Autopilot buddy code is changed?

[jsmay311]

I did a follow-up test. Not only did adjusting the volume via the steering wheel scroll wheel after 9.5 minutes not prevent the nag, I actually got a forced disengagement got the first time at the 10 minute mark (or so).

Idk why I got a forced disengagement this time and not on my previous test drive, where I went the full 10 minutes without touching the steering wheel twice and only got a “nag” each time. Maybe the software gives you a pass on the first drive where it suspects a cheat device, or it only gives nags on the first two 10-minute intervals and then gives you a strike on the third.

And, just like someone else reported earlier, I got an “FSB Beta Warning” or strike recorded against me, even though I have FSD Beta disabled in the settings and was just using regular Autopilot. So now I’ve only got 4 FSD Beta strikes left.

Maybe tomorrow I’ll try repeating the test but after 9.5 minutes I’ll give the steering wheel a tug instead of just flicking a scroll wheel.

I did another round of Autopilot tests with a weight strapped to the steering wheel to simulate a cheat device:

First, I drove 9.5 minutes on Autopilot without touching the steering wheel. Then I grabbed the wheel and put some torque on it, and then I went another 9.5 minutes without touching it. I didn’t get any nags or forced disengagements.

In a separate test, I avoided applying any torque to the steering wheel but I used both scroll wheels a bunch and used the left stalk to initiate several lane changes. After~10 mines I got a forced disengagement.

Lastly, I drove ~30 minutes while only touching the steering wheel once every few minutes to apply some torque. I got zero nags or disengagements.

So it appears that the key is avoiding going a full 10 minutes at a time without manually applying torque to the steering wheel. Scroll wheel and stalk inputs don’t appear to have any effect, which makes sense since they don’t do anything to mask the presence of the weight / cheat device on the steering wheel

 

[AlanSubie4Life]

I did another round of Autopilot tests with a weight strapped to the steering wheel to simulate a cheat device:

First, I drove 9.5 minutes on Autopilot without touching the steering wheel. Then I grabbed the wheel and put some torque on it, and then I went another 9.5 minutes without touching it. I didn’t get any nags or forced disengagements.

In a separate test, I avoided applying any torque to the steering wheel but I used both scroll wheels a bunch and used the left stalk to initiate several lane changes. After~10 mines I got a forced disengagement.

Lastly, I drove ~30 minutes while only touching the steering wheel once every few minutes to apply some torque. I got zero nags or disengagements.

So it appears all that matters is avoiding going a full 10 minutes at a time without manually applying torque to the steering wheel.

Thanks for taking these (two?) strikes for the team.

So with a wheel weight, you'll get a strike after 10 minutes if you don't apply torque manually, no matter what you do with the buttons/scroll wheels, and stalks.

Seems pretty awesome, though I'd prefer a much shorter interval. I continue to be amazed that anyone can get a strike (without really trying, that is). It seems basically impossible if you're holding the wheel with two hands (appropriately imbalanced of course!) and looking at the road ahead.

 

[2101Guy]

Thanks for taking these (two?) strikes for the team.

So with a wheel weight, you'll get a strike after 10 minutes if you don't apply torque manually, no matter what you do with the buttons/scroll wheels, and stalks.

Seems pretty awesome, though I'd prefer a much shorter interval. I continue to be amazed that anyone can get a strike (without really trying, that is). It seems basically impossible if you're holding the wheel with two hands (appropriately imbalanced of course!) and looking at the road ahead.

I can’t tell if you are for or against wheel weights. Lol

 

[hybridbear]

I did another round of Autopilot tests with a weight strapped to the steering wheel to simulate a cheat device:

First, I drove 9.5 minutes on Autopilot without touching the steering wheel. Then I grabbed the wheel and put some torque on it, and then I went another 9.5 minutes without touching it. I didn’t get any nags or forced disengagements.

In a separate test, I avoided applying any torque to the steering wheel but I used both scroll wheels a bunch and used the left stalk to initiate several lane changes. After~10 mines I got a forced disengagement.

Lastly, I drove ~30 minutes while only touching the steering wheel once every few minutes to apply some torque. I got zero nags or disengagements.

So it appears that the key is avoiding going a full 10 minutes at a time without manually applying torque to the steering wheel. Scroll wheel and stalk inputs don’t appear to have any effect, which makes sense since they don’t do anything to mask the presence of the weight / cheat device on the steering wheel

Very interesting. What kind of a weight are you using for your testing? Something that is completely fixed to the steering wheel or a weight that dangles & swings?
And, for the forced disengagement after 10 mins there's no warning at all?
And, what software version are you running? There have been multiple software versions reported for when this behavior was added to AP.

 

[jsmay311]

What kind of a weight are you using for your testing? Something that is completely fixed to the steering wheel or a weight that dangles & swings?

I read somewhere that the autopilot weights were ~1 lb, so I just used a couple 0.5 lb barbell collars/nuts. And I strapped them nice and tight so they wouldn’t dangle/swing/wiggle/jiggle/etc.

And, for the forced disengagement after 10 mins there's no warning at all?

Basically, yes.

As I mentioned in my first post in this thread, during my first test, I did two 10-minute stints and only got harmless nags at the end of each. It wasn’t until my second test drive and my third overall 10-minute interval that I got a forced disengagement and accompanying FSD Beta strike. There was no blue flashing or any other indication that the disengagement was imminent. (The “nag” message actually does pop up about one second before the big red steering wheel image pops up, but I think it’s already too late at that point.)

Here’s a crappy video I took of the moment it disengaged with my phone looking up from the cup holder. It’s blurry but the nag message reads “Apply slight turning force to steering wheel” and the disengagement message reads “Autosteer unavailable for the rest of this drive. Hold steering wheel to drive manually.”

So my assumption is that you get nagged the first two times as warnings of sorts, and then kicked out every time after that.

And, what software version are you running? There have been multiple software versions reported for when this behavior was added to AP.

2022.36.20 (FSD 10.69.3.1)

You make a good point tho. Tesla can (and probably will) tweak this behavior at any point in a future software update. So one shouldn’t expect these test results to hold true forever.

 

[Benjimatt]

Just throwing this out there. (Im on standard autopilot, non-FSD)

I first noticed this

 

[Benjimatt]

Just throwing this out there. (Im on standard autopilot, non-FSD)

I first noticed this when my vehicle updated to 2022.40.4 and I started getting kicked out of Autopilot without warning
With wheel weight: After 10 min kicked out automatically
Using scroll buttons only instead of wheel weight: After 10 min kicked out automatically

Once I updated to 2022.40.4.1 everything seems to be back to normal
With wheel weight :After 10 min alert to puts hands on the wheel (blue warning)
Using scroll buttons only instead of wheel weight: After 10 min alert to puts hands on the wheel (blue warning)

Currently I have decided to stay on 2022.40.4.1 so that I can drive on the highway peacefully.

 

[hybridbear]

There had been uncertainly before about what version includes this code change, it has now become clear that it is in 2022.36.20.

 

[hybridbear]

Is there any change to this 10 minute cutoff in 2022.40.4.10 which is FSD Beta 10.69.3.3?

 

[nori]

does anybody pass the mission 2022.40.4.2? I almost did. could quit waring 15min, use weight and spring.

 

[Blacktes24]

Exactly. Why would a weight be more dangerous if the driver is still paying attention?

Genuinely curious what thought processes people have

bingo. AP kind of sucks when u have to mess with the wheel non stop. I can watch the road perfect and have my hand lightly on the wheel but it's not enough for the tesla as it wants pressure or movement. And I can tell something changed after the last update(no I don't have a wheel weight) but i keep my hand in the same spot/pressure a lot of times and I had it kick out 2x on me. not with a warning but from using too much resistance by using my hand.

but in the end the ap kind of sucks. i'm used to autopilot in tractors in fields where they drive with sub inch accuracy and you don't touch the wheel ever other than to turn at the end of the field. I understand why the car needs to be different than the tractor but being used to autosteer in tractors for 20+- years it makes Teslas AP seem not so impressive and a almost more of a nuisance than anything.

 

[Blacktes24]

People nod off all the time while driving.

Drowsy Driving | NHTSA

Learn the signs of drowsy driving and get resources on how to prevent the dangerous combination of being tired, fatigue and sleepy behind the wheel.

www.nhtsa.gov

Tesla driver slept as car was going over 80 mph on Autopilot, Wisconsin officials say

A 38-year-old driver from Palatine, Illinois, was ticketed and accused of sleeping behind the wheel of a 2019 Tesla, according to a sheriff's report.

www.nbcnews.com

Cops Chased a Tesla for 7 Miles While the Driver Apparently Slept | Digital Trends

Cops in California recently spotted a Tesla Model S motoring along the highway at 70 mph with the driver apparently asleep at the wheel. The case is still being investigated, but it appears that the Tesla's self-driving features and the officers' skillful driving helped to safely end the incident.

www.digitaltrends.com

People have medical emergencies while driving, sometimes dying behind the wheel.

If properly used, a Tesla vehicle will nag, warn, then alert the driver of inattention. That alarm could wake them up. Otherwise, the car will turn on the hazard lights and begin slowing in the lane until it's at a complete stop.

How does a wheel weight aid in these cases?

people fall asleep all the time in regular cars too. they just end up crashing sooner than they would in a tesla with AP.

 

[Dewg]

people fall asleep all the time in regular cars too. they just end up crashing sooner than they would in a tesla with AP.

If you were paying attention (pun intended) to my post, as long as the system is being used correctly, AP will slow the vehicle to a stop with the hazards on if the driver stops satisfying the wheel torque requirement or the cabin camera (if available) detects eyes are not on the road. The point is precisely to not crash the car.

 

[hybridbear]

What’s the status of this with 2022.44.25.5 FSD Beta 10.69.25? We’re currently on 2022.36.20 FSD Beta 10.69.3.1. I don’t want to update if there will be more strike issues. We only get 3 on our S because of no cabin camera. Right now our car is waiting for wifi to download & I’m not allowing it to connect to wifi.

 

[brancky3]

What’s the status of this with 2022.44.25.5 FSD Beta 10.69.25? We’re currently on 2022.36.20 FSD Beta 10.69.3.1. I don’t want to update if there will be more strike issues. We only get 3 on our S because of no cabin camera. Right now our car is waiting for wifi to download & I’m not allowing it to connect to wifi.

Can't tell you for FSD, but I'm on 2022.44.25.3 on my Model 3 and I got an 'hands-free defeat device detected' after 100 miles of driving with it today. No other warnings before that

 

[Blacktes24]

Can't tell you for FSD, but I'm on 2022.44.25.3 on my Model 3 and I got an 'hands-free defeat device detected' after 100 miles of driving with it today. No other warnings before that

well damn that sucks and is annoying. what do they do with that then? that sounds worse than just kicking out. And thats even without FSD?

 

[brancky3]

well damn that sucks and is annoying. what do they do with that then? that sounds worse than just kicking out. And thats even without FSD?

I dunno, I just took it off and restarted autopilot because I only had 30 miles to go. The rest of the drive was through downtown Atlanta on the freeway so I wouldn’t have gone hands free anyway.

 

[jsmay311]

I dunno, I just took it off and restarted autopilot because I only had 30 miles to go. The rest of the drive was through downtown Atlanta on the freeway so I wouldn’t have gone hands free anyway.

Did you have to stop and put it in Park before using Autopilot again after the disengagement?

 

[brancky3]

Did you have to stop and put it in Park before using Autopilot again after the disengagement?

No not in this case, because it didn’t technically give me the “red flag”, it just said defeat device detected and to remove it. I turn off autopilot, removed it, and turned it back on.

 

[Blacktes24]

No not in this case, because it didn’t technically give me the “red flag”, it just said defeat device detected and to remove it. I turn off autopilot, removed it, and turned it back on.

It would be interesting to know if it is giving strikes or just telling people to remove them. I also don't fully understand what strikes are or what it means if you get too many. lol