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

UK FSD Discussion

This site may earn commission on affiliate links.
Believe it if/when you see it and wonder what new is broken..sigh..

The single stack implementation of highways seems to be pretty good. Certainly more positive than negative, in fact I can’t actually recall anything significant poor in single stack highways, although that is only North America’s so far.

The aim is to roll that out to all vehicles, irrespective of which bits of functionality are enabled or not. So whether you have FSD or not, you will benefit from a more fully maintained AP implementation. It wouldn’t surprise me if more of that is now running beneath the hood than perhaps the UI and release notes might indicate. There seem to be a few more notable changes in driving behaviour being reported recently (2023.x releases) than (with a few notable exceptions) perhaps the previous few years.
 
  • Like
Reactions: M3noob
It wouldn’t surprise me if more of that is now running beneath the hood than perhaps the UI and release notes might indicate

I wouldn't want to be running a project where some-and-some was integrated on something this complicated (and incomplete in both Old and New code! - those two moving-targets would bring a host of new issues)

But that's pure speculation on my part, they may indeed be doing that, especially if the old code was designed to be piecemeal replaced by functions from the new code - but that approach always leaves developers one-hand-tied-behind-back on how innovative they can be with New Code) ... so personally I would stick it out for "replace".

I also wonder if they need to encourage FSD purchasing take-up - in which case they could take a commercial decision to leave AP festering for some time ... conversely they might want a shut-out bid with competitors in which case the commercial decision could be to leapfrog the inclusive-AP to something more capable than the competition; flies in the face of "Robotaxi" adding value to cars, and thus why the sticker price for FSD has been exceptionally high. Tesla wouldn't hang existing FSD purchasers out to dry ... would they? 🤩
 
  • Funny
Reactions: M3noob
That really only says they were running single stack, which necessitates running the beta branch right now. Smarter AP, basically, rather than any hope of city streets.

Definitely progress, if it's an answer to phantom braking, lorries leaping between lanes, that kind of thing.
 
I wouldn't want to be running a project where some-and-some was integrated on something this complicated (and incomplete in both Old and New code! - those two moving-targets would bring a host of new issues)

It already does share part of the stack already. NN’s etc. And we know recent developments for non USS park assist rely on the new occupancy networks and FSD beta previously did a seamless transition between old and new stacks and bits like sumon and auto park are old tech even for FSD beta. So already a precedent set for mix and match. A staged migration, as Tesla seem to have recently done with FSD beta highways) of functionality should pose no more significant risks than a big bang approach - less if anything - worked on both (large high profile systems that would have made the relevant news if there would have been the slightest hiccup) and all pretty uneventful if properly executed.
 
Tesla's EAP still has too many issues, even on 'simple' roads like motorways. FSD needs to be much, much improved. Yesterday we experienced our most aggressive phantom braking event on the M6. It hit the brakes really hard, scaring both me and my wife. Thankfully the guy behind was paying attention. He braked hard and luckily had an empty lane to swerve into. There was no car in front of us for several hundred feet. We were approaching a bridge and think maybe the car saw something in the shadow.

Phantom braking is much rarer for me now, but yesterday was bad enough that my wife has now banned AP use when she is in the car.
 
  • Like
Reactions: Pandinium
There’s a lot of hype on this side of the Atlantic about how good FSD Beta is in the US. I’ve never used it so I can’t comment. Fred Lambert, one of the journalists with Electrek who has FSD beta, maintains there are still way too many disengagements, and despite several software updates there has been no significant improvements in the last year. Others, though, think it’s fantastic.

I consider myself to be an eternal optimist, but when it comes to Tesla I’ve learnt to believe it when I see it. And with FSD I haven’t seen much so far.
 
  • Like
Reactions: ACarneiro
Really hope they bring back either a trial or add rental ability - I'm not willing to drop £3.5k or £7k on either feature set without some form of test.

Do look forward to the tech being available tho.
 
There’s a lot of hype on this side of the Atlantic about how good FSD Beta is in the US. I’ve never used it so I can’t comment. Fred Lambert, one of the journalists with Electrek who has FSD beta, maintains there are still way too many disengagements, and despite several software updates there has been no significant improvements in the last year. Others, though, think it’s fantastic.

I consider myself to be an eternal optimist, but when it comes to Tesla I’ve learnt to believe it when I see it. And with FSD I haven’t seen much so far.

The only FSD youtuber I tended to watch often was DirtyTesla although I feel like he's gone a bit too "pro" Tesla these days, plus the content has kind of got boring for me.

BUT. Watching the way his car handles itself vs the way mine does? Wow.....if I could even just get what he has, then I'd comfortably let my car drive me to work and back every day and that would be enough for me 😂
 
  • Like
Reactions: Buckminster
Tesla's EAP still has too many issues, even on 'simple' roads like motorways. FSD needs to be much, much improved. Yesterday we experienced our most aggressive phantom braking event on the M6.
Yes, there was a comedy moment on the A1(M) last week where it decided to take a random exit, signalled, swerved, and stamped on the brakes. Pretty sure the driver of the HGV behind me was wearing his breakfast cereal afterward.
 
There’s a lot of hype on this side of the Atlantic about how good FSD Beta is in the US. I’ve never used it so I can’t comment. Fred Lambert, one of the journalists with Electrek who has FSD beta, maintains there are still way too many disengagements, and despite several software updates there has been no significant improvements in the last year. Others, though, think it’s fantastic.

I consider myself to be an eternal optimist, but when it comes to Tesla I’ve learnt to believe it when I see it. And with FSD I haven’t seen much so far.
Hi from USA. Don't believe everything you read. It needs ALOT of work.
 
Could I just go for a 2-year bump and say [E]AP still can’t read road signs properly yet? A9 in Scotland last week 50mph🤦


FSD good luck.
 
Could I just go for a 2-year bump and say [E]AP still can’t read road signs properly yet? A9 in Scotland last week 50mph🤦


FSD good luck.
The originally linked tweet did suggest some improvement in that area

 
Temporary speed limits are a problem (time based or the 'when lights flash' ones)... also speed limits that only apply to certain classes of traffic and one with riders like 'half a mile ahead'.

FSD would need to read bus lane times as well..

It's not that an AI can't learn this stuff, but I just don't see any progress towards that.

It's not just the UK that gets wierd.. came across these US ones looking for examples.. I wonder if FSD gets these right..

main-qimg-9a9b69f969f4cdaa9697e5adebbef1a9-lq.jpeg
sign.png