact numbers, but it's several lifetimes worth of driving to make any difference at scale.

Newsflash: it's not going to go a year without an intervention this year. Maybe not this decade.

lol

There’s a difference between necessary interventions and interventions though. Cruise and Waymo both went driverless after they reached about 5-10k miles between disengagements because the vast majority of disengagements did not prevent a collision. I agree that FSD won’t get there on the current hardware.