Software Update 2018.21.9 75bdbc11

[smk23us]

Last 2 days i drove nearly 2000 km on new firmware, on my AP1 nov. 2014 S85.

In my case, stering wheel torque sensor seams to be more sensitive to touch than before, so basicly i get less false nags than before and overal the same amount or less then before.

Unfortunatly now AP1 has this weird bug, that 3 times tried to stick (or even cross) straight line on highway. Twice it happened when i was going from gas station exit to highway and put AP on as soon as i entered highway, and car pulled imidetly to the right line driving on the line! It was something new and a bit scarry.
Once it did simillar thing on left lane.
I tried to put on and off AP twice and it did the same, as soon as i set it on. Had to drive a while and than try again to center in the lane correctly.

I dont know what they did, but AP1 was quite reliable before.

 

[Joe F]

I messed around with the nags today just to see what the update had done. I hovered my hands over the wheel to induce the nag several times. Then I experimented with how much force was required to dismiss it. What I found was that in my car it was variable.

I recall at some point you were going to try comfort steering, but don't remember reading if you had.

If you haven't already, I'd toggle AutoSteer off/on. I've seen things in the past which were broken after an update (summon recently) where the option was on, but just wasn't working properly. Toggling brought it back to life...

I still think something's broken in your M3, despite what the service center told you. Especially since steering resistance is variable (provided at the same speed.) Just doesn't feel right to me.

And their telling you to use the stalk or brake to disengage AS is BS given the manual states, clearly, the steering wheel can be used to disengage AS. This from the Model S manual:

"When you leave the road, or disengage Autosteer by using the steering wheel, you can increase your set speed again, if desired."​

Although, to make things more confusing, the manual further down lists methods to cancel AS, and turning the wheel is not listed. Oversight? Don't know...

Also, death grips and constant downward pressure is not required to have "hands on wheel" recognized. Again from the manual, as well as my own (and others) experience:

Autosteer detects your hands by recognizing light resistance as the steering wheel turns, or from you manually turning the steering wheel very lightly (i.e., without enough force to retake control). When your hands are detected, the message disappears and Autosteer resumes normal operation.
​

Time to escalate the problem...

 

[boonedocks]

Before and after I reset mine:

From this:

To this:

 

[Joe F]

Before and after I reset mine:

Most experienced here reboot after a new install from what I've read over the years. I thought it was SOP, and was surprised you did not.

And I saw your violently wiggling the wheel the first time you posted it, and just SMH. I understand the frustration, but expecting that to register and cancel the nag was a tad overboard IMO.

Glad the reboot cleared whatever setting was hosed for you! It's a computer, and a reboot is prudent from time to time. I'm just surprised that this problem was solved by a reboot. Good find.

I rebooted when summon was failing and it did not fix it. Toggling the features state did.

 

[boonedocks]

Most experienced here reboot after a new install from what I've read over the years. I thought it was SOP, and was surprised you did not.

And I saw your violently wiggling the wheel the first time you posted it, and just SMH. I understand the frustration, but expecting that to register and cancel the nag was a tad overboard IMO.

Glad the reboot cleared whatever setting was hosed for you! It's a computer, and a reboot is prudent from time to time. I'm just surprised that this problem was solved by a reboot. Good find.

I rebooted when summon was failing and it did not fix it. Toggling the features state did.

The first video was to show how far I could turn the wheel before cancelling the nag. I had to go almost to the point of disengaging AP to get it to cancel. I could have turned it once and it would not really have shown the extremeness. I always reboot but not necessarily the ICU or powered off the car for a full 5 minutes.

 

[Joe F]

The first video was to show how far I could turn the wheel before cancelling the nag.

Yeah, but, that's not turning the wheel. That's sending mixed (and confusing) signals to the system.

If I instructed you while you were running, "turn left, no, turn right, no, turn left" repeatedly many times a second, what would you do?

 

[boonedocks]

Yeah, but, that's not turning the wheel. That's sending mixed (and confusing) signals to the system.

If I instructed you while you were running, "turn left, no, turn right, no, turn left" repeatedly many times a second, what would you do?

Well thank goodness my steering nag is fixed. I can't go back and do a single left or right now because I no longer have that problem. However if you see the 2nd video you will see I can't even nudge it close to the extreme of the first turn in the first video....My point still stands...The first of the left / right turns should have disengaged it. I won't argue with you anymore since obviously that's what you want.

 

[MorrisonHiker]

I didn't have any problems with nags this morning...probably because I didn't even bother trying Autopilot on the drive to work.

Honestly though, I rebooted after the poor performance on the drive home Wednesday. Thursday's drive home was much better and I don't remember any nags. The spot were it automatically dropped the set TACC speed on Wednesday didn't cause any issues on Thursday. I did have it do the TAKE OVER IMMEDIATELY warnings twice in a row for no reason that I could tell. Perhaps it was confused by a sign located in the middle of a crosswalk but it had no problem with that crosswalk the previous day.

 

[GlmnAlyAirCar]

Most experienced here reboot after a new install from what I've read over the years. I thought it was SOP, and was surprised you did not.

Not to be contrarian, but the "reboot after update" is just voodoo. I have a June 2015 S70D with 110,000 miles. I have accepted every update. Never rebooted after an update. And I have never experienced any problems after an update.

Plus, I can't imagine the MCU doesn't reboot on its own after an update, especially for an update that modifies the OS.

 

[Joe F]

I won't argue with you anymore since obviously that's what you want.

Sorry, I wasn't arguing at all. I thought it was a discussion. Didn't mean for it to seem harsh.

Just trying to point out, hopefully for others sake who saw and reacted to the first time you posted the video, that, that is not how things work when the system is working properly. Like crying in baseball, there is no shaking in AP.

 

[Joe F]

Not to be contrarian, but the "reboot after update" is just voodoo. I have a June 2015 S70D with 110,000 miles. I have accepted every update. Never rebooted after an update. And I have never experienced any problems after an update.

Plus, I can't imagine the MCU doesn't reboot on its own after an update, especially for an update that modifies the OS.

How then does that explain the change @boonedocks saw after rebooting?

Or, why is a reboot required when the rear camera goes brain dead and either freezes the screen incorrectly on either the last frame displayed or totally blank? Or a reboot clearing a GPS problem?

I tend to follow the wisdom of those who came before me. Perhaps there were problems in earlier releases that caused issues the first time the kernel booted, like timing problems where the H/W was probed but was in a not ready state for instance.

I agree with you that, ideally, it should not be required, but there's no such thing as a bug free program, except for maybe that one last bug.

 

[MarcG]

Hey, I uncovered that the Nag Warning time is directly related to your speed. It appears to be: 40-45 sec at 40-50 mph, 30-35 sec at 65 mph; 25-27 sec at 70 mph; 23-25 sec at 75 mph; 18-20 sec at 80 mph. I'm pretty sure my Model 3 was 60 sec for all speeds before. My Model S HW1 was 3 mins for all speeds.

Yep, it’s called distance

Found that out a couple of days ago per the thread I created here:
New logic for “Hold Steering Wheel” nags in 2018.21.9

 

[clostridium]

I've not been happy with the auto wipers at any stage since they never seem to wipe frequently enough nor do they like going to a continuous wipe mode except in a torrential downpour.

It seems like the wipers were pretty decent and rarely needed overriding a version or two after they were introduced except they tended to run a bit fast for the conditions. Now they seem to have regressed a bit and I’ve noticed that they won’t ramp up speed sufficiently in heavier rain as you have also noticed. I’m wondering if people complained that they were running too fast in lighter rain and Tesla overshot in their “fix.”

 

[chillaban]

Wow, on highway driving, 2018.21.9 is really shining, even compared to 2018.18! I was just able to do the whole Tehachapi mountain pass with zero interventions, and it never hit the pavement grooves (which are about 4 inches inside the borders of the lanes)

This is unprecedented for me. It felt like the car is on rails. Seems much better at understanding how to approach curves on a hill, and much more likely to preemptively start turning rather than playing catch-up once it’s almost out of the lane.


EDIT: I should also mention my speed was set at 75 the whole time and traffic wasn’t too heavy so the lowest speed it reached was 65. It previously could perform acceptably when I was blocked by a truck but would struggle around sharper curves at speed.

 

[Nick's Life]

Yep, it’s called distance

Found that out a couple of days ago per the thread I created here:
New logic for “Hold Steering Wheel” nags in 2018.21.9

Are you sure its distance and not speed? I tested 40/45, 65, 70, 75, and 80 mph without recording if there were cars in front of me. Perhaps a combination of both.

 

[MarcG]

Are you sure its distance and not speed? I tested 40/45, 65, 70, 75, and 80 mph without recording if there were cars in front of me. Perhaps a combination of both.

Yep, here - let’s do some simple math on the average of your numbers:

40-45 sec @ 40-50 mph:
42.5 * 45 / 3600 = 0.53 miles

30-35 sec @ 65 mph:
32.5 * 65 / 3600 = 0.59 miles

25-27 sec @ 70 mph:
26 * 70 / 3600 = 0.51 miles

18-20 sec @ 80 mph:
19 * 80 / 3600 = 0.42 miles

So just around 0.5 miles on average given the ranges quoted, which is what I’ve been finding per my other thread.

 

[fasteddie7]

Far too many threads on this topic I guess.

Here's an example from a different thread that I thought was in this thread.

While I will agree the sensing method Tesla choose is not ideal, it is what we have today. Those who wish to just keep their hands anywhere but on the wheel must never have experienced shorts changing encounters on the highway at speed where milliseconds count in an avoidance maneuver.

There are credible reports though of people who have experience with other Tesla cars, who find their newer Tesla requiring more steering effort to either disengage AS, or to cancel the nag. @Az_Rael is one I'd believe in this post, given her history. It's quite possible Tesla has changed steering H/W at some point, or something inside the box has changed, such that it is not calibrated the same from those of us who do not have problems with sensing "hands on wheel." In that case, Tesla needs to check into the problem further and modify what triggers sensing.

If I get a chance, I'll shoot and post a video showing how little effort it takes to cancel the nag. If others can't duplicate this simple single finger nudge of the wheel, they need to open a service center call ASAP as something has changed.

So maybe I won't have to shoot a video. I didn't watch it earlier, but @fasteddie7 posted his findings in this thread, at least for his older AP1 Tesla. Hey Fast, have you compared it with the AP2 Tesla yet? My late 2016 Ap2 has no problem with canceling nags at all.

I have a similar experience with ap2. It does indeed look like autopilot changed in order to nudge us to using as intended, hands on the wheel, and not elsewhere but jiggle the wheel every 30-45 seconds. I do see others that don't have the same experience, might be slightly variance on manufacturing, I recall having a loaner that when I used the wheel to disengage autopilot the wheel would violently jerk, and I've never experienced it elsewhere.

 

[Twiglett]

true, and not widely-enough publicised, Daniellane

other than the manual of course

 

[ICUDoc]

True.
But not not widely-enough publicised.
Maybe if we all keep mentioning the limitations of Autopilot, and the necessary conditions to use it safely, people will stop abusing it and it will be less hobbled by the company. A few bad apples ruining it for the whole bunch.

other than the manual of course

 

[LS..]

Anyone loose their Internet connection on 21.9? Service thinks its due to the latest firmware.

I've been out for the last three days...2017 X100D AP2.5 MCU1
Thanks.