Autopilot goof. Good thing I was paying attention :)

[tomas]

I got my hand slapped and one or more of my posts on this thread was removed for "snippiness". I apologize to the community and participants and especially whoever reported my post if whatever I said offended your sensibilities and/or you feel it had the potential to make your TSLA values decline. I am unsubscribing from this thread now.

You misunderstand. There is no issue with informed debate and strong opinions. The issue is with name calling and pejorative treatment of the others on the forum. That is against the forum rules and you stepped unambiguously over the line. And you continued in your apology.

 

[SammichLover]

Finally tried it today. Was pleasantly surprised that all the people complaining about how the torque requirement had been increased were full of it. Apparently those people just weren’t holding the wheel before (as I suspected)!

It isn't that simple, for a few reasons. First, you're in a Model 3. The Model S has a different steering wheel, and a years older design made and released under different software, so not a good idea to assume it is as easy on it.

Second, I had difficulty to start with when I got the car. You have to find the spot between firm enough to satisfy the "nag" requirement but not too resistive to knock the AP off. You happen to have landed in the right spot right away. I struggled until I found out it was a lot easier for me when holding it at "4 o'clock" only, which isn't particularly natural for me driving from but, as has been proven by events, it is a good home spot for my hands for when I do need to take over when something very unexpected arises.

The last thing is that at a given speed it may work without constant tensions, when the road is very straight, if you're relying on the wheel to move a little to create resistance there can be a lot of time between that and now at the 80mph+ range the nags are down to 10 seconds between. Very easy to get to blue flash, and even past that.

 

[SammichLover]

I'm sure that 7 hours in the 3 without EAP would be less tiring than 7 hours in the Prius because the 3 is a more comfortable car. But I'm also sure that 7 hours with EAP is far less stressful than 7 hours without it. And safer. I'm not going to perform an experiment that involves being less safe than necessary.

I've driven long distances in a lot of vehicles over the years, including rather comfortable ones.

Nothing, absolutely none of them, was even close to Model 3 using EAP. It isn't even really what I'd call physical difference, not one you can really locate, almost entirely mental effect and a general "I have no energy so I need to sleep" type of feeling.

 

[SammichLover]

Maybe its the Prius...try the same drive (maybe too late since you are moving) w/o autopilot in the Model 3 and compare your fatigue level with the Prius.

I've driven "luxury" cars, and "spacious" vehicles with good ergonomics on seats and physical controls, long distances. This isn't the niceness of the cabin or anything like that. The Model 3 seating has similar "seated for a 1000 miles" issues as you'll find elsewhere on very good seats. It isn't a physical fatigue that you can point to that it addresses. It is mental/energy. No other vehicle, save maybe for the couple thousand vehicles out there (?) equipped with SuperCruise on roads that match their whitelist, comes anywhere close to this.

The Model 3 is currently the flat out best long distance traveling vehicle in the US. It gets close in winter in the colder zones, due complications that put a bit of a pinch on range, but that's it.

 

[AlanSubie4Life]

You have to find the spot between firm enough to satisfy the "nag" requirement but not too resistive to knock the AP off.

Yeah, mostly this. It seemed relatively natural for me probably because of how I am used to holding the wheel.

Probably very annoying to some people. Tesla forces conformity. Fight the establishment!

The last thing is that at a given speed it may work without constant tensions, when the road is very straight, if you're relying on the wheel to move a little to create resistance there can be a lot of time between that and now at the 80mph+

I did not have any problems on 163 N which is straight for fairly long stretches but not super straight. For the sake of argument, let’s suppose I was theoretically doing 85 for this test Clearly I am not relying on the wheel to create resistance...

My point was that I got the impression here that somehow the torque sensing changed with ULC NoA. If your hands are detected to be on the wheel, it will change lanes. As far as I can tell there is no higher threshold required.

 

[SammichLover]

I did not have any problems on 163 N

You edited out an important part of my sentence? <edit> has been put back in I see, and talked about, but still focused on yourself rather than what seems to be a fairly large number of people (not me, mind you)

 

[SammichLover]

My point was that I got the impression here that somehow the torque sensing changed with ULC NoA

It does change the requirements, it appears to need triggering basically all the way through rather than a momentary tap that's good for several seconds.

That's very different than what people are used to having to provide.

 

[AlanSubie4Life]

You edited out an important part of my sentence?

Did not edit. Not intentional. Just quoting on my phone. Ninja editing now...post was missing something due to internet connection issues.

 

[AlanSubie4Life]

It does change the requirements, it appears to need triggering basically all the way through rather than a momentary tap that's good for several seconds.

That's very different than what people are used to having to provide.

I did not notice. Just holding the wheel. Maybe it is different, but effectively for me it is not. There was no way for me to tell anything was different. Maybe it is, but just saying effectively it was completely seamless (for me) with no behavior change required for me.

 

[SammichLover]

Maybe it is different, but effectively for me it is not.

For you. In a way you didn't understand it was different, just that Model 3 has not been showing you this, so you were accusing people of not having their hands and the wheel. :/

Kinda narcissistic, right?

And it isn't a "maybe", as explained in the first post of mine you responded to. The thumbwheel is also a way to trigger indication you have your hands of the wheel, and it does momentarily satisfy NoAP (the blue flash stops momentarily) but not enough for it to finish the lane change (well someday I'll try spin the volume up and down until the lane change completes ).

 

[AlanSubie4Life]

For you. In a way you didn't understand it was different, just that Model 3 has not been showing you this, so you were accusing people of not having their hands and the wheel. :/

I say "for me" realizing full well that other people's experience has been different. There's a reason for that difference, obviously - people are doing different things. Not being narcissistic.

I really wasn't accusing anyone of anything. <joke> As Steve Jobs would say, maybe I was just saying they were holding it wrong? </joke>

I was simply sharing my experience, for some people (who never had nag issues with the prior version of NoA), that they can continue to expect no issues, if they just keep doing what they were doing, or at least there is a good chance of it. I legitimately was very unclear myself on whether this would be the case, from reading this forum, prior to receiving the update. I was relieved that I had no issues and everything worked seamlessly. That's all I was saying. Of course by now anyone who cares already has the update and already knows how it works, and whether it works seamlessly for them!

And it isn't a "maybe", as explained in the first post of mine you responded to.

I said maybe, because I have no way to know if it is different. I understand what you're saying about the scroll wheel - that is different for sure (based on what you say). But I have no way to know whether this means that it requires torque on the wheel all the way through the lane change. Seems probable that the torque sensor might also require a long period input, but I have no idea.

ADD EDIT:
Also, not sure what there is to disagree with here, but I suppose you'll let me know. Maybe you never had nag issues before and now you do? I didn't say it was a guarantee - I just said there was a good chance. EDIT AGAIN: It's because I was lying that I hadn't accused anyone of anything.

 

[SammichLover]

I really wasn't accusing anyone of anything.

Apparently those people just weren’t holding the wheel before (as I suspected)!

....you'd have to be certifiable to take ..... your hands off the wheel even for an instant...

 

[AlanSubie4Life]

Ha. Guess you got me there. I have a VERY short memory. Though I stand by the third quote above.

 

[SammichLover]

Ha. Guess you got me there. I have a short memory. Though I stand by the third quote above.

Yeah, like a few minutes or so. Or it is motivated misremembering and you just make stuff up to fill the gap. Ex.

I didn't say it was a guarantee - I just said there was a good chance.

Because that's pretty bad that it keeps on happening over and over like this.

 

[AlanSubie4Life]

Yeah, like a few minutes or so. Or it is motivated misremembering and you just make stuff up to fill the gap. Ex.



Because that's pretty bad that it keeps on happening over and over like this.

It's actually more like less than 24 hours.

Sorry man. Feel free to point out the "over and over again". Not sure what you are referring to. I try to be as honest as possible, and I agree I definitely could have rephrased that post yesterday to be less accusatory. Regardless of the reasons, some people have issues with the ULC, that is a fact. I understand that. I was just glad it worked ok for me and got carried away.

 

[SammichLover]

It's actually more like less than 24 hours.

Because you don't bother read posts that you're replying to?

 

[AlanSubie4Life]

Because you don't bother read posts that you're replying to?

Not sure exactly what you're saying. It's true that I didn't go back and read my posts from yesterday; if I had I would not have claimed that I did not accuse people of anything.

I do read your posts. I'm sorry I accused you of misleading people, that was not my intent. This morning on my phone I had some issues with quoting and connectivity that made me reluctant to scroll back through and read the thread and quoting was being annoying. And then I lost connectivity and lost my edits to my post before posting it, and had to redo them. Again, sorry, and I hope we can move on.

It was my intent to share my experience, not to accuse people of "doing it wrong." Clearly my post from yesterday was not consistent with that intent. Sorry. I apologize.

 

[Roys3]

I think this scenario is similar to the problem I've been experiencing on mountainous interstates (which is where I have a use for AP). When a truck lane is starting the AP wants to center itself in the widening lane, then it sees a new line starting right in front of the car and either goes ungracefully to the left scaring passengers or aborts, especially if the truck lane is starting a curve with any real bend or rise where it can't get much of a look ahead. Here's an example, although this one isn't bad since you can see ahead, but AP does not handle these gracefully.

I also do not like AP going to the left on entrance/exit ramps when the lane widens and then swerving back over. It's unnerving. If there was only one of these situations every 25 miles or so, maybe it wouldn't be too bad, but in the mountains, every couple miles there is a truck lane and/or entrance/exit ramps. I'm sure it's better in the flat lands, but needs more work in the mountains.

View attachment 399506

Same here.
I want my "Beware, Student A.I." sticker - and my "A.I. Instructor" certificate.

 

[daniel]

I think this scenario is similar to the problem I've been experiencing on mountainous interstates (which is where I have a use for AP). When a truck lane is starting the AP wants to center itself in the widening lane, then it sees a new line starting right in front of the car and either goes ungracefully to the left scaring passengers or aborts, especially if the truck lane is starting a curve with any real bend or rise where it can't get much of a look ahead. Here's an example, although this one isn't bad since you can see ahead, but AP does not handle these gracefully.

I also do not like AP going to the left on entrance/exit ramps when the lane widens and then swerving back over. It's unnerving. If there was only one of these situations every 25 miles or so, maybe it wouldn't be too bad, but in the mountains, every couple miles there is a truck lane and/or entrance/exit ramps. I'm sure it's better in the flat lands, but needs more work in the mountains.

View attachment 399506

When I drove up to Canada it was mostly two-lane roads, and in the mountains there were frequent passing lanes where slower cars are supposed to stay to the left. I learned quickly that the Model 3 would stay in the lane directed by the dashed lines. Sometimes the dashed line led the car into the slow lane, ad sometimes into the fast lane. If there was no dashed line marking the expected lane, then it was anybody's guess. It was a matter of no effort at all to disengage or override EAP if it was going to go into the fast lane, and re-engage it a moment later. This is not a goof or a bug or a defect. It's a situation that EAP cannot handle. True FSD will bave to be able to handle it. But EAP is not that, and is not claimed to be that.

The first couple of times I might have said, "Oh, fooey, I have to override." But that was the part of me that still wanted to play the How Long Can I Keep EAP Engaged? game. Once I settled into just using it to be a safer driver, it was no problem to take over and then re-engage EAP a moment later.

Yes, people will do stupid things with it and get themselves killed. People do stupid thigs and get themselves killed all the time, and if they're not driving EAP as though it was FSD then they'll be dong something else equally dangerous.

 

[SammichLover]

(well someday I'll try spin the volume up and down until the lane change completes ).

FYI tried and this does NOT work. Doing so will continue to keep AP running indefinitely, you don't get the red-hands w/beep and then the red-wheel disengage. So it still satisfies the normal AP attention requirements. However NoAP will never start the lane change it wants to do, either, no matter how much you spin the volume and/or speed adjustment thumbwheels.

Maybe the programmers have created a mirror, with some changes, of the attention check to use in NoAP? I thought I'd checked this before when NoAP came out and the thumbspinner would work? Not super sure, since I decided NoAP wasn't worth continuing to use. That brings up the questions of if this was an intentional change in behavior, whether it will get changed [back?] in the future, or if AP's behavior will also get changed to follow this?