Second thoughts on FSD

[WannabeOwner]

Shame AP impressed on some and gets it so wrong on others

I don't think there is much where "AP gets it so wrong" (happy to hear other viewpoints though), except when using it outside what it is currently intended for - of course Tesla allow any usage you like - I've manged to engage AP on a single track road because a few yards of central divide paint on a blind bend brought the T-AP-Logo on , and YouTubers are posting how well AP is progressing (or not) - including one who used spray-paint on his lawn and got AP to engage there . If you stick to Dual Carriageways and maybe Major single-lane A-Roads (even then it may dislike the road and cut the speed back well below 60 MPH limit) then I can't remember having a single occasion where AP got it wrong that I wasn't expecting (so the "brow of a hill" and "Car in front pulls over to reveal parked car" type issues)

People have been killed driving into a concrete divider, where AP followed the wrong white lines because a) they were badly maintained and b) road workers were supposed to put a sandbag defence there, when they removed the armco, but "hadn't bothered". In those instances there was enough time for driver to react (as demonstrated by follow-up YouTubes driving the same stretch of road and getting the same AP behaviour, and then taking over manually) so clearly they were, very sadly, not paying attention

 

[goRt]

Interesting Which? has waded into the FSD claims to the ASA (onwards to the NL authorities I guess):
Tesla Model 3: don’t be fooled by ‘full self-driving’ claims – Which? News

 

[Peteski]

I'm not sure how or even if AP and AEB work together. It's clear that when driving manually, AEB will only wipe off an initial 25 mph when it anticipates a collision, but thankfully I've never got beyond the AEB warning signal to experience it (warning sensitivity set to early). In AP1 or AP2 the car will of course attempt to make a complete stop, so AEB is kind of redundant. I presume AP will slam on the brakes just as hard as AEB would have done, although you would expect AP to react sooner to an impending collision.

I don't know the history of Tesla AEB development and functionality and how that corresponds to AP1 and AP2. I expect it is somewhat independent except for sharing the same sensors as AP.

 

[WannabeOwner]

I presume AP will slam on the brakes just as hard as AEB would have done, although you would expect AP to react sooner to an impending collision.

That's a good point:

AP : "Car in front is slowing, start applying regen/brakes"
AEB : "Car in front is slowing, but applying the brakes isn't urgent ... yet"

If nothing else the first one has less chance of you getting hit from behind

 

[MrBadger]

I don't think there is much where "AP gets it so wrong" (happy to hear other viewpoints though), except when using it outside what it is currently intended for

Phantom braking for overhead structures (see video below), annoying tacc braking and thinking that it needs to suddenly slow down on a fast elevated section because a lower road has a different speed limit for starters...

All within the current control domain of autipilot.

 

[Peteski]

Interesting Which? has waded into the FSD claims to the ASA (onwards to the NL authorities I guess):
Tesla Model 3: don’t be fooled by ‘full self-driving’ claims – Which? News

I agree entirely with that article. Tesla are really pushing the boundaries when it comes to hyping up FSD and its real capability.

 

[WannabeOwner]

Phantom braking

I don't regard that as "so wrong", but fully understand that may not be the norm. My definition of a category such as "so wrong" would be "tries to drive me into the ditch / oncoming traffic" and I can't remember a single time that has happened.

I drive with foot-over-accelerator, when phantom breaking happens I can counter promptly. I am rarely able to guess when that is likely, so I am just compensating for it when it happens.

 

[Peteski]

Phantom braking is extremely annoying and potentially dangerous. Latest software seems to have regressed a little too in this respect. Although could be just the sunny weather with more harsh shadows around.

 

[Durzel]

I don't regard that as "so wrong", but fully understand that may not be the norm. My definition of a category such as "so wrong" would be "tries to drive me into the ditch / oncoming traffic" and I can't remember a single time that has happened.

I drive with foot-over-accelerator, when phantom breaking happens I can counter promptly. I am rarely able to guess when that is likely, so I am just compensating for it when it happens.

Problem is in a lot of situations if the car randomly heavy brakes there's an above average chance someone will go into the back of you.

Obviously people should leave plenty of distance and expect the unexpected, but in reality people don't expect others to randomly brake hard. People often don't leave the sort of gaps that would be safe, and in certain situations leaving big gaps will just mean someone else drives into them.

You'd look like you're trying to carry out a crash-for-cash scam, and looking like you're brake testing people is sure to lead to road rage.

 

[Mr Miserable]

Just modified my order and binned the FSD.

 

[ivangr]

FOMO really.

I must admit I have been checking and unchecking FSD on the order page, and I like to think of myself as wise to sales tricks. I know it's limited in functionality now, I know it's likely to be eclipsed by whatever is out by the time I would be looking to change cars anyway. Total head logic would dictate that the wisest choice would be to ignore it this time around, and see where things are in 2-3 years when I change up to the latest model. But, I keep considering it.

The idea that it (apparently?) costs £7000 to retrofit, and with Musk talking about a $1000 price rise on August 16th, it does give me pause for thought about whether I should just get it now - accept it as a sunk cost, and hope it sees the development promised sooner rather than later.

It doesn't help that I like toys

same for me. i’ve stuck with it now. i like toys. my worry is i wouldn’t get it and then just obsess about it until i have it.

also, i reckon the elon has achieved enough of the “impossible” to earn some faith. reusable rockets, mainstream electric cars and so on.

 

[MrBadger]

If anyone is unsure, imho better to leave it on until latest possible point. Then when rumoured August price rise comes, you are hopefully locked in at lower one.

 

[tess19]

Or when they offer free delivery

 

[Mr Miserable]

rumoured August price rise

Have I missed something here? Please rumour me

 

[RjbT3]

FOMO really.

I like to think of myself as wise to sales tricks.

It doesn't help that I like toys

I completely agree with your post. For me it's FOMO too. I know that it's nowhere near actual FSD capability and probability won't be for years. But I also know that if Tesla introduce some new cool features, I'd want to upgrade at the higher price down the line. It's definitely a bit of FOMO and new toys.

Interesting Which? has waded into the FSD claims to the ASA (onwards to the NL authorities I guess):
Tesla Model 3: don’t be fooled by ‘full self-driving’ claims – Which? News

Which? are incredibly irritating when it comes to Tesla. Did anyone read their full review of the Model 3? They clearly did zero research when writing that review and it made me seriously doubt how much research they do on products and services full stop. They should stick to Hoover's and washing machines.

 

[Peteski]

Which? are incredibly irritating when it comes to Tesla. Did anyone read their full review of the Model 3? They clearly did zero research when writing that review and it made me seriously doubt how much research they do on products and services full stop. They should stick to Hoover's and washing machines.

I would agree in general about Which being rubbish, but having read that FSD article I do happen to agree with it. It will be interesting what you think when you get to experience FSD first hand.

 

[MrBadger]

Have I missed something here? Please rumour me

Musk posted that a $1000 price rise for FSD was coming in August - probably to coincide with (yet another) advanced summon release that may be the start of ramping up FSD functionality.

I think feedback from that may influence some as to whether whatever FSD means is going to be something they want to buy into - me included.

 

[Peteski]

Musk posted that a $1000 price rise for FSD was coming in August - probably to coincide with (yet another) advanced summon release that may be the start of ramping up FSD functionality.

I think feedback from that may influence some as to whether whatever FSD means is going to be something they want to buy into - me included.

Have you read this thread below from the FSD forum?

Enhanced Summon, where are you?

The history of FSD development does not give me any confidence in it ramping up anything like quick enough to be of any practical use in our next Tesla. They've barely even completed EAP functionality (just moved a few of its features into the "new" FSD and created basic AP to differentiate it). At least AP is now free, so that is some progress at least! AP only needs auto-lane change (or a better workaround for manual lane changes) to put it right on par with current FSD. Unless you really do put serious value on gimmicks like Summon and auto park.

 

[MrBadger]

I’m in it for the long haul so Ok about getting on the rollercoaster as long as something comes of it - I’m not expecting L4 or L5 but a decent L3 covering most typical cases when statute allows will possibly do it for me. TBH, I would be surprised if Tesla and others were not at L4 with a fairly large control domain within 5 years. I wouldn't even hazard a guess at L5 and TBH, at present, I have no interest in that level of autonomy.

As I understand it, many of the niggles are caused by issues that HW3 should resolve with being able to run more cameras at higher spatial and temporal resolutions. That will probably filter down to AP improvements too but there are some FSD features I’m particularly interested in. That said, I believe that Tesla will need to add to their inputs, ie to allow V2I and V2V, otherwise they risk being left out in the cold. It would be nice to think that those with FSD will get these for free/nominal price, but I doubt it.

 

[MrBadger]

... AP only needs auto-lane change (or a better workaround for manual lane changes) to put it right on par with current FSD.

@Peteski is this similar to what you are talking about ? Just got 2019.24.4

I got 2019.24.4 last night and I noticed this morning that using the turn signal didn't disable Autosteer (I just have standard AP, no FSD). Has anyone else noticed this?

I don't use Autosteer often, but I'd rather not have to disable TACC in order to change lanes if I do have it on and I hadn't had this happen before.

and following discussion...​