Enhanced Summon & Autonomous driving UK laws

[GeorgeSymonds]

Wrong document. Try here:

https://unece.org/fileadmin/DAM/trans/main/wp29/wp29regs/2020/R079r4am2e.pdf

It explains the time limits and has some diagrams.

Thanks, much appreciated

Having read it though I don’t see the issue. Minimum 3s warning to other motorists (you could argue a little less would be ok but the point of indicating is to warn others after all) and then get on with it. I don’t see anything about aborting once the actual lane change commences so if that’s occurring it’s not the regulations, if it aborted before the lane change commences, maybe the indicator is on and times out, then I’d share a bit of frustration but again, surely nav on autopilot should know it has time to change lanes based on the closing speed of other vehicles.

 

[Pink Duck]

In terms of AP abort there's mention on page 85 of S owner's manual that a trigger condition is if the outside lane edge marking can't be recognised. However, it also has to adhere to the overall manoeuvre execution time limit per:

5.6.4.6.5. The lane change manoeuvre shall be completed in less than:
(a) 5 seconds for M1, N1 vehicle categories;
(b) 10 seconds for M2, M3, N2, N3 vehicle categories

(including the indicating time period)

 

[GeorgeSymonds]

Who’s preventing it being completed? It’s automated, it’s waited for the the 3-7 seconds or whatever Tesla coded, and then Tesla control the steering. I don’t get why this is a regulation issue, and 5s to change lane seems plenty of time.

I’m genuinely curious because if there is something I’d support the argument, I just don’t see it

 

[DrJFoster]

Could this be a "way you drive" thing? On motorways I tend to indicate when I intend to change lanes, not when I want to, ie I'm happy its safe, I've checked around me, and the indication is really no more than a courtesy to others and a "just in case I missed something", and stops the lane departure warning going off. When I use AP with nav on autopilot turned on I do the same, I indicate and while I wish it would change lanes straight away, I'm already happy all the conditions have been met for it to succeed. Occasionally it doesn't even try, spooked by something. Nav on AP doesn't always seem to offer the right opportunity so I largely don't bother now.

Some people you see drive differently, they indicate when they want to change lane and not quite when they intend to change lane, and then seem to sit there until the gap appears (sometimes its not long, they're anticipating the gap) or someone makes space and flashes them in. I can understand it failing in any of these latter situations.

The system timing out to complete a lane change though is surely a system issue or the driver asking when its not possible.

 

[MrBadger]

Who’s preventing it being completed? It’s automated, it’s waited for the the 3-7 seconds or whatever Tesla coded, and then Tesla control the steering. I don’t get why this is a regulation issue, and 5s to change lane seems plenty of time.

I’m genuinely curious because if there is something I’d support the argument, I just don’t see it

Its usually user error preventing it from being completed in time. I've had it happen to me a couple of times in 2 years and one occasion, it was pretty scary as driver originally behind me decided to floor it (much too soon) as I started to pull back left and fill the space I just came out of and the space I was about to lurch back into.

The user error is not holding the wheel correctly at point of instigating the lane change. When you indicate, the car checks that you are 'in control' which in other situations will trigger the nag, but not during lane change most of the time. So the clock starts running until it times out and cancels indicators - not a problem other than other road users probably thinking you are stupid for indicating for nothing.

The issue comes under certain timing conditions. If after you indicate, the car then detects that you are holding the wheel as required to satisfy being in control, but there was a delay before you satisfied the condition, then there is a time window when the car can start the manoeuvre but then run out of time mid manoeuvre causing the car to abort and swerve back into the lane that it just came from. There is a small window of opportunity for this to occur, detecting user input late into the window, so it not very frequent - as above, probably twice in 2 years - not the sort of event you tend to forget.

The reality is, like the disengagement when lateral forces is exceeded, whilst its the UNECE r79 regs limiting things, its Tesla's poor implementation that turns it into a big (possibly safety) event rather than an inconvenience/annoyance. To reduce user error, if the car does not recognise you are in control early into the manoeuvre, don't go putting torque against the wheel (or wiggle steering) too late into the lane change.

 

[gangzoom]

Do you have FSD or EAP? There really isn't an issue with lane change, other than it require more space than some people are comfortable with. I use it every day, and haven't had one abort for well over a year. I've also never had a cornering issue on a motorway, and don't use AP on other roads.

I have to say FSD lane changes are pretty good now. We did 300 miles of M-way driving this weekend and AP did 95% of lane changes. I cannot remember the last 'aborted' lane change.

Even phantom braking seems to be drastically reduced, just one episode in 5hrs of driving, and that was because of a van slowing down in the overtaking lane which caused the car to brake as it doesn't like undertaking.

 

[GeorgeSymonds]

The issue comes under certain timing conditions. If after you indicate, the car then detects that you are holding the wheel as required to satisfy being in control, but there was a delay before you satisfied the condition, then there is a time window when the car can start the manoeuvre but then run out of time mid manoeuvre causing the car to abort and swerve back into the lane that it just came from. There is a small window of opportunity for this to occur, detecting user input late into the window, so it not very frequent - as above, probably twice in 2 years - not the sort of event you tend to forget.

Looking at the regs, the "are you holding the wheel" check is in the first phase, and the actual lane change is in the second phase, each have individual time thresholds so a late confirm that you're holding the wheel shouldn't make any difference to the exeution of the lane change once started, unless its taken so long that the space is no longer available, but then the physical lane change shouldn't start if its not sure it can complete (accepting you might get a tool who decides to change into the lane next to you despite you indicating you're about to do the same.

Either way, I agree that it doesn't seem like a regulation issue to me, it's very much an implementation and/or driver issue.

 

[MrJaffro]

Interesting specifics on the regulation. The timings, and regs in general, have always seemed plausible but not ideal. I can see the human rational in them at least.

I just have a general disagreement with the approach of 'do exactly this', trying to prescribe an approach suitable for every scenario in the tiny detail doesn't seem realistically achievable. And I fear will be restrictive to progress overall, while also creating some safety situations.

 

[TestPilot]

I've never had it get 75% across the road and revert. I find the delay Tesla have put in annoying but that has nothing to do with the regulations that I can see, it's just a myth propogated by a few and latched onto by the many.

The regulation blame thing really feels like nothing more than a desire to blame someone other than Tesla.

I have and it's frankly terrifying for your passengers.

 

[Dilly]

I suspect that successive updates have quietly made some tweaks to improve overtaking. Last year it was very much hit and miss but my last long drive of 300 miles had a 100% success rate. Sadly, Tesla are like Apple. They make “improvements” but don’t mention them.

 

[DrJFoster]

I have and it's frankly terrifying for your passengers.

I can imagine, but I can't make the connection to the regulations, nowhere does it say "abort half way through" it just says "if you're going to do it, get on with it and do the lane change within x seconds once you start to change lane"

Other makes of car do lane change, I've not heard them complaining.

It does all seem to come full circle to the regulations are just a set of parameters, maximum cornering forces, min /max time to be indicating before changing lane and then how long you have to change lane - nothing there that is "knobbling the cars" - Tesla just don't seem interested in meeting the spec.

What would be the alternatives. Indicate to change lane and allow the car to sit there flashing away for 10 or 20 seconds? Nope.. not good. Or taking 15 seconds to change lane once it starts to move I think that would not be good either, it would look like a car drifting out of lane as much as changing lane.

 

[TestPilot]

I can imagine, but I can't make the connection to the regulations, nowhere does it say "abort half way through" it just says "if you're going to do it, get on with it and do the lane change within x seconds once you start to change lane"

Other makes of car do lane change, I've not heard them complaining.

It does all seem to come full circle to the regulations are just a set of parameters, maximum cornering forces, min /max time to be indicating before changing lane and then how long you have to change lane - nothing there that is "knobbling the cars" - Tesla just don't seem interested in meeting the spec.

What would be the alternatives. Indicate to change lane and allow the car to sit there flashing away for 10 or 20 seconds? Nope.. not good. Or taking 15 seconds to change lane once it starts to move I think that would not be good either, it would look like a car drifting out of lane as much as changing lane.

I can't remember what the exact rules are but it's something like 15 seconds total indication including manoeuvre and 5 seconds for manoeuvre itself right?

Imagine it being applied to a human in this scenario:

Driving in middle lane in busy but moving traffic with cars passing reasonably close to eachother in passing lane. You would like to move into passing lane to overtake a truck.
You begin to indicate to signal your intent, hoping some kind soul will let you out (as said, and as usual, there aren't gaps)
After 13 seconds someone slows down and flashes you to move across, you now have 2 seconds remaining to complete the lane change, you have a choice:
a) Cancel indicating, sit in your lane
b) Stop indicating and start indicating again to reset your timer, check it's still safe, move over
c) Try and sharply move over in your remaining 2 seconds

With FSD what you have to do is option a or option b, because if you try option c the car will take too long and abort the change. Without the limitation for both human and FSD your sensible option is
d) Safely complete the lane change, thank the driver who let you out.

 

[DrJFoster]

I can't remember what the exact rules are but it's something like 15 seconds total indication including manoeuvre and 5 seconds for manoeuvre itself right?

Imagine it being applied to a human in this scenario:

Driving in middle lane in busy but moving traffic with cars passing reasonably close to eachother in passing lane. You would like to move into passing lane to overtake a truck.
You begin to indicate to signal your intent, hoping some kind soul will let you out (as said, and as usual, there aren't gaps)
After 13 seconds someone slows down and flashes you to move across, you now have 2 seconds remaining to complete the lane change, you have a choice:
a) Cancel indicating, sit in your lane
b) Stop indicating and start indicating again to reset your timer, check it's still safe, move over
c) Try and sharply move over in your remaining 2 seconds

With FSD what you have to do is option a or option b, because if you try option c the car will take too long and abort the change. Without the limitation for both human and FSD your sensible option is
d) Safely complete the lane change, thank the driver who let you out.

That's the exact scenario I described a few posts earlier as a potential explanation but not one I'd really support. Are we suggesting when using Navigation on Autopilot that as soon as it reaches a car in front driving slower than it wants it should put its indicator on and wait until a space appears? Either way, the rules can hardly be construed as completely hindering the system because of what they are

 

[TestPilot]

That's the exact scenario I described a few posts earlier as a potential explanation but not one I'd really support. Are we suggesting when using Navigation on Autopilot that as soon as it reaches a car in front driving slower than it wants it should put its indicator on and wait until a space appears? Either way, the rules can hardly be construed as completely hindering the system because of what they are

This is why FSD is hard right, because what *should* it do? Quite commonly on UK roads there won't be a big enough gap for many minutes. Should the car just sit there doing 50mph for 15 minutes while cars slowly pass by?

 

[MrJaffro]

This is why FSD is hard right, because what *should* it do? Quite commonly on UK roads there won't be a big enough gap for many minutes. Should the car just sit there doing 50mph for 15 minutes while cars slowly pass by?

This is it right. Prescribing exact behaviours doesn't fit all scenarios and, in my opinion, is why our driving tests are based around proof of control of the vehicle and understanding of rules and dangers. If everything we're predicable and followed the exact uniform behaviour then it might work out. But humans and the world in general is weird and judgment is always needed.