That’s a good point. Maybe it works much better in regions where this doesn’t apply, but as this is the UK sub forum it is what it is.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Auto lane change on FSD
- Thread starter tabpharmer
- Start date
Whilst we are on this topic has anyone noticed this quirk: When I activate autopilot one NOA road it will want me to change lane immediately. Usually to the right. Before then realising the road is empty and ask you to change back to the correct lane? I get a "lane change to follow route" nag. Strange one. To be honest I complain about a lot of the software on the Tesla but apart from having to nudge the wheel to confirm (which is a regulatory thing I understand) I do like NOA lane changes. it's reasonably consist and only struggles in certain types of traffic. Also I've never had a problem with automatically exiting motorways apart form the previously mentioned fact "Mirror, Signal, Manoeuvre " happens in one startling and dramatic instant. I would prefer the car to signal intent BEFORE spearing you onto a slip road.
MrBadger
Badger out
Seems to be widespread behaviour.
The nag is not a regulatory thing, but as I understand it, the car having to check driver is in control is. The nag is a consequence of Tesla's implementation when it doesn't detect you in control of the car, ie in Tesla way of doing things, your hands holding the wheel properly.
Louis0w
Member
Thank you, and yes - agreed, it is what it is here in the UK, until it hopefully changes. Just think t's important for others to understand that, technologically, our cars are more capable than our own daily experiences would suggest. Anyone using AP with auto-lane change before mid 2019 can confirm that. Mid 2019 is when Tesla changed/limited the AP software to conform to UNECE R79 for the European market.
Louis0w
Member
My understanding is the UNECE rules require the system to "provide a means of detecting that the driver is holding the steering control". Without pressure sensors in the wheel or a wheel facing camera, the lateral movement provided by a light 'tug' is sufficient means for Tesla to confirm the driver is holding the wheel. UNECE R79 also prescribes the nag time intervals and what steps the car should take to warn the driver they need to hold the steering wheel.
Yeah I get that. My point is I hold the wheel with a bit of resistance to stop the usual nag. When you lane change it always asks you to apply a slight turn to confirm. Even if you are already holding the wheel. This can't be necessary. Indeed it means auto lane change, well, isn't
MrBadger
Badger out
I don't always get the nag when changing lanes. I suspect its part down to how I hold the wheel and providing inconsistent torque over a period of time - under in lane driving conditions, it only seems to have to detect this torque ever 15 seconds or so, so detecting the necessary torque any time within a 15 second period is fine to put off the nag. However, with an auto lane change, the detection of (in Tesla's case) required torque, has to occur at a specific point in time, so if I am not satisfying the requirement at that point in time, its either not going to start the manoeuvre until it does, or give up on the manoeuvre after a period of time. As long as I am satisfying the torque requirement at that point in time, I get a prompt lane change with no further intervention (other than the indicator).
Mark-R
Member
It is only asking you to apply a slight turn for the lane change becuase it thinks you aren't holding the wheel. If you apply gentle but constant torque (see the video below) you don't get the nag when it wants to initiate a lane change - you just need to confirm it with the indicator.
Cardo
Active Member
I was testing this on a 50 mile motorway trip, the other day. I don’t think you need to apply pressure to initiate the lane change, it just needs to have detected pressure within the last x seconds to be happy you are still ”paying attention”. I tested this by applying pressure before trying a lane change, and as long as I had shown I was holding the wheel moments before, if I then hit the indicator the car would change lane without requiring a further nudge.
That's not what I experience at all. Is my system broken? I hold the wheel to stop the normal autopilot nag, always with a bit of clockwise torque. When I initiate a lane change (or the car spots one) I'm further asked to apply a slight turning force to the wheel. This is above what I am already applying to stop the autopilot nag. Is this not normal behaviour?
Cardo
Active Member
I’m wondering whether there is a higher threshold on the torque required between lane keeping and lane changing.
Mark-R
Member
It sounds like you might not be applying quite enough torque, or perhaps the torque you are applying is a bit too variable? It took me quite a bit of trial and error to get it right, but now I can let the car drive in AP without the nag and it is very rare for the car to abort an automated lane change.
Louis0w
Member
This is my experience too, Auto lane change will initiate 'hands off' (though is not recommended) so long as the car confirms you are being attentive. Attention is confirmed by the application of lateral movement to the steering wheel within the time limit specified in the regulations; I believe this is within 15 secs of last attentive confirmation.
No that can't be it. I hold my wheel exactly like the guy in the video and I never get the autopilot nag. Therefore I assume it must know I'm touching the wheel. As soon as it wants to change lane or I initiate a lane change it asks for more turning force to be applied.
Probably down to the tolerance of your steering wheel torque sensor. Some require a bit more pressure than others to activate.
I have had EAP since Feb 18 on my MX and I don't remember it being any better before the change. It has always been dog slow to change lanes and prone to aborting or taking several attempts to initiate (and not from user error). I do still use ALC most of the time regardless of the issues, just to remain in AP, but it's really not a good system in the UK. I would never pay for it again. I have the 3 month FSD trial with my new M3 and that is frankly just as rubbish too even with the latest hardware. Along with the irritating steering wheel nags (when you hold the wheel naturally two-handed without deliberately forcing any torque) the whole thing is a joke. Might as well just use TACC and steer it yourself - as your hands are on the wheel anyway.
I wish I could decrease the required steering torque. If it required just the minimal torque to detect that the driver holds the wheel, that would be a great improvement. As it is now, it is too much.
Of course, other methods to detect driver attention could also be an improvement, but I think, torque would be quite OK, if it wouldn't require as much torque as it does now.
Hack the sensor, someone? As long as it reliably detects the driver's hand, why not?
Of course, other methods to detect driver attention could also be an improvement, but I think, torque would be quite OK, if it wouldn't require as much torque as it does now.
Hack the sensor, someone? As long as it reliably detects the driver's hand, why not?
Similar threads
