Any hack to remove the autopilot nag?

[S4WRXTTCS]

That’s not a feature I’d pay for anymore.

I'd say everything about AP has been ruined.

TACC doesn't work anymore because of all the false braking events. They're in there because Tesla is trying to make it do more than it should do.

AP is reliant on a Torque Sensor that isn't sensitive enough so it just nags you.

NoA is completely broken because the "without confirmation" waits too long to change lanes.

It's so bad that the adaptive cruise control in my Jeep Wrangler is more convenient than anything EAP/FSD offers at this point. Now of course I can't road trip in that because of poor fuel economy, and its way TOO LOUD.

 

[cucubits]

TACC doesn't work anymore because of all the false braking events. They're in there because Tesla is trying to make it do more than it should do.

AP is reliant on a Torque Sensor that isn't sensitive enough so it just nags you.

NoA is completely broken because the "without confirmation" waits too long to change lanes.

Sadly for Tesla I find none of these problems with AP1. Looks like they still haven't measured up...

 

[S4WRXTTCS]

Lol good find, I missed that. But clearly that's being overlooked. The emphasis is on AP and Tesla.I believe the accident could have been avoided had he not been on his phone.

I think the NTSB is being pretty dismissive of that because of all the studies done with L2 system, and driver distraction. That people are using these systems as an excuse to do distracted things.

It's hard to fight human nature so I think the NTSB is just giving into it.

Where in other countries they take personal responsibility way more seriously.

 

[S4WRXTTCS]

Sadly for Tesla I find none of these problems with AP1. Looks like they still haven't measured up...

I had AP1 on a Model S from 2015 until 2018 when I traded it in on the P3D.

I absolutely loved TACC on the Model S. It hardly ever had any false braking, and was WAY smoother than the P3D TACC.

I didn't like AP on the Model S though because it did too many micro corrections, and occasionally would have truck lust. I didn't have any issues with the nag. The steering wheel torque sensor was simply more sensitive than what I have in the P3D.

 

[Knightshade]

Sadly for Tesla I find none of these problems with AP1. Looks like they still haven't measured up...

I don't have any of them with AP2.5 either- and a LOT more functionality than AP does.

Well, I can sorta agree the no-confirmation lane changes are slower than the ones when you use the blinker yourself I guess... but I haven't found it enough of an issue to not use it or anything.

Apart from that the nags are no issue if you hold the wheel correctly, and I've had no "phantom" braking issues in almost 20,000 miles almost all of it using EAP. It sometimes brakes in a way I might not, though even that's rare- but never "for no reason".

 

[Fernand]

Leaving out the people who are happy with how it works for them now, on their car

===> For the others, who are not (so happy) can I please ask a question: <=====

If the driver attention "interlock" was working as intended, i.e. 1) ready to nag but perfectly satisfied with a driver's hand(s) on the wheel in position to take over, and 2) it didn't require additional unnatural "jiggles", or 3) drop out of autopilot from said "jiggles", would you be OK with it?

If not, why not?
.

 

[Knightshade]

FWIW I'm not "happy" with the torque sensor nag system- I just don't find it onerous enough or hard enough to use correctly to not otherwise enjoy a system that works far better than anything I've used on a half dozen other brands of car.

I'd certainly prefer a works-awesome camera system for driver attention- but even Caddy who's I guess the "leader" in that specific metric- has had enough problems with their system it's being significantly revamped.... and given the ship has sailed on such a system being designed into the S, 3, X, or Y it pretty much is what it is at this point for Tesla.

And much like other things not designed-in like rear-cross-traffic it's pretty clear they keep designing to the idea FSD will be here Any Day Now so why worry about making L2 better in the long term when we'll be L5 driving cross country Real Soon Now... (and thus a Tesla driver's a lot safer backing into spaces, same way the car will automatically some day- since they put a radar on the front but not the back)

 

[Fernand]

@Knightshade so you'd still hate sensing the hand on the wheel, even if the torque sensing worked perfectly?

'Cause I think with a little effort, Tesla could make the sensitivity adjustable, tunable for the driver, and the system thereby working much much better.
.

 

[S4WRXTTCS]

I don't have any of them with AP2.5 either- and a LOT more functionality than AP does.

Well, I can sorta agree the no-confirmation lane changes are slower than the ones when you use the blinker yourself I guess... but I haven't found it enough of an issue to not use it or anything.

Apart from that the nags are no issue if you hold the wheel correctly, and I've had no "phantom" braking issues in almost 20,000 miles almost all of it using EAP. It sometimes brakes in a way I might not, though even that's rare- but never "for no reason".

I get one false braking event on average about once every 400 miles. It typically happens on the same Snohomish to Portland, and back trip on I5 (and parts of 405 sometimes). The last one it was showing that it was tracking the car next to me, but slightly ahead of me. It was a little more serious braking than the typical TACC false braking. It was about 20mph of speed reduction for nothing.

Now I haven't stopped using TACC nor have I become one of those "I hover my foot about the accelerator" types. But, I am hugely disappointed in it.

As to Lane changes:

On my Model 3 there is absolutely no contest between confirmed or user initiated lane changes, and non-confirmed lane changes. The initiated lane changes are correctly timed, and meet the "as good as me or better" criteria pretty much every time. The non-confirmed lane changes make me look like an idiot with the delay.

I do have an appointment for the HW3 upgrade in Mid-March, and I remain hopeful that it will fix the non-confirmed lane changes like others with HW3 have reported.

 

[S4WRXTTCS]

Leaving out the people who are happy with how it works for them now, on their car

===> For the others, who are not (so happy) can I please ask a question: <=====

If the driver attention "interlock" was working as intended, i.e. 1) ready to nag but perfectly satisfied with a driver's hand(s) on the wheel in position to take over, and 2) it didn't require additional unnatural "jiggles", or 3) drop out of autopilot from said "jiggles", would you be OK with it?

If not, why not?
.

My frustration with the torque sensor is well below other aspects of AP that bug me. The false braking, the lack of accurate maps for NoA to work properly, the tendency for AP2 to recenter itself in the right lane anytime there is a merge, etc.

Most of the benefits of a proper driver attention system is in future-proofing for when the system gets really good, but where its still L2 or L3. It also future proofs it for future sales in countries where regulations will require a proper driver monitoring system.

It also allows Tesla to accurately say whether a person was paying attention or not. They currently make claims like "drivers hands weren't detected in X amount of time" when the system has no real way of knowing whether the hands were on the steering wheel or not.

 

[Fernand]

@S4WRXTTCS I sincerely hope that upgrade makes you less hugely disappointed. I'm looking forward to it too. But, honestly, and this may not be the ideal congregation to say this in, but I'm surprised when people feel that way. Is it a matter of expectations? Maybe I'm relatively not as hot-shot a driver as some. Or is their car malfunctioning?

Because pimples and all I'm pretty hugely happy as punch with the way my Nicki works now, after a little tuning to make the torque sensing more responsive. It's a joy to sit in, to operate, and it gets me there, almost as un-tired as riding honorable bullet traino. My NOA works fine. It's the Jetsons, dude, I don't care what "level" it's called. And I don't mind doing some things manually, it's 2020, no other car comes close. It'll outgrow the pimples, because they ain't warts or fissures or melanoma. Can't wait for the next chapters.
.

 

[drtimhill]

Remember. The nag was not a Tesla idea. Original Autopilot had no nag. Only when some owners began to post pictures of themselves in the back seat or eating meals with both hands, did the pressure come on Tesla to add the nags.

Whenever you are annoyed by the nag, put the blame where it belongs. On the self absorbed idiots that did not think before posting doing dangerous things while on autopilot.

Exactly. Which is why I push back hard on the whole "nag defeat" thing. It's fodder to the "Ban Autopilot" brigade. And who will be the first people to whine if that happens? The "nag defeat" guys (you know who you are).

 

[ElectricIAC]

I'd say everything about AP has been ruined.

TACC doesn't work anymore because of all the false braking events. They're in there because Tesla is trying to make it do more than it should do.

Agreed. TACC is 100% awful about keeping a consistent speed anymore between...

a) Resetting your set speed randomly to some variation of mapped PSL (+0 up to +5), and sometimes significantly lower if it loses this data (because it doesn’t read PSL signs)
b) Adjacent Lane Speed is buggy as all get out and frequently tracks single cars adjacent failing to overtake.
c) Phantom braking (however this appears to occur less often with HW3)

 

[pscarroll]

Yes there is, purchase FSD, park your car in a garage for two years (plugged in of course), don't use it, come back after those two years, presto... No more nags.

 

[Knightshade]

@Knightshade so you'd still hate sensing the hand on the wheel, even if the torque sensing worked perfectly?

'Cause I think with a little effort, Tesla could make the sensitivity adjustable, tunable for the driver, and the system thereby working much much better.
.

As I say, it mostly works fine for me already.... I didn't say I hate it, just that I'd prefer a better system but recognize that'd require significant HW changes and there's roughly 0 chance of Tesla doing that short of NHTSA forcing them, a thing they seem totally disinterested in doing.

 

[BikerPeaBody]

Yes there is, purchase FSD, park your car in a garage for two years (plugged in of course), don't use it, come back after those two years, presto... No more nags.

so you saying FSD will be here in 2 years? completely no driver attention nor intervention needed??

 

[Fernand]

It's funny, that when you think about it, it seems it's the guys who believe that FSD is a specific endpoint, like an SKU, and should be delivered very soon who are the most "disenchanted". It's like they don't accept shortcomings because its' SUPPOSED to be driving your kids to school tomorrow. Well, it's not. It's not a deliverable, it's not even fully spec'ed. The spec is still evolving.

As somebody who's developed real time control systems for years, I know how easy it is to get over-optimistic on things that have never been done before, so I have no ill feelings towards the developers/company. It's a lot harder than anyone expected. And what they've accomplished so far is very impressive, to me anyway.

I'm pretty sure that for any car to self-drive without any human supervision is several years out. What we can expect is gradual evolution. I expect that it will NOA, i.e. self-drive, on city streets, while keeping a hand on the wheel, and requiring occasional intervention, fairly soon, as these things go. Nothing wrong with that. But we won't hear the end of it - what? requiring hands on the wheel?

BTW I can't see using TACC. Who's using THAT? Without auto-steer and without ability to change lanes by tapping the turn signal, it's just a cruise control, and not very useful.
.

 

[Fernand]

@Knightshade I'd like to understand what you'd prefer to a sensitive and working hands on wheel detector. Every alternative, like cameras, seems more intrusive and less reliable to me, less able to confirm that the driver can take over quickly. I mean, if the driver's face/eyes are detected by camera to be "paying attention", but he's lying back in the seat and can't grab the wheel quickly, what's the advantage? Or if he's paying attention but his hands are tied up.

What exactly bugs you about using torque sensing?
.

 

[Knightshade]

@Knightshade I'd like to understand what you'd prefer to a sensitive and working hands on wheel detector. Every alternative, like cameras, seems more intrusive and less reliable to me, less able to confirm that the driver can take over quickly. I mean, if the driver's face/eyes are detected by camera to be "paying attention", but he's lying back in the seat and can't grab the wheel quickly, what's the advantage? Or if he's paying attention but his hands are tied up.

I can think of a bunch of examples where either system is better than the other. Quick example of each-


Hand on wheel, but other hand holding phone reading a book or playing a game or whatever-not looking at the road at all- the camera system is vastly superior isn't it?

Looking out at the road, but holding a big messy sandwich and eating it with both hands so that if you need to take over you have to put the food down someplace then move to the wheel? The wheel torque sensor is much superior to avoid that.


Overall, if we pretend it's a binary choice- EITHER insure there's "some interaction with wheel" OR "eyes are on the road" I think the second one is probably at least marginally safer... you're at least insuring the driver is AWARE of what the car (and surrounding cars/animals/whatever) is actually doing... versus the wheel system where you're insuring there's a slight speed advantage for the driver to take over but not that he'll ever notice he NEEDS TO.

It's not a binary choice of course- but if you make it one a well-working camera system probably kills fewer lazy people (and is certainly harder to "defeat" than the torque sensor is.)


But as I say- Tesla has pretty clearly already made its choice on attention systems- and won't be changing unless someone forces em to- and NHTSA seems actively disinterested in regulating this stuff... (that's among the things NTSB was complaining about at the news conference)

What exactly bugs you about using torque sensing?
.

Mainly how much people complain about it

I mean the fact it's so much more easily defeated/fooled isn't great either of course.

BTW I can't see using TACC. Who's using THAT? Without auto-steer and without ability to change lanes by tapping the turn signal, it's just a cruise control, and not very useful.
.

As long as Tesla restricts autosteer speeds to 45-no-higher in places the database doesn't have speed limits, and 5-over-max anywhere it doesn't think is a proper spot for AP, then TACC will be an option for folks who are on a road where all traffic is going significantly faster than AP allows, especially where there might be some stop and go, to make driving significantly nicer/easier than not using anything.

 

[DopeGhoti]

I saw a video a while back by a guy who showed a proof-of-concept of another means to detect hands-on-wheel by, rather than relying on steady torque in one direction or the other, sending commands to the wheel to -- and this next word is important -- imperceptibly pull in alternating directions, and measure the resistance that way. Granted, it was a POC shown by the creator so of course it would appear to work well, but... it appeared to work well.