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

2019.40.1.1

This site may earn commission on affiliate links.
@Plastic6, you are wasting you time asking for people to stop spamming us with useless messages. I have sent many messages asking for the same thing.

The easiest way to stop seeing these useless posts is to click on their user name and then click on Ignore. I have clicked on Ignore on as many as 50 in a day in all of the threads that I subscribe to.

Ok, I'm gonna ignore that. Like this post, if you found that comment useful..........
 
Last edited:
  • Funny
Reactions: APotatoGod
I wonder if this means that "automatic city driving" is in shadow mode now.

Seriously, do you believe in Santa Claus? The way you think of shadow mode is how kids think of Santa. Verygreen (you know, the person you cite for the auto city driving code discovery) has repeatedly debunked this "shadow mode" existence. They are not operating side code and comparing anything (it would be visible to VG).
 
Seriously, do you believe in Santa Claus? The way you think of shadow mode is how kids think of Santa. Verygreen (you know, the person you cite for the auto city driving code discovery) has repeatedly debunked this "shadow mode" existence. They are not operating side code and comparing anything (it would be visible to VG).

Uh?! Tesla itself confirmed shadow mode exists during autonomy investor day. See screenshot below.

upload_2019-12-1_15-55-23.png

And here is how Tesla described Shadow Mode during Autonomy Investor Day:

"When we originally have algorithms that we want to try out, we can put them on the fleet and see what they would have done in a real world scenario."
 
Last edited:
Uh?! Tesla itself confirmed shadow mode exists during autonomy investor day. See screenshot below.

View attachment 483604

And here is how Tesla described Shadow Mode during Autonomy Investor Day:

"When we originally have algorithms that we want to try out, we can put them on the fleet and see what they would have done in a real world scenario."

Did you read my post? It's a lie no matter how much you want to believe, Mulder. Verygreen, your exact source, had debunked their lies repeatedly but you'll only parrot the Tesla lies as if they are more credible than the actual code itself.

And, for the record, I much more respect you responding to posts unlike the idiots who just click disagree but remain in ignorance.
 
Did you read my post? It's a lie no matter how much you want to believe, Mulder. Verygreen, your exact source, had debunked their lies repeatedly but you'll only parrot the Tesla lies as if they are more credible than the actual code itself.

If it is not shadow mode, then what is the new code "CityStreetsBehavior" doing?

And, for the record, I much more respect you responding to posts unlike the idiots who just click disagree but remain in ignorance.

Thanks.
 
If it is not shadow mode, then what is the new code "CityStreetsBehavior" doing?



Thanks.

VG will let us know after he digs in but it's likely the driving code for city FSD itself but it's not running and comparing what it would do vs the actual driving decisions. That's the fantasy part.

So be excited, I am, but don't think Tesla is somehow testing the code in any way because it's partially rolled out. This is how Tesla rolls out feature code. They did this with NoA and VG noticed it but initially it was called something else like Drive on Autopilot.
 
VG will let us know after he digs in but it's likely the driving code for city FSD itself but it's not running and comparing what it would do vs the actual driving decisions. That's the fantasy part.

So be excited, I am, but don't think Tesla is somehow testing the code in any way because it's partially rolled out. This is how Tesla rolls out feature code. They did this with NoA and VG noticed it but initially it was called something else like Drive on Autopilot.
I don't know man. I agree with diplomat33 personally. But maybe you're right and it's just what I want to believe.

I felt like when the ability to set AP to zero MPH showed up it was an early version of traffic light and stop sign abilities. I was under the impression (although I have no facts) that they've been testing this in shadow mode for some time now.

Either way, the fact that anything related to Automatic City Driving is showing up in the code is good news... Even if it's still a long way off.

Can you point me to where Verygreen has debunked Tesla's claims of running functionality in shadow mode? Thanks!
 
  • Informative
Reactions: APotatoGod
Been on 40.1.1 for a couple of days now and scheduled departure time is all messed up.

My car has been topping off charging all through the weekend and started charging after an errand as soon as it was plugged in on Saturday. I have scheduled departure set for 7:00am. Also I have my max charge set to 90% and it is now at 92% after it’s last “top-off”. Very odd “new” charging behavior.

2019 Model S Raven P + L+
 
Been on 40.1.1 for a couple of days now and scheduled departure time is all messed up.

My car has been topping off charging all through the weekend and started charging after an errand as soon as it was plugged in on Saturday. I have scheduled departure set for 7:00am. Also I have my max charge set to 90% and it is now at 92% after it’s last “top-off”. Very odd “new” charging behavior.

2019 Model S Raven P + L+
I'm on 36.2.2. and Scheduled Departure is behaving the same as you have described.
 
Did you read my post? It's a lie no matter how much you want to believe, Mulder. Verygreen, your exact source, had debunked their lies repeatedly but you'll only parrot the Tesla lies as if they are more credible than the actual code itself.

And, for the record, I much more respect you responding to posts unlike the idiots who just click disagree but remain in ignorance.

If I’m not mistaken very green has mentioned that he sees some but not all of the code a while back. If he’s reading this he can correct me.
 
Been testing 2019.40.1.1 for a few days and holy S*** lane changes are significantly better.
I use NoA with stalk confirmation and as soon as I hit that stalk, the lane change starts with the first blinker, and executes much faster with sharper wheel angle (only when the lane is open, of course). It will also execute this into tighter gaps. Zero zig-zag cancels thus far as well. Bravo Tesla AP team.

PS, to settle the unrelated arguments above, yes it appears Tesla lied about shadow mode and doesn't really "compare" much active driving. According to Green (and autonomy day without the embellishment) they put out campaigns on specific cars on specific days to test false positive and false negatives for a new behavior they are testing, but that's about it. Think 'cut-in-detector looking for cut-ins in the background and when it sees one that wasn't predicted, uploads that as false neg; or when it predicts one that doesn't actually happen, uploads that as false pos; but only on certain cars at certain times; prior to launching a feature.' It takes hours of nerdy searching through Greens twitter posts to see his interpretation of the shadow-mode code but, it's there.

Anyway, 2019.40 is fantastic and very exciting. Dare I say lane changes went from confused grandpa to super-human?
 
@verygreen is the autopilot trip log format different with 2019.40.1.1? From my own usage of Autopilot on local streets, the ability to drop the set speed to 0mph should result in much lower Braking Disengagement and potentially lower Steering Disengagement numbers for sharp turns that can now be taken at say 10mph but not 18mph. This could help Tesla better identify problematic intersections or curves that can't be as "simply" fixed with a slower approach speed.

Would be interesting to see if Tesla uses "0mph set speed" as triggering condition to send additional data for traffic light or stop light training. But then again, the percentage of Tesla drivers who push the limits of Autopilot on local streets is probably quite small, but perhaps the data quality could be worthwhile anyway

Or maybe the "lower than 18mph" set speed is a precursor to speed adjustments similar to how Autopilot drops the set speed when going through interchanges as well as Autopilot automatically slowing down for curves on highways ignoring the set speed.
 
So besides the "scheduled departure time" being fubarred in this update, NOA is also a complete mess. First of all auto lane change is as others have said. Super quick and as most people (human) drivers would do, so there's that.

For NOA on 40.1.1 My first commute today to try it out and on my first 17 miles of expressway on I985 Southbound in NE Atlanta at EVERY single exit ramp, just as I pass them, my car would immediately decrease the speed from 75mph to anywhere for 60 down two 45. - This is NOT phantom braking but actually resetting the speed and displaying that new set speed on the binnacle. - NO ONE IN FRONT OR BESIDE ME and well over 1/2 mile before the nest entrance ramp. As you can imagine especially with a Raven model S with extra regen, the cars behind me did not take kindly to this. After the third occurrence and right cue when I would guess it was going to happen again I turned OFF NOA, just as it has been since the 2ned week if was introduced. I am in the process of writing up an email to Tesla but after several other emails to them including video links and Tesla CSR just emailing back saying "reboot needed" / "NOA not intended for HOV lane usage" "schedule an appointment your car needs service" etc etc etc...I already know it will fall on deaf ears.
 
Last edited:
  • Informative
Reactions: DrDabbles
For NOA on 40.1.1 My first commute today to try it out and on my first 17 miles of expressway on I985 Southbound in NE Atlanta at EVERY single exit ramp, just as I pass them, my car would immediately decrease the speed from 75mph to anywhere for 60 down two 45. NO ONE IN FRONT OR BESIDE ME and well over 1/2 mile before the nest entrance ramp. As you can imagine especially with a Raven model S with extra regen, the cars behind me did not take kindly to this. After the third occurrence and right cue when I would guess it was going to happen again I turned OFF NOA, just as it has been since the 2ned week if was introduced.

I don't think the phantom braking you're describing is related to NoA - did it actually improve after you turned it off? The issue seems to be a combination of:
- "Fleet Speed" - a sounds-good-in-theory-until-you-try it feature where Tesla seems to collect data from other cars and their average speed, theoretically to detect where people naturally slow down before sharp curves, etc. In practice it seems to react to "There's always traffic here so people slow down, so I'm going to slam on the brakes even though the road is clear now"
- Bad speed limit info, with disagreement over whether it's from OpenStreetMap or Google. You'd see this by TACC changing the set speed.
- Imprecise lane detection, with the car believing you're actually in the exit ramp lane. May or may not see this in the TACC display, but I suspect it triggers one of the first two items - either it thinks you're in an off ramp with a lower speed limit, or the Fleet Speed data is imprecise so it's trying to slow down as others who took the ramp did.

I have noticed that AutoStreet does a slightly better job in 40.1.1 at not swerving wildly into and out of merge lanes when the lane lines disappear. It was actually reasonable enough that I left autosteer on when I was in the right lane and dealt with the occasional slight drifting rather than the old behavior which was to immediately dart to the right as though it was cutting off whoever might be trying to merge in.
 
I don't think the phantom braking you're describing is related to NoA - did it actually improve after you turned it off? The issue seems to be a combination of:
- "Fleet Speed" - a sounds-good-in-theory-until-you-try it feature where Tesla seems to collect data from other cars and their average speed, theoretically to detect where people naturally slow down before sharp curves, etc. In practice it seems to react to "There's always traffic here so people slow down, so I'm going to slam on the brakes even though the road is clear now"
- Bad speed limit info, with disagreement over whether it's from OpenStreetMap or Google. You'd see this by TACC changing the set speed.
- Imprecise lane detection, with the car believing you're actually in the exit ramp lane. May or may not see this in the TACC display, but I suspect it triggers one of the first two items - either it thinks you're in an off ramp with a lower speed limit, or the Fleet Speed data is imprecise so it's trying to slow down as others who took the ramp did.

I have noticed that AutoStreet does a slightly better job in 40.1.1 at not swerving wildly into and out of merge lanes when the lane lines disappear. It was actually reasonable enough that I left autosteer on when I was in the right lane and dealt with the occasional slight drifting rather than the old behavior which was to immediately dart to the right as though it was cutting off whoever might be trying to merge in.

I didn't write that clear enough. The car actually RESET the speed to those speeds and showed that NEW speed set on the binnacle. ie it showed 45mph as the set speed from 75mph and / or showed 65mph at three other instances. I had to bump the stalk up until I reset it to 75 mph again. It was NOT phantom braking at all.

Also that Interstate does not change speeds and has literally been there forever and non of our AP1 / AP2 / AP 2.5 or HW3 cars has ever changed speed from the set TACC or AP speed including the limited time I used NOA on my HW2.0 S and when I first picked up my HW3 car this year..
 
Last edited: