Things my wife said about Navigate on Autopilot tonight

[shellderp]

If AP is better than you, then you shouldn't be allowed to drive. That's as honest an opinion as I can possibly give.

what do you expect me to do with a p100d? not launch it at every possible opportunity?

 

[mswlogo]

I am not going to debate this point beyond saying that logically makes no sense. The correct terminology is geocoded speed limit. As I recall you simply said "posted speed limit". My understanding is the NVidia software, when the S used MobileEye (or whatever it was called) DID actually read posted speed limits and stop signs and traffic lights.

That's what I choose to call it. Doesn't matter if you agree with what I call it. I understand why you interpreted it that way.

They are all "PSL" (Posted Speed Limit). How you get that information varies. Saying "Posted" Speed Limit helps refer to what the legal "Speed Limit" is as opposed to some other "Speed Limit". On Radar forums, it's discussed to death and always referred to as PSL. You can get the PSL Electronically, by Human Eye Ball or Camera Eye Ball.

We can agree to disagree what you want to call it.

My point is at a Bridge the Car has TWO (Posted) Speed limits it gets from the maps (the Highway you are on and Cross Road) and the car (TACC) can momentary confuse the two (little blip on GPS position). And obviously it can't be reading the sign of both. It gets everything from the maps. This is all speculation. But the "-20 Offset Trick" seams to work for me and others. Order of magnitude less phantom brakes. The ones that do happen make sense, albeit overly conservative to brake or brake as hard as it does. But the "Bridge Shadow" ones appear to be eliminated. Also remeber on a system like this, it will always lean conservatively. If it thinks there is chance the (Posted) Speed Limit is lower it will take it. You could also read my "Posted Speed Limite" as "Legal Speed Limit".

 

[FMinMI]

I'm sorry you find "using your cars features as the manufacturer directs and intends" to be unacceptable.

I don't think that's a reasonable opinion but I suppose you're welcome to hold it.




And neither does Tesla.

Again you appear to be confusing autosteer and TACC.

TACC says nothing about limited access roads- just to avoid its use on city streets.




Where did they say anything about the speed limit. That's a minimum listed there, not a max.

Andthere's tons of highways where traffic speed is often 20-30 for periods of time.

Often long periods. Often even slower. Some places call this "rush hour"

Which is one reason you can engage it below 18 as long as there's a car in front of you.





Now here I agree with you. That SHOULD be something you can toggle... (and obviously if turned off AP wouldn't work)

But I also think it should be an option you have to actively change at least 1 or 2 deep in a menu- because otherwise you'll start getting people whining about how they rear-ended someone because they "didn't realize" they were only using CC instead of TACC if it's in any way "easy" to be unsure which mode you picked.

Sorry, but I never said anything about autosteer/autopilot in any of my posts (please go back and re-read them), just about TACC's poor performance (from a jerkyness/slam on the brakes standpoint), enough so to scare both passengers and drivers at times regardless of where it is used, so no, I am not confusing AP with TACC at all. All my comments dealt only with TACC (although in the OP he/she does mention AP).

Next point - I quoted right out of the manual that:

Traffic-Aware Cruise Control is primarily
intended for driving on dry, straight roads,
such as highways and freeways. It should not
be used on city streets.​

Highways and freeways, by definition, are limited access roads.

Finally, as for why I brought up the speed comment, I was showing how in the TACC section of the v9 manual (page 82) there are contradictions with its use - it talks about where it should only be used (see previous quote) yet it allows TACC to start at 18mph or above. YET they also say in THAT SAME SECTION that you should not (!diamond symbol) use it "on roads where traffic conditions are constantly changing." The latter is DEFINITELY a perfect definition of rush hour driving, at least in my city. So, the only real place that you can use it (if you adhere to the manual and all its warning) is on long, non-curvy, dry highways with traffic flowing ideally. That's BS.

But getting back to my first post to the OP, I still contend (and there have been many others who have voiced similar complaints in this forum and others) that the current programming/implementation of TACC sucks compared to other vehicles that have "TACC-similar" (under different names) cruise control and needs improvement.

I do agree with the last part of your post that if they do give the option for plain cruise control, there needs to be some way to alert the driver that is what is 'working' so as not to end up with a number of rear-end collisions.

Enough said...

Cheers.

 

[FMinMI]

Then your impression is demonstrably wrong.

All of these games were developed by people who don't work at Tesla.

Most of them by people who worked at Atari. Decades ago.

And the Atari Linux emulator came out in the 1990s. Also not written by people employed by Tesla to write it.

The racing game is a mobile phone game ported by the game developer, not Tesla to run on Linux (which again is what the Tesla runs)

Likewise the upcoming games like Fallout Shelter.


Tesla's "coding resources" going to this amount to "loading the already existing software someone who doesn't work for Tesla wrote and adding a button to launch it"


... what?

FSD didn't exist in 2014.

It didn't go on sale until October 2016.

For $2000 less than you just claimed.

So the games just 'appear on the screen' - no programming needed by Tesla to get them running? OK.

As for FSD, I screwed up - someone mentioned he had a 2014 and was pissed about not being able to get FSD that he paid for. Guess he was complaining about paying for AP only to be informed that the "promised FSD coming later" for his vehicle will never happen since he has HW Version 1. My bad.

But you are wrong about FSD - it was a $5K option in 2016. I have a 2016 and that is the price on my invoice, which also required an additional $3K for EAP, so $8K total. It of course has bounced around recently.
Maybe you got EAP and FSD confused when you said $3K ?

Cheers.

 

[DrDabbles]

what do you expect me to do with a p100d? not launch it at every possible opportunity?

What, is every other opportunity a difficult ask?

 

[Knightshade]

So the games just 'appear on the screen' - no programming needed by Tesla to get them running? OK.

Pretty much- yes.

All the programming was done by other people when they wrote the games- most of them decades ago... and by the fella who wrote the Atari emulator for Linux- also decades ago now.

When you buy a game at the store, that someone else wrote, and install it on your computer-usually telling it where to install, and maybe entering a few settings to set it up... do you consider that you doing "programming"? Because it's damn near the same thing in this case.

As for FSD, I screwed up - someone mentioned he had a 2014 and was pissed about not being able to get FSD that he paid for. Guess he was complaining about paying for AP only to be informed that the "promised FSD coming later" for his vehicle will never happen since he has HW Version 1. My bad.

HW1 was never promised FSD.... so still not sure what you're talking about.

FSD is a term they literally created in 2016.

But you are wrong about FSD - it was a $5K option in 2016

Nope. It was not.

. I have a 2016 and that is the price on my invoice, which also required an additional $3K for EAP, so $8K total.

Please post a picture.

Because what it should actually say is what I told you.

$3000 for FSD.

The $5000 was for EAP.

Maybe you got EAP and FSD confused when you said $3K ?

Nope, that was you.


Tesla announces all production cars now have fully self-driving hardware - Electrek

Notice the pictures direct from Tesla?

Like this one?

Tesla announces all production cars now have fully self-driving hardware - Electrek

$3000 for FSD. At launch.

Or this one?

Tesla announces all production cars now have fully self-driving hardware - Electrek

$5000 for EAP. At launch.


Like I've told you twice now.

 

[Knightshade]

Sorry, but I never said anything about autosteer/autopilot in any of my posts

Then why are you posting in a thread whose topic is literally autopilot?

(please go back and re-read them), just about TACC's poor performance (from a jerkyness/slam on the brakes standpoint), enough so to scare both passengers and drivers at times regardless of where it is used, so no, I am not confusing AP with TACC at all. All my comments dealt only with TACC (although in the OP he/she does mention AP).

Next point - I quoted right out of the manual that:

Traffic-Aware Cruise Control is primarily
intended for driving on dry, straight roads,
such as highways and freeways. It should not
be used on city streets.​

Highways and freeways, by definition, are limited access roads.

So first, no, highways are not by definition limited access (I can quote you US federal highway code if you like)- but second, it gives such as, not any exclusive language.

Second, it says "such as" which is not "only on"

In contrast autosteer says

Autosteer is intended for use only on highways and limited-access roads with a fully attentive driver

Not the "only" and the specific mention of limited access roads?

But getting back to my first post to the OP, I still contend (and there have been many others who have voiced similar complaints in this forum and others) that the current programming/implementation of TACC sucks compared to other vehicles that have "TACC-similar" (under different names) cruise control and needs improvement.

and as I said you're welcome to your opinion, I think it's deeply wrong though.

Every other companies system I've used has been terrible in comparison.

Prior to buying my 3 I tried systems from Lexus, Toyota, Infiniti, Nissan, Ford, and Caddy (though not Supercruise since the only car they sell it on sucks)...Probably forgetting a couple... all of them followed, even at nearest setting, WAY too far away so I would constantly be cut off and fall even further back... and none of them would even go down to 0 and resume (most cut off around 25-30)

I understand some of them have finally figured out stop/start in the last year or so, but no idea if they've improved elsewise- and their autosteer, for those who even offer it, remains largely inferior from all tests I've seen.

EAP is literally the reason I bought the car. Otherwise I'd have kept my Lexus.

(full disclosure, 95% of my driving is on interstates- probably much less of a big deal if I was doing 95% city driving)

 

[Octo]

I am not confusing AP with TACC at all. All my comments dealt only with TACC (although in the OP he/she does mention AP).

TACC is part of AP so what’s there to confuse?

 

[FMinMI]

Then why are you posting in a thread whose topic is literally autopilot?

Because although his post says AP, the jerkyness he mentions and the other comments others made deal with TACC, and as someone else mentioned, TACC is part of (E)AP/FSD..

So first, no, highways are not by definition limited access (I can quote you US federal highway code if you like)- but second, it gives such as, not any exclusive language.

"Highways are numbered routes, designed to carry more traffic, greater distances with more controlled access points (driveways or side streets)."
"Limited Access - A highway or section of highway designed for travel by registered motor vehicles. Access is limited to intersections, and driveways are generally not allowed. Freeways are a common type of limited access highway."

Second, it says "such as" which is not "only on"

In contrast autosteer says

Everything quoted was directly from the TACC section, page 82 not from any other section and how by strict reading of all the items in it, you cannot [Do Not] use it on roads where traffic conditions are constantly changing (e.g. rush hour traffic).

Every other companies system I've used has been terrible in comparison.

Prior to buying my 3 I tried systems from Lexus, Toyota, Infiniti, Nissan, Ford, and Caddy (though not Supercruise since the only car they sell it on sucks)...Probably forgetting a couple... all of them followed, even at nearest setting, WAY too far away so I would constantly be cut off and fall even further back... and none of them would even go down to 0 and resume (most cut off around 25-30)

I have experience with only 3 vehicles with "TACC" - all have done better than my MS. There are many posts in this and other forums about how it is jerky and has phantom braking. And most replies do not argue against that fact so either it is true to most owners OR folks who think it is fine as-is are not responding and are skewing the data. Also, maybe your M3 has improvements that my MS (and others) lack.

Pretty much- yes.

All the programming was done by other people when they wrote the games- most of them decades ago... and by the fella who wrote the Atari emulator for Linux- also decades ago now.

When you buy a game at the store, that someone else wrote, and install it on your computer-usually telling it where to install, and maybe entering a few settings to set it up... do you consider that you doing "programming"? Because it's damn near the same thing in this case.

That analogy is pretty crazy... But lets assume it only takes a Tesla programmer 10 minutes to install it. That, in my opinion, was 10 minutes that could have been spent working on way more important code to make TACC/EAP/FSD work better, safer and come to the market sooner. Any time spent on 'fun' stuff just pushes a delayed product out even further and that is not what Tesla needs right now.

Nope, that was you. [Regarding getting the price of EAP and FSD confused]

Yes, I eat crow here. FSD was $3K, then went to $5K (and now $6K?). I thought the $5K price tag was always the same. As for FSD, it was coined as you said in late 2016 but the folks who bought in 2015 and early 2016 were told "autonomous driving" and "self- driving" would come with future software enhancements. But when Oct 2016 came around, they found out they would not get FSD. I am glad I was not looking for a MS in 2015 as I too would have been pissed once I heard this.

HW1 was never promised FSD.... so still not sure what you're talking about.

Here is one link where "autonomous driving" was discussed/promised in early 2015. There was also stuff in the Teslamotors blog that said something similar. Had I been thinking about getting a MS in 2015 and read this, I would have expected to get what FSD is supposed to do.

Elon Musk Says Self-Driving Tesla Cars Will Be in the U.S. by Summer

That's it. I am done. TACC as-is today on my MS is crappy, needs improved programming and is way more important than games, karaoke or video streaming. You want to something to do while charging? Go help some old person load up there car (in MI and OH, most SC are at Meijers supermarkets), tidy up the area around the superchargers picking up crap that blew in, talk to the person in the ICE car who comes and asks you about driving an EV - do something extroverted instead of burying your head in a game...

 

[AlanSubie4Life]

Just a quick video showing the objectionable (to my wife) behavior (wasn't the specific situation that bothered my wife, but similar). To me, this was felt as a noticeable, unpleasant, mild jerk, which I would not have generated if I had been driving.

Note here:

Follow distance/time setting 5. NoA in use. Keep a close eye on the green/black line (which is plotted on a nonlinear scale FYI), and the brake lights of the vehicle in front.

1) Brakes applied by vehicle in front. No immediate response from the car.
2) Finally, it responds to the reducing distance to the van. Note that it very likely did not engage the brakes, since the regen was not maxed out (though it was quite high). Also note that this felt quite jerky - note how quickly the regen comes on - it was an almost complete, sudden release of the accelerator.
3) It was necessary for the TACC to reduce speed in this case, of course - the final speed was less than the initial speed by about 10mph. However, it clearly could have been more smooth.
4) The delay in response to the slowing of the vehicle in front in this case actually increased the jerk factor (rate of change of acceleration). It's pretty clear that a skilled, attentive, human driver would have eased off slightly upon appearance of brake lights, and maintained a healthy follow distance, with minimal jerk.
5) Notice the car is back on the accelerator again near the end of the video (then gets back into regen again!). There really would have been no need for this if it just hadn't slowed down quite as quickly when it did respond, and if it had responded sooner! It's apparently kind of under-damped.



Again, it's clear to me that in this case, a more moderate onset of regen could have led to the same net change in velocity, with both lower peak acceleration (magnitude of the bar), and lower peak jerk (rate of change of the bar).

It does show the trickiness of the problem, though. Too quick response to that initial slowing of the tracked vehicle would presumably lead to a lot of slowing down/speeding up all the time (which some people have complained about, including me). In this case, it was actually fairly slow to respond (perhaps due to the higher following time of 5?), but since the slowing continued, it suddenly decided it had to slow down quickly. The correct, optimal response is somewhere in between.

This also shows to me that I can drive more efficiently than Autopilot, at least while in traffic - much less energy slosh (which is inefficient) if you keep the bar short. People talk about AP being more efficient, but clearly that can't be the case if this is the typical behavior (though the differences in efficiency vs. optimal driving will be pretty minimal).

I feel like if they just put a cap on the peak jerk, unless an emergency is detected, they would have much smoother results AND better efficiency. Maybe Tesla control engineers should read the Wikipedia page on jerk? (I'm kidding - I'm sure they have already...but they may have other priorities for now...I hope that is the reason for this behavior, anyway.)

Jerk (physics) - Wikipedia

In general, for the people "doubting" this behavior (I'm not sure whether there actually is anyone who doubts it), you can just keep an eye on this green/black bar out of the corner of your eye while driving while in TACC/Autosteer (AP), or while in NoA. It tells the tale more clearly than any verbal description can.

 

[wcorey]

Because although his post says AP, the jerkyness he mentions and the other comments others made deal with TACC, and as someone else mentioned, TACC is part of (E)AP/FSD..




"Highways are numbered routes, designed to carry more traffic, greater distances with more controlled access points (driveways or side streets)."
"Limited Access - A highway or section of highway designed for travel by registered motor vehicles. Access is limited to intersections, and driveways are generally not allowed. Freeways are a common type of limited access highway."



Everything quoted was directly from the TACC section, page 82 not from any other section and how by strict reading of all the items in it, you cannot [Do Not] use it on roads where traffic conditions are constantly changing (e.g. rush hour traffic).


I have experience with only 3 vehicles with "TACC" - all have done better than my MS. There are many posts in this and other forums about how it is jerky and has phantom braking. And most replies do not argue against that fact so either it is true to most owners OR folks who think it is fine as-is are not responding and are skewing the data. Also, maybe your M3 has improvements that my MS (and others) lack.


That analogy is pretty crazy... But lets assume it only takes a Tesla programmer 10 minutes to install it. That, in my opinion, was 10 minutes that could have been spent working on way more important code to make TACC/EAP/FSD work better, safer and come to the market sooner. Any time spent on 'fun' stuff just pushes a delayed product out even further and that is not what Tesla needs right now.

Yes, I eat crow here. FSD was $3K, then went to $5K (and now $6K?). I thought the $5K price tag was always the same. As for FSD, it was coined as you said in late 2016 but the folks who bought in 2015 and early 2016 were told "autonomous driving" and "self- driving" would come with future software enhancements. But when Oct 2016 came around, they found out they would not get FSD. I am glad I was not looking for a MS in 2015 as I too would have been pissed once I heard this.



Here is one link where "autonomous driving" was discussed/promised in early 2015. There was also stuff in the Teslamotors blog that said something similar. Had I been thinking about getting a MS in 2015 and read this, I would have expected to get what FSD is supposed to do.

Elon Musk Says Self-Driving Tesla Cars Will Be in the U.S. by Summer

That's it. I am done. TACC as-is today on my MS is crappy, needs improved programming and is way more important than games, karaoke or video streaming. You want to something to do while charging? Go help some old person load up there car (in MI and OH, most SC are at Meijers supermarkets), tidy up the area around the superchargers picking up crap that blew in, talk to the person in the ICE car who comes and asks you about driving an EV - do something extroverted instead of burying your head in a game...

Frank, I traded in my 2012 Prius Plugin. I paid extra on it to get Adaptive Cruise Control aka TACC. It was FAR better than TACC. They also had LaneKeep aka AutoSteer but I couldn't get it with Adaptive Cruise Control. I am not happy either but, no, I am not going to dump the car because it does something the Prius didn't. It gets OTA software updates.

 

[North75]

Just a quick video showing the objectionable (to my wife) behavior (wasn't the specific situation that bothered my wife, but similar). To me, this was felt as a noticeable, unpleasant, mild jerk, which I would not have generated if I had been driving.

Note here:

Follow distance/time setting 5. NoA in use. Keep a close eye on the green/black line (which is plotted on a nonlinear scale FYI), and the brake lights of the vehicle in front.

1) Brakes applied by vehicle in front. No immediate response from the car.
2) Finally, it responds to the reducing distance to the van. Note that it very likely did not engage the brakes, since the regen was not maxed out (though it was quite high). Also note that this felt quite jerky - note how quickly the regen comes on - it was an almost complete, sudden release of the accelerator.
3) It was necessary for the TACC to reduce speed in this case, of course - the final speed was less than the initial speed by about 10mph. However, it clearly could have been more smooth.
4) The delay in response to the slowing of the vehicle in front in this case actually increased the jerk factor (rate of change of acceleration). It's pretty clear that a skilled, attentive, human driver would have eased off slightly upon appearance of brake lights, and maintained a healthy follow distance, with minimal jerk.
5) Notice the car is back on the accelerator again near the end of the video (then gets back into regen again!). There really would have been no need for this if it just hadn't slowed down quite as quickly when it did respond, and if it had responded sooner! It's apparently kind of under-damped.



Again, it's clear to me that in this case, a more moderate onset of regen could have led to the same net change in velocity, with both lower peak acceleration (magnitude of the bar), and lower peak jerk (rate of change of the bar).

It does show the trickiness of the problem, though. Too quick response to that initial slowing of the tracked vehicle would presumably lead to a lot of slowing down/speeding up all the time (which some people have complained about, including me). In this case, it was actually fairly slow to respond (perhaps due to the higher following time of 5?), but since the slowing continued, it suddenly decided it had to slow down quickly. The correct, optimal response is somewhere in between.

This also shows to me that I can drive more efficiently than Autopilot, at least while in traffic - much less energy slosh (which is inefficient) if you keep the bar short. People talk about AP being more efficient, but clearly that can't be the case if this is the typical behavior (though the differences in efficiency vs. optimal driving will be pretty minimal).

I feel like if they just put a cap on the peak jerk, unless an emergency is detected, they would have much smoother results AND better efficiency. Maybe Tesla control engineers should read the Wikipedia page on jerk? (I'm kidding - I'm sure they have already...but they may have other priorities for now...I hope that is the reason for this behavior, anyway.)

Jerk (physics) - Wikipedia

In general, for the people "doubting" this behavior (I'm not sure whether there actually is anyone who doubts it), you can just keep an eye on this green/black bar out of the corner of your eye while driving while in TACC/Autosteer (AP), or while in NoA. It tells the tale more clearly than any verbal description can.

It mainly seems to me that TACC prioritizes maintaining the absolute following distance over "smoothness". I typically drive withshort follow distance settings. 3, 2, or even 1 depending on the traffic level. I think when you are following closely it's more reasonable to expect that TACC would need to prioritize the absolute distance over smoothness. This is the way you would probably drive manually if you were trying to maintain a close following distance (some might say tailgating, although I think even at 1 it's a reasonable gap)
I think people that typically drive with a large following distance are using that space as a buffer so they can drive more smoothly and expect TACC to behave the same way.

For me personally at small follow distances, I think TACC behaves similarly as if I was driving. Perhaps not quite as smooth in all situations, but overall it's quite good and doesn't bother me at all.

Maybe Tesla needs a way to increase the following distance tolerance as the setting increases to optimize smoothness over following distance for those that want to drive that way.

 

[DopeGhoti]

Maybe for TACC (but not for EAB), Tesla could make the smoothing/damping factor correlate directly with configured follow distance? Or even just make it correlate not with configuration, but with actual (or, well, detected actual) distance.

 

[Dan_LA]

Wife: "Why is it so jerky?"

Me: "Well, it can decelerate somewhat abruptly while on freeway interchanges..."

Wife: "It's not just around corners, it's all the time on the freeway; why do you use it if it is so bad?"

Me: "Well, I'm trying it out; I think maybe you notice less if you're driving"

Wife: "Well, I notice it. It's ok to try it out if you're the only one in the car, but it isn't cool to do it otherwise"

Me: ...

Wife: If you keep trying to use it, I'm going to stop riding in this car, and we're going to have to take my car*. It's not relaxing.

Me: <disengages Autopilot for remainder of trip>

<a couple minutes pass>

Wife: It's much smoother now. Did you turn it off?

Me: Yes, I did.

Wife: I'm much more relaxed now.


* My wife's car is a Chevrolet Spark EV. That tells you something.

You can't make this stuff up!

Here's hoping that Tesla discovers low pass filters soon.

OMFG do we have the same wife? Had this talk like 5 times probably. But my wife added that she hate the sound of AP, it wakes up the kids in a long trip hah.

 

[EVS Motors]

Navigate on Autopilot is horrible, it makes so many unnecessary lane changes. lol I end up turning mine off because its so annoying and just use standard autopilot.

 

[DrDabbles]

Follow distance/time setting 5.

TrY SEtTinG iT To sEvEN

Obviously the problem is you and not the software doing these things.

 

[wcorey]

Navigate on Autopilot is horrible, it makes so many unnecessary lane changes. lol I end up turning mine off because its so annoying and just use standard autopilot.

I use NoA all the time on limited access roads. What I do no do is have it set to automatically make those lane changes. I simply ignore the permission request. After a few seconds it goes away.

 

[Phlier]

Just a quick video showing the objectionable (to my wife) behavior (wasn't the specific situation that bothered my wife, but similar). To me, this was felt as a noticeable, unpleasant, mild jerk, which I would not have generated if I had been driving.

Note here:

Follow distance/time setting 5. NoA in use. Keep a close eye on the green/black line (which is plotted on a nonlinear scale FYI), and the brake lights of the vehicle in front.

1) Brakes applied by vehicle in front. No immediate response from the car.
2) Finally, it responds to the reducing distance to the van. Note that it very likely did not engage the brakes, since the regen was not maxed out (though it was quite high). Also note that this felt quite jerky - note how quickly the regen comes on - it was an almost complete, sudden release of the accelerator.
3) It was necessary for the TACC to reduce speed in this case, of course - the final speed was less than the initial speed by about 10mph. However, it clearly could have been more smooth.
4) The delay in response to the slowing of the vehicle in front in this case actually increased the jerk factor (rate of change of acceleration). It's pretty clear that a skilled, attentive, human driver would have eased off slightly upon appearance of brake lights, and maintained a healthy follow distance, with minimal jerk.
5) Notice the car is back on the accelerator again near the end of the video (then gets back into regen again!). There really would have been no need for this if it just hadn't slowed down quite as quickly when it did respond, and if it had responded sooner! It's apparently kind of under-damped.



Again, it's clear to me that in this case, a more moderate onset of regen could have led to the same net change in velocity, with both lower peak acceleration (magnitude of the bar), and lower peak jerk (rate of change of the bar).

It does show the trickiness of the problem, though. Too quick response to that initial slowing of the tracked vehicle would presumably lead to a lot of slowing down/speeding up all the time (which some people have complained about, including me). In this case, it was actually fairly slow to respond (perhaps due to the higher following time of 5?), but since the slowing continued, it suddenly decided it had to slow down quickly. The correct, optimal response is somewhere in between.

This also shows to me that I can drive more efficiently than Autopilot, at least while in traffic - much less energy slosh (which is inefficient) if you keep the bar short. People talk about AP being more efficient, but clearly that can't be the case if this is the typical behavior (though the differences in efficiency vs. optimal driving will be pretty minimal).

I feel like if they just put a cap on the peak jerk, unless an emergency is detected, they would have much smoother results AND better efficiency. Maybe Tesla control engineers should read the Wikipedia page on jerk? (I'm kidding - I'm sure they have already...but they may have other priorities for now...I hope that is the reason for this behavior, anyway.)

Jerk (physics) - Wikipedia

In general, for the people "doubting" this behavior (I'm not sure whether there actually is anyone who doubts it), you can just keep an eye on this green/black bar out of the corner of your eye while driving while in TACC/Autosteer (AP), or while in NoA. It tells the tale more clearly than any verbal description can.

Yeah, wow... that was a very quick deceleration, followed by a very confused acceleration/deceleration moment of confusion. Scrubbing off 10mph that quickly couldn't have been pleasant. Your wife has my condolences.

Your vid is a very good example of one of the things that NoA does very poorly. Unfortunately, the list of what it does poorly is very long at this point.

 

[wcorey]

TrY SEtTinG iT To sEvEN

Obviously the problem is you and not the software doing these things.

People talking about what number to set follow distance to. When I was 15 1/2 they taught up 1 car length for every 10 mph. There absolutely should not be a set distance as it varies depending on how fast you are going. The software knows how fast you're going and how far away the car in front of you is. It should be pretty simple but clearly nobody taught those developers the 1 car length per 10mph. I've seen nothing in the owners manual or written on here to suggest that numeric setting is not speed adjusted.

 

[Phlier]

People talking about what number to set follow distance to. When I was 15 1/2 they taught up 1 car length for every 10 mph. There absolutely should not be a set distance as it varies depending on how fast you are going. The software knows how fast you're going and how far away the car in front of you is. It should be pretty simple but clearly nobody taught those developers the 1 car length per 10mph. I've seen nothing in the owners manual or written on here to suggest that numeric setting is not speed adjusted.

I could be wrong about this, as I'm simply repeating what I've read a few times here on the forums, but supposedly, each number on the follow distance is 1/2 second of time. So it's time based, not speed or distance. Well, actually I guess you *could* say it's distance, as it's a distance based on the time it would take to proceed that particular distance based on current speed.