How about Elon Musk dismissing it as minor? Would that count as an additional source?
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
FSD Beta 10.69
- Thread starter Buckminster
- Start date
sleepydoc
Well-Known Member
Yes, but how accurate are Elon’s tweets?How about Elon Musk dismissing it as minor? Would that count as an additional source?
View attachment 862297
Boston version could be
Last night FSD tried to slow down to turn onto my road but without warning slammed on the brakes so hard the gallon of milk in the passenger seat flew into the glove box then landed on the floor. Thank-fully the milk did not break. Definitely more phantom braking for me which has increased from .2 which had the most phantom braking events in several releases. Hurry up 69.3.
FSDtester#1
Large Member
FSD has been programmed not to care about milk.Boston version could be
Last night FSD tried to slow down to turn onto my road but without warning slammed on the brakes so hard the gallon of milk in the passenger seat flew into the glove box then landed on the floor. Thank-fully the milk did not break. Definitely more phantom braking for me which has increased from .2 which had the most phantom braking events in several releases. Hurry up 69.3.
If it was a 6 pack of Sam Adam's, it would of slowed down smooth as silk...
AlanSubie4Life
Efficiency Obsessed Member
Round trips, right? Not really possible to conclude, otherwise. You have an elevation range of about 400 feet worst case in Minneapolis area. Over 16 miles worst case that would alter results by about 40-50Wh/mi. (1.6kWh/1k ft *0.4kft / 16mi). (Reverse vs. forward would differ by up to 80-100Wh/mi.)
No idea the exact routes here so hard to say one way or the other.
And of course should compare to manual driving too.
PACEMD
Active Member
10.69.2.3 for my route has eliminated the phantom braking that had only shown up on the last 2 update versions. I had rarely in 54k miles had phantom braking until 10.69.2.2 et al.......now it's gone........cool.......
Your phantom braking is gone and mine is much worse such is the world of FSD testing!
SK360
Member
PACEMD
Active Member
That's fine, all that means is that you are completely wrong and I am the only one right, right?
PACEMD
Active Member
No seriously, that's why I expressed it as for my route. On those mentioned versions, there was consistent PB at certain regions of the interstate. Not now. It does not at all surprise me that other routs would produce different results. Your phantom braking my vary........
1. Elon's dismissed it as not major as posted above
2. No change to release notes
3. Anecdotally from most of the reviews here, 100,000 testers and Youtubers, no one has claimed any noticeable differences between 10.69.2.2 and 10.69.2.3. You'd think if there was something major, SOMEONE would've noticed
willow_hiller
Well-Known Member
How about Elon Musk dismissing it as minor? Would that count as an additional source?
View attachment 862297
There are three levels of versioning in software development. Major, minor, and patch. Usually this is denoted by the dots in the version number, but Tesla doesn't seem to be following that very closely for the 10.69 branch. Elon saying the next version will be major does not necessarily mean that the latest version was just a patch, it could be a minor version with some significant bug fixes or tweaks.
PACEMD
Active Member
Ok, I have a confession. I paid Tesla to move the phantom braking virus from my computer to yours. Not something I am especially proud of, but it had to be done........
sleepydoc
Well-Known Member
They were oneway trips. The trip that was 221 Wh/mi was a return trip. As I recall, it was 248 Wh/mi on the way there. More traffic and was using the air conditioning a bit. Minneapolis is pretty flat. The return trip was actually climbing slightly (86 feet over 16 miles if you can count that!) It’s hard to say for sure but I doubt I could do much better driving manually. Maybe a bit but 221 is about the lowest trip economy I’ve seen.
In watching the regeneration bar, it seems to be using the brakes less with 69.2.3. I might be imagining it, but I’m not seeing it as much. It also seems to be doing a bit better at braking earlier. Just a bit, though - it could still be a lot better.
The other thing I’ve thought of is how much energy the FSD computer uses vs manual driving. If it’s doing a lot of processing it could be using more energy but the difference may be negligible compared to all the other consumers. Either way FSD is at least capable of driving quite efficiently. Once it’s more developed it would be interesting to have a direct comparison by having 2 cars drive the identical routes.
EVNow
Well-Known Member
Here is the latest disengagement rate for me. As you can see 69.x is behaving much better than earlier releases.
One thing not captured above is the # of zero disengagement trips.
Before 69.x I had logged 6 zero engagement drives (3 had one intervention each). After getting 69.x I've logged 14 zero disengagement drives (2 had one intervention each). Infact yesterday I had a 20 mile zero disengagement drive (10 of those freeway, where I actually used NOA). The car took me all the way from the parking lot of a mall to my home with no interventions/disengagements - a fairly simple drive, but a first nonetheless.
The reasons I think are
- I've started trusting FSDb a little more and don't disengagement for jittery behavior. This is almost entirely a result of CULT videos.
- FSD is handling turns much better, and I have very clear cases where it handles turns now that it just couldn't handle earlier (like ULT into a small street with a vehicle waiting in the lane we are turning into)
- FSD's handling of roundabouts is somewhat random i.e. depends on whether there is a vehicle coming from the left and whether FSD stops for that vehicle. Since a lot of my disengagements are at roundabouts, if FSD handles the roundabout correctly, the chances of a zero disengagement drive are much higher.
About 60% of all disengagements I have had are because of roundabouts. But in 69.x roundabouts account for 76% of all disengagements !
One thing not captured above is the # of zero disengagement trips.
Before 69.x I had logged 6 zero engagement drives (3 had one intervention each). After getting 69.x I've logged 14 zero disengagement drives (2 had one intervention each). Infact yesterday I had a 20 mile zero disengagement drive (10 of those freeway, where I actually used NOA). The car took me all the way from the parking lot of a mall to my home with no interventions/disengagements - a fairly simple drive, but a first nonetheless.
The reasons I think are
- I've started trusting FSDb a little more and don't disengagement for jittery behavior. This is almost entirely a result of CULT videos.
- FSD is handling turns much better, and I have very clear cases where it handles turns now that it just couldn't handle earlier (like ULT into a small street with a vehicle waiting in the lane we are turning into)
- FSD's handling of roundabouts is somewhat random i.e. depends on whether there is a vehicle coming from the left and whether FSD stops for that vehicle. Since a lot of my disengagements are at roundabouts, if FSD handles the roundabout correctly, the chances of a zero disengagement drive are much higher.
About 60% of all disengagements I have had are because of roundabouts. But in 69.x roundabouts account for 76% of all disengagements !
Last edited:
I don't know if it is meaningful or not but I recorded multiple failures to stop for a particular stop sign for an unprotected left turn on 2.2 and today on 2.3 was the first time I can remember that it has ever stopped for this particular stop sign before making a left turn. For me this was a milestone.
Of course, this was followed by fantom braking for approaching cars so I waited until I got closer to town to reengage.
EVNow
Well-Known Member
sleepydoc
Well-Known Member
I’ve noticed that disengagement-free drives are definitely becoming more common, too.
The other thing I’ve noticed is that the lane selection is slowly improving. A large portion of my disengagements used to be because FSD couldn’t figure out how to stay out of a turn lane and go straight. It still has problems on a regular basis but they’ve improved. There are several places where I always had to take over in previous versions but now only have to take over half the time.
The two other scenarios that regularly cause issues are unprotected right turns/right turns with yield lanes and doglegs where you make a right, then need to quickly merge over a lane or two to make a left.
My observation is that 10.69.2.3 is a better release than the previous versions. I’m more comfortable with local driving.
Highway which was okay under 10.69.2.2 also appears to be better as well. I like the reaction when other drivers have abruptly moved into my lane. My car backed off smoothly like I would have.
Improvement.
Highway which was okay under 10.69.2.2 also appears to be better as well. I like the reaction when other drivers have abruptly moved into my lane. My car backed off smoothly like I would have.
Improvement.Wait, WHAT?!?! Your car is driving in the oncoming lane? And no one is talking about that?! That's huge, and should be an immediate recall.
Similar threads
- Article
