-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
NoA is amazing.
- Thread starter Microterf
- Start date
powertoold
Active Member
So which one is it? You didn't disengage it for 2,000 miles, or you disengage it before it does something stupid because you expect it?
I know of at least 4 exits within 10 miles of my house that utterly confuse it into trying to take the wrong exit. Hard to see how you found 2,000 miles of highway that never once confused it.
I didn't have to disengage it on that trip.
The second one isn't about it asking asking wrong exit, it's about when it takes the exit ramp, it goes all the way to the white line on the edge. I disengaged that at that point when I first started using it because I didn't know if it would hit thr guard rail, etc.
EVRider-FL
Active Member
MXLRplus
Active Member
I've driven several kinds of autosteering systems. Here's my take on NoA's status:
Roughly 10% of my annual mileage is empty freeways. In those situations, every brand of auto-steering is good.
NoA doesn't work correctly in the Riverside County / Orange County / LA County region. Which is 16 million people, and 90% of my driving.
Reasons:
1) System does not look far enough ahead.
2) System does not judge relative speeds of surrounding vehicles accurately.
3) Navigation maps at complex interchanges have too many errors.
4) Car does not initiate directional changes in a timely fashion, such as curves and off-ramps. Appears to have about 1-2 car lengths of lag.
5) Lacks predictive AI to spot potentially dangerous situations. Today's driver needs to spot cars that are driving erratically more than ever before due to increasing cell phone usage in heavy traffic.
6) Cannot spot dangerous roadway debris, or detect lead cars avoidance of debris.
So I have turned NoA off. It offers nothing that standard AP lacks. When it comes to safety systems a 99% competent feature (1 error per 100 minutes) is not acceptable, and it's not at 95% yet.
Roughly 10% of my annual mileage is empty freeways. In those situations, every brand of auto-steering is good.
NoA doesn't work correctly in the Riverside County / Orange County / LA County region. Which is 16 million people, and 90% of my driving.
Reasons:
1) System does not look far enough ahead.
2) System does not judge relative speeds of surrounding vehicles accurately.
3) Navigation maps at complex interchanges have too many errors.
4) Car does not initiate directional changes in a timely fashion, such as curves and off-ramps. Appears to have about 1-2 car lengths of lag.
5) Lacks predictive AI to spot potentially dangerous situations. Today's driver needs to spot cars that are driving erratically more than ever before due to increasing cell phone usage in heavy traffic.
6) Cannot spot dangerous roadway debris, or detect lead cars avoidance of debris.
So I have turned NoA off. It offers nothing that standard AP lacks. When it comes to safety systems a 99% competent feature (1 error per 100 minutes) is not acceptable, and it's not at 95% yet.
It's just that extraordinary claims require extraordinary evidence. When most people can't get 10 miles without a disengagement, the claim that you did 2000 miles is extreme. It also feels a lot like it probably has some sort of bias, conscious or not. "Well, I disengaged for that, but I didn't expect NoA to handle that, so it doesn't count." or "Let's disengage now, because our exit is coming up" even though that is the literal purpose of NoA. You could even have been manually changing lanes (again, possibly unconsciously) to keep it away from difficult situations. You may just have a very high tolerance for really iffy driving that others would have disengaged for (or raged over when it just camps behind someone going 20 under). You yourself mentioned that 50 miles were driven with AP off- why? If NoA is perfect and doesn't need disengagement, why was it disengaged? 50 miles of driving on surface streets to get power, food, and sleep is a lot.
If you do a trip like this again, consider a dashcam pointed over your shoulder at the dash to back up an experience like this.
The 50 miles driven is because I don't have FSD Beta and therefore had to drive to take turns to charge.
I'm sorry that your experience has been less than optimal. It sounds like in certain areas or scenarios, it's pretty useless. I let the car initiate the lane changes. There are plenty where it would change the lane and change it back for no apparent reason. I always let it do that unless there are people that it will piss off.
My use case is that as soon as I turn onto a highway, I engage NoA. I let it do the merge, and I let it do everything it can, and I disengage it when I get to the end of my exit ramp and I have to get off.
I am taking another trip at the end of this month to Florida. I will bring a notepad to keep track of any interventions and any disengagents. Thank you for the suggestions.
. As far as recording it, I do have a few go pros, but I don't know how useful it will be as the batteries don't last every long. Especially for a 19 hour drive.
I'm sorry that your experience has been less than optimal. It sounds like in certain areas or scenarios, it's pretty useless. I let the car initiate the lane changes. There are plenty where it would change the lane and change it back for no apparent reason. I always let it do that unless there are people that it will piss off.
My use case is that as soon as I turn onto a highway, I engage NoA. I let it do the merge, and I let it do everything it can, and I disengage it when I get to the end of my exit ramp and I have to get off.
I am taking another trip at the end of this month to Florida. I will bring a notepad to keep track of any interventions and any disengagents. Thank you for the suggestions.
. As far as recording it, I do have a few go pros, but I don't know how useful it will be as the batteries don't last every long. Especially for a 19 hour drive.
Gopros work with external USB power. Long drives like that make for fun TimeWarp videos.
I'll have to check that out. I've only used them a couple of times. I do have plenty of battery packs. I'll try a couple of videos before my Florida trip to test it out.
I found mad max to be a little too harsh especially with people in the car. If you are on a divided highway the other settings transitions wont make your passengers get sick, But if the road is busy it's braking to fit into the spot in the next lane will be harsh no matter what. It is almost like it gets nervous in missing its off ramp if it its coming up . I did a long trip with the latest version and it is not bad at all ( double dot releases, tweak parameters - and there are A LOT of parameters- and some times they are worse not better for me) . it handled two round abouts for the first time since I have had it since 2017 ( albeit a little tentatively) and I'm glad no one else was in the round about....it is getting there!
Ahh, marketing. Tesla is amazing at it.
"Mad Max" does not change how aggressive a lane change is. It changes how often it suggests a lane change for slower traffic in your lane. Mad Max just means it will move over for a car only 1mph slower while normal might wait for 3 MPH. It's described in the info on the screen in the setting, or in this blog post: Introducing Navigate on Autopilot
I've only ever used it on mad max. It's nuts how non-aggressive it still is. It wants to move over into the right lane 1+ miles ahead of an exit even if that lane of traffic is going 20 under the speed limit. It means it's worthless in any kind of traffic.
"Mad Max" does not change how aggressive a lane change is. It changes how often it suggests a lane change for slower traffic in your lane. Mad Max just means it will move over for a car only 1mph slower while normal might wait for 3 MPH. It's described in the info on the screen in the setting, or in this blog post: Introducing Navigate on Autopilot
I've only ever used it on mad max. It's nuts how non-aggressive it still is. It wants to move over into the right lane 1+ miles ahead of an exit even if that lane of traffic is going 20 under the speed limit. It means it's worthless in any kind of traffic.
I routinely take 4-5 hour trips and NoA is mind blowingly awesome. It was worth the purchase price!
MXLRplus
Active Member
$10,000 on a $37,000 car that comes with AP standard? Huh.
I'm not poor, but that seems a lot for the added working functions that FSD provides over AP as of April 2021. I personally advise buyers to wait.
I bought my MX knowing that FSD did not work* but I purchased it anyways to see what it could and could not do. The auto lane change simply replaces the blindspot mirrors or digital review mirror found on other cars.
Soon after I started testing NoA, it nearly hit another car when it became confused on a Y-split (241) and oscillated between two lanes abruptly 3 times. And that was with me having my hands resting on the wheel. I had to jerk the wheel like a drunk to avoid the other car.
* NoA failed during the 15 min demo I took with the Tesla salesman driving his proven test loop in March 2020; it completely missed the offramp and the driver had to take control.
Last edited:
I have a late 2020 Model S, so it wasn't 37k. And I purchased FSD for $6k at that time. I didn't say you were poor; I don't even know you. I think NoA was worth what I paid for. Everyone else can make their own decision on their own. I didn't even say that it doesn't make mistakes -- I said I think its awesome, its useful to me and I'm glad I bought it. There are dime a dozen people taking shots at FSD/Autopilot daily. And its really counter-productive. The company is trying to advance technology in multiple ways, let them do their thing. If it's not your cup of tea, then don't buy their products.
That's cool that you bought a car a few months ago and had a ton of info about what FSD is and the history on it when you made your purchase.
There are people that paid for FSD in Jan 2017 with a promise that it was only 6 months away and a published full self driving video. Given the company that is "trying to advance technology in many ways" is completely unwilling to have any kind of discussion on a refund for these individuals, and has given multiple timelines that have been completely missed, they do have a valid complaint. Tesla has brought this upon themselves by charging for features ahead of time, not offering FSD transfers to other vehicles, and giving crazy timelines, it's not just people being "counter-productive". People are angry because they were lied to, and because Tesla continues to claim they are very close when they clearly are not.
Remember, this is what was sold in 2016. How would you feel if you paid $8K for this in 2016 and you have what FSD is now, 4.5 years later:
I use NoA on almost every trip out of the city. I find it handles heavy traffic pretty well on the whole. Not having to deal with steering whilst crawling along is quite pleasant. Wish i could better control how close it follows the car in front somtimes (the distance control thing doesn't seem to make a huge difference in traffic in my occasional experiments).
I still take control in roadworks or if the road layout is a total *sugar* show (not unusual in and around NYC). However, by far my biggest frustration is that it takes some exit ramps waaaay too quick. I let it go for as long as i can hoping it will start to slow down, but if it doesn't start braking by the time i reach my risk limit, i take control back. Maybe i'm too much of a conservative driver.
I still take control in roadworks or if the road layout is a total *sugar* show (not unusual in and around NYC). However, by far my biggest frustration is that it takes some exit ramps waaaay too quick. I let it go for as long as i can hoping it will start to slow down, but if it doesn't start braking by the time i reach my risk limit, i take control back. Maybe i'm too much of a conservative driver.
Similar threads
