Has anyone noticed this with 10.9? It doesn’t show me the 10.9 release notes. These are the release notes for 10.8. I’m wondering if I didn’t really get 10.9…even though I have the 30.10 version.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
10.9 FSD
- Thread starter Buckminster
- Start date
Has anyone noticed this with 10.9? It doesn’t show me the 10.9 release notes. These are the release notes for 10.8. I’m wondering if I didn’t really get 10.9…even though I have the 30.10 version.
CWhite
2020 Model S LR+
A full rewrite means doing the code again for things like ‘this is a yellow line’ , can’t believe they are doing that. A lot is changing, to be sure, but code branches that are working fine should be ported over.
EVJAutoJunkie
EV Junkie
There is no code for "this is a yellow line" it is done by the NNs and they are being totally rewritten and retrained for FSD Beta 11.
rpo
Member
My experience is that it is roughly 0.1 miles between interventions on city streets. Last night, it repeatedly was crossing the yellow line by a foot on corners. Left corners, right corners, did not matter. The animation even showed the car over the line too.
rpo
Member
I am fearful of getting pulled over every time I use it on city streets, so that seems like a good analogy. The constant sudden steering jerks would get most cop's attention along with driving in the gutter on one street near my home. It wants to put the right tires about a foot from the curb while leaving 12 feet between the left wheels and the yellow line.
They aren't .. the NNs will simply continue to evolve .. if/when they switch highway/NoA to the new stack it will basically mean changing the route planner logic to use the new vision stack (very gross simplification here, of course).
10.9 has been a pretty big regression for me. A lot of the same stuff others are noticing. We do have like 2-3 ft of snow on the ground here and roads are blanched with salt, some are still snow covered, so maybe that’s playing a part in poor performance. It’s seems super janky even on clear streets in the city. I’ve had the passenger side rear wheel roll over a curb while making a right turn I think twice now. That’s never happened before. Lots of phantom breaking where other roads “t” into the road I’m traveling on. Rural unmarked roads, it drive so close to the right side, I feel like it’s gonna put me in the ditch. I don’t even use it on those roads anymore. Still way to long to complete SOME turns. 10.8 was way better for me.
According to Elon “better than a Human Driver” is right around those curbed corners and wheels.
Ski
I have not seen anything public from Tesla. There is one post by frequent contributor elasalle of one driver’s detailed experience, of 6,000+ miles driven on versions from 10.3 up thru 10.9.
#321,857 in perpetual investor thread (sorry, can’t seem to figure out how get the link to the post to insert.)
TL;DR: disengagements per mile go from 0.22 to 0.04.
While it’s anyone’s guess how representative this is of Tesla’s aggregate data, I find it valuable because it captures relative changes from one version to the next for the same driver over (hopefully) the same routes.
Sure is interesting to see one driver's experience across versions, but 0.04 disengagements per mile is VERY different from my personal experience. I have FSD engaged for > 95% of my 30-minute all-non-highway commute every weekday, and I love using it, but I think my disengagement rate is roughly 1 disengagement/mile, and it's been pretty constant through the last several versions aside from a big increase on 10.8/10.8.1, which couldn't complete a right turn without crossing the yellow line and also wanted to cross solid lines to drive on the shoulder approaching a right turn (I haven't had the chance to drive 10.9 much yet to verify if these are fixed).
I think over 90% of my disengagements fall into one of these categories:
- Mis-classifies mailbox as a pedestrian and wanted to cross double yellow line to give the "pedestrian" appropriate space (usually it's those thick plastic mailboxes like this one)
- Tries to cross double yellow line to go around queue of cars in front waiting at a stop sign or traffic light that is out of sight (around a corner or over a hill)
- Poor lane selection (got into a turn lane when going straight, changed into a lane that ends in a few hundred feet, changed lanes away from an upcoming turn within 0.5 mile of the turn or sometimes within a few hundred feet of the turn)
- Doesn't stay right on roads with no yellow center line, especially at stop signs where it would prevent other cars from turning onto the road I'm leaving.
- Approaches sharp turns wayyy too fast (45 mph speed limit technically, but yellow warning signs indicate 15 mph is appropriate for this curve)
So I suppose if the roads in the area of that other driver don't have that big plastic mailbox type, or roads with stop signs / traffic lights can be out of sight when queueing, or unmarked roads, or medium speed roads with some extra sharp turns, I could believe 0.04 disengagements/mile, but it sure doesn't work that way where I drive.
Daniel in SD
(supervised)
Tesla, TSLA & the Investment World: the Perpetual Investors' Roundtable
They should've built GF4 in The Netherlands, that would've gone much smoother! oh, wait..
teslamotorsclub.com
tmoz
S85D, Y LR, ex Prius PiP
Thanks - that was one data point. As many have said, what type of road one uses to test FSD has a big impact on disengagements.
What is needed is a standard route to measure disengagements.
And of course, everyone's "typical" drive will include a different mixture of those difficult situations, so, as the saying goes, Your Mileage May Vary.
However, a "standard route" really is needed to quantify the various trouble situations.
israndy
Supercharger Hunter
What's the math to covert that into miles per disengagements? Seems like an easier number to understand.
If it's 0.25 that's 4 miles per, if it's 0.5 that's 2 miles per, so 0.05 would be 20 miles per
So went from more than 4 miles per to more than 20 miles per
Not my experience, but I do most of my city driving around my city (surprise!) and I hit the exact same issues each drive (map data not updated). I get the advantages of the beta like it doing the traffic lights and stop signs, and the removal of the +5 mph limit on certain highways, but I am pretty much unable to use it with my wife in the car or with anyone behind me going into a turn at an intersection or round about.
rpo
Member
If you mean hitting the button to save the report, then yes. Does Tesla actually look at those? I've done a few dozen of them.
EVNow
Well-Known Member
They have 10,000 labelers (not on contract, but apparently "internal"). They could potentially handle 300k clips a day - 30 per person. If each takes 5 minutes, still about 3 hours of the day. If labeling these are a large part of their job, they can do it ... but if its a small part (i.e. Tesla wants them to label mostly some other images/clips, it won't be possible).60,000 Beta testers submitting just 5 a day = 300,000 video clips a day or 12,500 per hour.They probably have a few interns that watch them a few 100 at time at 10x speed.
