Vision only following distance

[insaneoctane]

I've got a 2018 TM3 with RADAR, but FSD beta installed so it's disabled. Been evaluating FSD beta for many months now and am very unimpressed specifically with following distance. It often follows too closely with the distance setting on max. It's also quite inconsistent, sometimes the distance is OK, while other times it's just hairy. My measure of uncomfortably close is less than one second. I'm guessing that vision just isn't as reliable depending on the characteristics of the vehicle you're following? Maybe something about the width or height screws up the distance estimate sometimes? Anyone else see this issue with FSD Beta?

 

[hgmichna]

The autopilot has never followed at a fixed distance. It has a distance range, and it follows fairly closely as long as the car ahead goes faster, even just a little faster, than the follower.

This is sensible behavior, because otherwise, for example, the autopilot would have to brake whenever an overtaking car goes into the own lane closely before the own car.

This has always been so, with or without FSD or FSD Beta.

 

[insaneoctane]

Thanks for your reply, but I disagree. I've found notable change since using AP on the freeway with RADAR disabled. One second (is "one Mississippi") behind a vehicle with the maximum "7" setting at freeway speeds is ridiculous IMO.

 

[KArnold]

In my observations, I'm not sure follow distance is functional in FSDB currently. Yesterday I was following a car on a suburban street - changed to follow distance to 7, waited maybe 30 seconds, changed it to 1 and waited again. No apparent difference.

 

[nvx1977]

I've got a 2018 TM3 with RADAR, but FSD beta installed so it's disabled. Been evaluating FSD beta for many months now and am very unimpressed specifically with following distance. It often follows too closely with the distance setting on max. It's also quite inconsistent, sometimes the distance is OK, while other times it's just hairy. My measure of uncomfortably close is less than one second. I'm guessing that vision just isn't as reliable depending on the characteristics of the vehicle you're following? Maybe something about the width or height screws up the distance estimate sometimes? Anyone else see this issue with FSD Beta?

Same car as you. yeah the follow distance is not very consistent. It used to be. Sometimes it's hard to tell if it's at the final following distance because AP now takes forever to get back up to set speed if it slows down for traffic. I have mine set to 2 (used to have it on 1 with radar). I haven't observed the too close issue, but maybe that's because being close is my preference and what is needed with the driving culture around here.

 

[S4WRXTTCS]

I have a 2018 with FSD Beta, and I don't have this issue.

I think you should do a camera calibration to see if that fixes it, and if it doesn't then bring it into service. At 70mph I didn't see any perceivable differences in following distances between Radar based TACC versus Vision based TACC. I typically use either a setting of 3 or 5 depending on traffic type.

Now I did see a lot of other differences like "slowing for bad weather" when it was raining out which radar based vision didn't do. I'm also experiencing what I consider to be excessive acceleration while in stop and go traffic. The smoothness of Tesla Vision TACC/AP just isn't where it needs to be on my 2018 Model 3 Performance.

Edit: It appears like the OP is referring to following distances while on FSD Beta during city streets, and the following distances don't apply. There is only the aggressiveness settings where the lowest setting is SUPPOSED to have longer following distances.

 

[S4WRXTTCS]

In my observations, I'm not sure follow distance is functional in FSDB currently. Yesterday I was following a car on a suburban street - changed to follow distance to 7, waited maybe 30 seconds, changed it to 1 and waited again. No apparent difference.

That's correct.

The following distances don't apply to city streets FSD Beta, but they still apply to divided freeways.

This does bring up a good test for the OP to do.

Disable FSD Beta with the slider and then test out the following distances to see what the behavior is.

 

[hugh_jassol]

The following distances don't apply to city streets FSD Beta, but they still apply to divided freeways.

It's exactly this. FSD Beta's follow distance is only affected by that one setting of Chill/Average/Assertive - the right-scroll setting only applies when using the "old" AP (i.e., on freeways)

 

[insaneoctane]

Interesting. Thanks for the feedback. I didn't know follow distance wasn't functional in FSD beta in city streets. Explains a lot. I have re-calibrated recently, but obviously that wouldn't change anything!

 

[Mark II]

I have also noticed an obvious difference in inconsistent following distance on freeway without radar (2018 vs 2022 Model S).

However, the biggest difference I have seen with AP without radar is that it only reacts to the car directly in front of me. I.e. the following distance on freeway with no radar does not account for 2nd car in front of the 1st car in front of me. With the 2018, if the 1st car directly in front of me was too close to the 2nd car in front of it the following distance would increase. Also if the 2nd car started to break my 2018 would start breaking even before the 1st car directly in front of me started to brake. Which overall felt safer to me.

Unrelated to the radar, I hate the fact the following distance is buried in a menu on the 2022 without any way to adjust using a stalk while driving. This is especially important because the car does not adjust following distance to compensate for rain.