UK FSD - What does it actually do on Model 3?

[Ant M3]

I have researched this so much, but honestly cannot find a clear answer, just an argument between beta and normal, paid and monthly, and whether its worth it or not. But nobody actually talks about what is does?

Can I just ask, if I pay the money to upgrade to FSD, what do I actually get that is different to standard autopilot in the UK?

Anything? I'm reading that it's basically the same.

Thank you

The answer is - not a great deal!
I’ve had it in mine for the last 3.5years and wouldn’t buy it again with hindsight . There are better ways to blow a few grand than in these features that have never worked.
Summon - doesn’t work
Auto park - worst system I’ve used
FSD - doesn’t work
Navigate on Autopilot - doesn’t work

I actively use the lane control and adaptive cruise control… they work will enough these days … but you get both of these with the basic autopilot functionality

 

[MrBadger]

I think the auto lane change function in EAP is crippled by the regulations over here in the UK. As I understand it regulations limit how long the car has to complete the manoeuvre, how dramatically it's allowed to turn and how hard it's allowed to decelerate/accelerate, and they also require human confirmation for any action.

Other vehicle manufacturers manage just fine. Its let down by Teslas poor implementation, not the regulations. They all play by the same rules.

 

[Mullermn]

Other vehicle manufacturers manage just fine. Its let down by Teslas poor implementation, not the regulations. They all play by the same rules.

I don't disagree. As I say, I think the problem is that the UK rules have been slapped on top of Tesla's driving logic as constraints but they have not properly re-engineered the way it works around them.

Basically I think (totally hypothetical, btw, I have no evidence) that you've got the primary logic in Autopilot going 'I want to change lane here', then you've got rules that say 'OK, well we need to flag that and get human confirmation', by which time the situation has changed and the lane change no longer makes sense.

Or, you've got Autopilot going 'I need to change lane, going to have to ac/decellerate fairly firmly to match speed with the destination without disrupting my current lane' and then the regulatory rules going 'nope', with the result that you're sat there with it telling you it knows it needs to change lane but never actually being able to do it.

And obviously as a background factor we've got the fact that the highway driving code is EOL pending replacement with the mythical single stack. I doubt it's gotten more than cursory development effort since they decided to change architecture and start again.

 

[MrBadger]

Basically I think (totally hypothetical, btw, I have no evidence) that you've got the primary logic in Autopilot going 'I want to change lane here', then you've got rules that say 'OK, well we need to flag that and get human confirmation', by which time the situation has changed and the lane change no longer makes sense.

It doesn’t work like that. The lane change prompt is nothing to do with the process. It’s all initiated by human intervention, be it the NoA confirmation or a manually instigated lane change. The issue is that when you trigger a lane change (or confirm a NoA trigger) the clock starts ticking. If the car does not detect you have applied enough torque to the steering, then it won’t start the lane change but the clock will still be running. If you don’t cross the white line in time (time dictated by regs but more than enough to achieve) then the car will violently swerve back from where you came. The issue is the car not reliably detecting that you are holding the wheel and the driver not responding to the nag (which for 99% of cases the driver is holding the wheel) quick enough.

Other reasons for other so called regs issues is the car driving full speed into a corner, suddenly, mid corner, realising that it’s not going to exceed the lateral limits and throwing a wobbly. Even a basic attempt at modulating speed would have prevented the issue but AP carries on regardless.

 

[Ferret]

FSD isn't worth it but I'm very pleased I paid for EAP, I often do hundreds of miles without taking over. Since the move to vision it has got a lot better too. I did 200 miles last Friday, the vast majority on EAP and it was perfect. Worth it just for the lane changes imho.

I suggest you trial it, as different people have different experiences.

 

[HaggisMan]

Similar experience with EAP here.

I think it depends on your typical journeys as to how much you will enjoy it. For us it’s plenty of long dual carriage ways regularly overtaking slower moving traffic doing say 60mph. EAP works well for us in that use case. No need for NoA or anything beyond “auto” lane changes really.

 

[Durzel]

I don't disagree. As I say, I think the problem is that the UK rules have been slapped on top of Tesla's driving logic as constraints but they have not properly re-engineered the way it works around them.

Basically I think (totally hypothetical, btw, I have no evidence) that you've got the primary logic in Autopilot going 'I want to change lane here', then you've got rules that say 'OK, well we need to flag that and get human confirmation', by which time the situation has changed and the lane change no longer makes sense.

Or, you've got Autopilot going 'I need to change lane, going to have to ac/decellerate fairly firmly to match speed with the destination without disrupting my current lane' and then the regulatory rules going 'nope', with the result that you're sat there with it telling you it knows it needs to change lane but never actually being able to do it.

And obviously as a background factor we've got the fact that the highway driving code is EOL pending replacement with the mythical single stack. I doubt it's gotten more than cursory development effort since they decided to change architecture and start again.

The big problem is that Tesla is uninterested in software development outside of their home country. If/when countries ease their auto driving regulations to enable FSD beta to be made available, that's when Tesla grants it, rather than Tesla actively developing software for those markets.

I can somewhat understand it in the sense that FSD beta is perhaps a homogenous blob that can't easily have restrictions grafted onto it, but Tesla already manage to provide a UNECE compliant Autopilot & other tools - e.g. limited autosteer, lane changes that have to be completed in X seconds, Summon that only works over short range Bluetooth, etc - so it's not beyond the wit of man.

The fact matrix lights still do nothing on European cars, despite the fact they have been legal in Europe for many years (only just recently legal in the States) speaks volumes. Again, zero interest in parallel software development.

FSD beta will be available in the UK (and Europe) at such time as UNECE regulations are relaxed enough that it can comply with them "out of the box", without any changes. I'm convinced at this point that Tesla's competitors will be ahead of it by that point in terms of self driving behaviour, Mercedes is already doing level 3 (not without its issues & limitations, but still).

The fact that I know that Tesla doesn't care about the UK (or Europe) beyond selling cars has certainly changed my attitude towards them in terms of future purchases, particularly functionality they sell where they make no mention at all of how crippled it is, or the (real - not Elon time) timelines for stuff being improved.

 

[WannabeOwner]

The fact that I know that Tesla doesn't care about the UK (or Europe) beyond selling cars

Hmmm ...

I don't argue that that the appearance, but I think its a stretch to assume they don't care. I think more likely its a question of when they will be able to care.

A "startup", albeit with piles of cash, makes it quite a challenge to create R&D outfits in order to satisfy multiple countries / markets and integrate that all into their mainstream. In the meantime the number of cars in the original California market is very substantial and I don't find it a surprise that that is the focus, for the time being.

 

[Sean.]

So in August 2017 I ordered our model X, with EAP optioned @ £4,700 - yikes now thinking back. But at the time, wide eyed excitement, expectation and promises took over better judgement... Although to be fair I did get a £5,600 ‘adjusted price discount’ on the final invoice before delivery. So you could argue in some twisted sense I got EAP for ‘free’. Well eerrrr no but anyway.

In the early days of ownership, EAP seemed pretty good, although quite ‘brave’ at times on A-roads. It certainly pushed on when I probably would’ve backed off a notch or two. Then came the serial stream of OTA updates and the infamous accidents in the US. Subsequent OTA updates kept gradually dumbing things down and the nagging notched up!

Then in June 2020 whilst in lockdown another strange fit of madness and unfounded expectation took hold! Tesla were offering a limited time discounted full-FSD option, via the app of £2,200 my brain and wallet was separated from reality again by the promise that it was ‘nearly there’…clearly they just needed the CASH badly at the time.

Well I do have a lovely VAT invoice though that clearly states “Full Self Driving Capability” AND to be fair I did get the HW2.5 to HW3.0 box of magical goodness upgrade a month later. But alas nearly 3 years later and EAP is stupider than ever and FSD…well I think I should rename it after myself for believing the hype twice over:

Flipping Silly Dork. Shame on me.