That makes me think the reason none of us have gotten upgraded to 2024.2.x is that they are holding us back to deliver FSDb V12 to us soon. (Bringing back the split such that we will again start to be months behind the leading edge of general releases until Christmas.)
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
FSD v12.x (end to end AI)
- Thread starter Buckminster
- Start date
-
- Tags
- #fixphantombraking autopilot chill is not chill elon is an ass fix fsd extreme late deceleration fsd does not commit to go fsd does not maintain proper speed fsd hard and late braking fsd has decision wobble fsd jerky driving fsd late deceleration fsd sometimes may not detect train crossings fsd stops too early too quickly fsd stops too far behind line fsd tailgating in city streets
You have been around long enough to know all the steps involved that must happen first (and this is not even counting a whole new thing). We need to see it actually rolled out (internal testing and Whole Mars is NOT a rollout) to some of the influencers.
- Rollout to some influencers maybe 12.2.1
- Update and rollout/test again
- Rollout to 5% to 10% of the Early Access and test a week or 2
- Update and rollout/test again
- Rollout to a few more Early Access and test a week or 2
- Update and rollout/test again
- If all works out start a slow and methodical rollout to more (maybe 50%)
- Then if all goes well rollout to the rest of the Early Access
I'd rather have 2024.x.x/v11.4.9 than 2023.44.30.x/v12.x. I don't like old things.
I used to live there too but after Elon did this to me for the 24th time I finally learned or my
is ruined.
Knightshade
Well-Known Member
...What?
First release was FSDb v9... the interior camera was not mandatory until V10.2... story on that here:
Say goodbye to cabin camera covers - FSD Beta V10.2 requires camera to be uncovered
The cabin camera has been a controversial feature since the very first Model 3 rolled off the production line in 2017. As the name implies, the tiny camera mounted above the rear-view mirror was pointed […]
driveteslacanada.ca
And it's only quite recently they beefed up just how much attention it was paying to you paying attention.... and even THEN if you simply wear sunglasses you can look anywhere you like with your actual eyes- the car can't tell- as long as your sunglasses face forward.... Nor can it see your phone if you hold it low-ish because of how poorly positioned that interior camera actually is for driver monitoring-- a task it was never originally meant for.
I'm honestly kind of lost about what you're even trying to argue at this point.
Last edited:
I think it is likely to be sooner than that, but even that would qualify as soon.
According to TeslaScope only HW3 for initial 12.2 release. That means Omar initially won't get the new build since he has a HW4 car 
I hope the first V12.2 impression video posted tomorrow does not belong to Omar.
Don't believe so since he on HW3. See last post. I have to say though he has done better with V12 since a couple of his videos included other people which gave an additional perspective that was helpful. Of course that is assuming TeslaScope's feedback is accurate.
V9 had 2,000 testers....What?
First release was FSDb v9... the interior camera was not mandatory until V10.2... story on that here:
Say goodbye to cabin camera covers - FSD Beta V10.2 requires camera to be uncovered
The cabin camera has been a controversial feature since the very first Model 3 rolled off the production line in 2017. As the name implies, the tiny camera mounted above the rear-view mirror was pointed […]
And it's only quite recently they beefed up just how much attention it was paying to you paying attention.... and even THEN if you simply wear sunglasses you can look anywhere you like with your actual eyes- the car can't tell- as long as your sunglasses face forward.... Nor can it see your phone if you hold it low-ish because of how poorly positioned that interior camera actually is for driver monitoring-- a task it was never originally meant for.
I'm honestly kind of lost about what you're even trying to argue at this point.
You understand how little that would impact statistics with the amount of drivers added since V10?
The point is that there is visual monitoring on FSD. FSD is safe because it has a human forced to monitor it. You are trying to build strawmen about the ridiculously small amount of users or even mileage without a camera...because you're bored? Whatever.
OxBrew
Active Member
Watch that video? That's just the start. Imagine that 6 lanes wide, but with no lanes. And people drive the wrong way for a couple blocks against the flow, and everyone just adjusts like they're a rock in the stream.
AlanSubie4Life
Efficiency Obsessed Member
Yeah that is what he said. This is when Safety Scores were admitted:
Safety Score
Anyway back to that data, it is from 2022. So anyway there was some primitive camera monitoring for this data set (can’t remember the exact evolution but it used to be really lax). It is much much better now.
They finally released some new data recently, which is even less useful now that freeway and city streets miles are all smashed together with single stack.
We really cannot assess safety level relative to not using FSD Beta, due to selection biases. I think the updated excellent and rigorous attention monitoring may well push safety levels (as measured) a bit higher. But this is purely speculative. And then another question is - you are comparing FSD beta with attention monitoring to increasingly distracted drivers not using FSD. Is this what you want?
Thinking….Does having FSD increase the chances of a driver using a phone when not using FSD, because they are used to being able to do so when using FSD? (It is still quite easy to use a phone without nagging when using FSD.)
How would one capture that effect?
Seems to me it would be better to make sure accidents don’t happen when the driver is not paying attention. They seem more likely in that case. (Has been an argument of @verygreen for years as I recall.) Maybe some day!
I always hoped that buying FSD would at some point help prevent accidents when it was not engaged. My main reason for buying it, in fact (I had HW2.5 so the upgrade was critical)! Unfortunately seems not to be the strategy being followed so far.
It would be great to have an “Active Background Safety Fully Engaged” category in these safety metrics. It could (and should) include attention monitoring at all times. As a critical autonomous safety feature, of course this would be rolled to all customers, even if they don’t have FSD, as Elon has said.
I feel like if/when v12 is good enough they should just roll it out in the background and run it all the time for all customers even those who didn’t buy it. I’m probably ok with very sudden stops in that case.
Teslas would lap the field when it came to safety! Imagine expanding the critical reduction in accident rate to the entire HW3/4 fleet!
I guess that is when we will know it is safer?
Last edited:
The first versions of FSD Beta didn't have a number. It was updated to the influencers once a week. The influencers started giving FSD Beta a release number and there was some divergence so Elon step in and started numbering FSD Beta. Here is a DirtyTesla video of 20.44.10.2 that doesn't have a FSD Beta version number.
I choose to live in fantasy and hope that whenever I get it, it works even half as well as in Omar's vids.
I just want it to work at least as well as FSDb v11.4.9. It already handles 95%+ percent of my daily drives with no interventions/disengagements. (I don't count rolling the speed up, since even a 23% offset doesn't keep up with the local traffic. 23% is way too much for the highway, but I don't drive much highway so rolling down for that is more convenient than rolling up for everything else.) I guess the number one thing I want fixed is the lane changes to "follow the route" that are actually away from the route. But that is likely a map data related problem, so no FSDb update is likely to resolve that. The second thing to fix is to not be afraid of everything when it first sees/identifies it, which the default reaction to is slowing down. There are minor things that would be nice as well, like positioning in turns lanes, etc. but that is just a preference/comfort thing. My insurance app thinks excessive braking is the #1 problem, so I guess fixing that would be nice, but I think the app is way too sensitive.
Did they say that it was already two groups: employees and special groups?
