Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Navigate on Autopilot is Useless (2018.42.3)

This site may earn commission on affiliate links.
My 2010 Mercedes E350 has less Phantom braking than my 2019 Model 3 with the full FSD package :-(
It is an enigma to me why my Tesla continues to have this problem. It was solved a decade ago so why on earth is Tesla struggling so much with getting it right???

The reason is that it had a bunch of other sources like camera and ultrasonic sensors and it fuse the result and weighs it
 
  • Helpful
Reactions: pilotSteve
Hmmm, I seem to have the opposite experience. For me the software is quite reliable. Sounds so bad for you that maybe you will consider selling you car and getting a Volvo :)
I did not say the driving experience is not good. The flat torque curve with no transmission and smooth power application makes for a great drive. What I said was the software is not reliable. I stand by that statement. Maybe you believe that it is a good marketing tactic for Elon Musk to tout features that do not exist, but I do not.
 
You absolutely couldn't possibly be farther from the truth. But maybe now that you've driven with AP for a couple months, you see how foolish this statement was. The fact that AP makes "different" decisions isn't a concern. The fact that it makes terrible "decisions" is the concern.
Confused by your "a couple of months" phrase. I've been using NoA since 2018 and use it every day. Used it this morning. No disengagements other than the two times two different insane drivers almost merged into my lane. Zero terrible decisions made.
 
Confused by your "a couple of months" phrase. I've been using NoA since 2018 and use it every day. Used it this morning. No disengagements other than the two times two different insane drivers almost merged into my lane. Zero terrible decisions made.
yes, I was thinking the same thing. I was confused. Works perfectly for me. I guess not well for everyone though.
 
My 2010 Mercedes E350 has less Phantom braking than my 2019 Model 3 with the full FSD package :-(
It is an enigma to me why my Tesla continues to have this problem. It was solved a decade ago so why on earth is Tesla struggling so much with getting it right???
Tesla could easily get rid of phantom braking but then the car would run into more stuff. I bet your Mercedes would run right in to stopped cars way more often if left unattended.
Autopilot needs to be safer in real world usage so Tesla has to compensate for all the people who abuse it by making it err on the side of caution. Simple adaptive cruise control doesn't have that issue so it doesn't have to try to "see" stationary objects.
 
  • Informative
Reactions: pilotSteve
The people who don't own FSD yet reading this thread, please just know that some people here are overstating how useless NoA is. I use it every day and therefore isn't useless. It drives me to work. It's the best feature of any car I've ever owned, hands down.

Does it sometimes make stupid decisions still? Yes. It is getting better? Yes. It is useless? Only if you have insanely high expectations for the software, I guess.
 
  • Like
Reactions: DanCar
Confused by your "a couple of months" phrase. I've been using NoA since 2018 and use it every day. Used it this morning. No disengagements other than the two times two different insane drivers almost merged into my lane. Zero terrible decisions made.

yes, I was thinking the same thing. I was confused. Works perfectly for me. I guess not well for everyone though.

I can say that I have had mixed results with NOA. It works fine for most routine situations but I've had it completely miss my exits, oversteer too strongly when getting into an exit, wait too long to merge onto the freeway, make lane changes that don't make sense etc...

That's the challenge with developing a system to work "everywhere". There are a lot of different roads, drivers, conditions to account for. NOA might indeed work great for you and work terrible for someone else.
 
Maybe my problem here is misunderstanding the word "useless." Like, if I use it but still 100% pay attention to how it drives me, is that useless? I guess maybe that is what people are saying. If we could have some sort of criteria by which we could deem it useful, that would help me. When is NoA "good enough" for wk52 and Dr. Dabbles?
 
I can say that I have had mixed results with NOA. It works fine for most routine situations but I've had it completely miss my exits, oversteer too strongly when getting into an exit, wait too long to merge onto the freeway, make lane changes that don't make sense etc...
Sure, I have too have had mixed results and all those issues. Doesn't make it useless.
 
The people who don't own FSD yet reading this thread, please just know that some people here are overstating how useless NoA is. I use it every day and therefore isn't useless. It drives me to work. It's the best feature of any car I've ever owned, hands down.

Does it sometimes make stupid decisions still? Yes. It is getting better? Yes. It is useless? Only if you have insanely high expectations for the software, I guess.
I love Autopilot but I don't use NoA so there are differing opinions. I'd much rather be able to change lanes myself without disengaging Autopilot than use NoA. I will say that auto lane change works fine now that there's no traffic anymore.
99% of time on the interstate I'm just staying in a lane so having a machine help (somewhat poorly and unpredictably) with the other 1% just isn't very valuable.
 
  • Like
Reactions: boonedocks
Traffic is nearly back to normal in DC and auto lane change is working better than it ever has. It's pretty aggressive these days (not in terms of how often it changes--I have that set to "mild"--but in how it changes lanes either when it wants or when I ask it to).
 
The people who don't own FSD yet reading this thread, please just know that some people here are overstating how useless NoA is. I use it every day and therefore isn't useless. It drives me to work. It's the best feature of any car I've ever owned, hands down.

Does it sometimes make stupid decisions still? Yes. It is getting better? Yes. It is useless? Only if you have insanely high expectations for the software, I guess.
What is your commute like? How often does NOA do something versus regular autopilot. How many freeway exits does it get you ready for? I'm assuming that is what you like about NOA?
 
What is your commute like? How often does NOA do something versus regular autopilot. How many freeway exits does it get you ready for? I'm assuming that is what you like about NOA?
My commute is, roughly, entrance 495 (DC beltway) east to I-95N to 32E to exit, about 26 miles in total. So, it takes all 4 exits just great and also does a good job of navigating the traffic, especially in the morning. In the evening, when I go from I-95S to 495W, I ask the car to change lanes very quickly from the far right to the far left in order to avoid backups on the New Hampshire exit which are extremely common during the evening commute. While I wish the car understood that this was the best strategy, I don't expect it to understand that fact for now.
 
It sure would be nice for Dabs to acknowledge that his definition of "terrible decision" isn't universal, and that in general, instead of speaking in vague terms like that, be explicit about why he disagrees with the decisions the car is making, so that we can judge for ourselves if we think those decisions are terrible. Like I said, I rarely give a side eye to the car these days, though I admit it happens on occasion. My guess is our thresholds for giving the side eye are just radically different.

Like I'll give an example of a questionable decision it made today. It wasn't terrible, but I did roll my eyes. It moved into the passing lane and then moved immediately back out of it. That's pretty dumb. But it terrible? No. Does it render it useless? Hell no.
 
  • Like
Reactions: mikes_fsd
Current version is 2020.32.2. NoA is absolutely wonderful if I only have to get in a lane and never get out of it. Anything more and it’s... pause for dramatic effect... useless. I don’t mean “it’s ok but I disagree with some of the decisions it makes”, I mean “useless”. I’ve literally never had a positive experience with it despite trying to give it a chance to not suck for a period of time after every update. The only good thing I can say is that it hasn’t killed me yet, but that’s not necessarily due to lack of trying.

The thread title was, and continues to be, apt.
 
  • Like
Reactions: Pollux
Tesla could easily get rid of phantom braking but then the car would run into more stuff. I bet your Mercedes would run right in to stopped cars way more often if left unattended.
Autopilot needs to be safer in real world usage so Tesla has to compensate for all the people who abuse it by making it err on the side of caution. Simple adaptive cruise control doesn't have that issue so it doesn't have to try to "see" stationary objects.

I think its important to point out that there are multiple causes of phantom braking. Where a lot of times the false braking is the result of inadequate testing, and not because they tuned it to error on the side of caution.

Like for me the primary cause of phantom braking seems to be semi truck trailers. This seems to happen just as easily in TACC as in AP. I don't see the need for it under TACC. Adaptive cruise control these days is pretty widely available on new cars so there is no reason for Tesla to make it more complicated than it needs to be.

Those that use AP tend to get false braking because of incorrect speed limits which is once again a failing on Tesla. There is no official way of updating speed limits. So these things continue to happen.

It's so bad that people have been changing where they put their foot. I've come across comments on FB/TMC/Reddit saying things along the lines of "I hover my foot over the throttle so I can catch it fast", and that's crazy because it's not a safe way to drive.
 
Like for me the primary cause of phantom braking seems to be semi truck trailers. This seems to happen just as easily in TACC as in AP. I don't see the need for it under TACC. Adaptive cruise control these days is pretty widely available on new cars so there is no reason for Tesla to make it more complicated than it needs to be.

They are actually making it less complicated for themselves. Instead of messing with the sensor input part of the control loop, they just skip the steering part of the control algorithm. That's simpler for them. But that also brings in all the downsides from what's needed for steering and other things.
The steering part is probably calculated but just not used.
 
Last edited:
  • Like
Reactions: Daniel in SD
wait too long to merge onto the freeway, make lane changes that don't make sense

These are probably my main two gripes, especially the timing of merging when joining the main road. (UK)

Many on ramps have little or no dead zone at the end of the merging lane, and most every time on certain junctions, even with no traffic on the main carriageway, NoA puts on the turn signal then resolutely clings to on ramp lane (without slowing down either) to the point that you have no option but to take over control.

It's frustrating to start (most) NoA sessions with a disengagement at the first manoeuvre.