bjorn nyland's test of tesla vision

[JupiterMan]

There's been lots of discussion at the UNECE level as well.. but it takes years for discussion to reach legislation. I think they're going about it the right way though.. allow full autonomy in limited circumstances then slowly relax that as the technology gets better.

I disagree, UNECE rules make autopilot less safe in those markets and also slows down innovation.

When things are left to multinational committees these things invariably take longer, and you cannot help but be suspicious that there are various national interests within UNECE whose purpose is to slow down innovation so as to assist their domestic auto industries.

 

[DrJFoster]

I disagree, UNECE rules make autopilot less safe in those markets and also slows down innovation.

When things are left to multinational committees these things invariably take longer, and you cannot help but be suspicious that there are various national interests within UNECE whose purpose is to slow down innovation so as to assist their domestic auto industries.

I think it can work both ways. I know some argue that the UNECE rules are hindering Tesla in Europe but when you sit down and read the regulations they make perfectly sensible reading.

One person was objecting about automatic lane change on motorways not allowing him to just put his indicator on and when the car can move out, even if that's 30 seconds, doing so. Since when was that the correct way to drive in the UK? You indicate as the last thing you do before changing lanes. The regulations enforce this. That's good.

 

[Durzel]

And of course the law/UNECE and the like deciding the what and how of all this is one thing, Tesla being able to meet whatever they come up with is another. Imagine if they say something as simple as "for law enforcement and wider driver awareness, any car operating under level 4 conditions needs to have a steady green light visible from all angles" - Tesla won't be able to comply without changes, at least not with the existing fleet. On a similar note, the number of people who thought we'd get adaptive headlights when the US approved their use was quite staggering, Tesla simply haven't got the capability to actually do it at the moment.

I'm sure lots of lobbying is going on over the capabilities and rules, what's certain is being on the inside and working with regulators and playing the game is going to be a politically smarter place to be than being on the outside trying to goad them into something.

I was genuinely surprised (again, I need to do something about that) when Tesla didn't have anything ready to go for when the US made adaptive headlights legal.

Even if you disregard the fact that they could have deployed this tech in Europe as soon as the cars came with the hardware, but didn't - again, because they don't care about Europe, they could have been developing this on test cars in the States for however long they needed to be ready to launch it when the regulations were changed.

As you say it tends to suggest that there isn't the resource to do it, or the capability, that it is making no difference to their sales volume so they don't feel the need to, or - worse - Tesla (aka Musk) isn't particularly interested in it because other manufacturers like BMW, Audi, Mercedes, etc have already got X generation tech in place, so there isn't anything for Tesla to improve upon - only emulate - which isn't really Musk's bag. Musk's shadow looms large over development in the cars - case in point the complete absence of Apple CarPlay/Android Auto.

It also makes me think - as a cynical man - that the reason the only functional use of matrix headlights in any market is to project "TESLA" during a gimmick light show is exactly because it is something the other manufacturers haven't done (like the lightshow itself). It's all showmanship but with little substance behind it, where it counts, where it delivers value to actual customers rather than observers.

As it is I'm reasonably convinced the matrix headlights will see hardware iterations & upgrades before anyone realises any benefit from them. As it is they came on cars in late 2020 I think, maybe early 2021, and nothing has happened with them since then. Could it be another year before Tesla can be bothered with it? If sales carry on as they are, why would they care? People who own cars with the hardware could be looking at buying a replacement car before they get any benefit from it, much like me with FSD.

 

[chwe]

I think it can work both ways. I know some argue that the UNECE rules are hindering Tesla in Europe but when you sit down and read the regulations they make perfectly sensible reading.

One person was objecting about automatic lane change on motorways not allowing him to just put his indicator on and when the car can move out, even if that's 30 seconds, doing so. Since when was that the correct way to drive in the UK? You indicate as the last thing you do before changing lanes. The regulations enforce this. That's good.

Not sure I agree with this.

Highway code says "Signals warn and inform other road users, including pedestrians (download ‘Signals to other road users’), of your intended actions" and that you should " give clear signals in plenty of time, having checked it is not misleading to signal at that time". If it is your intention to change lanes then signalling your intention and then moving across when it is safe to do so would appear fine regardless of whether you wait 1 second or 30 seconds to do this.

 

[Jason71]

I was genuinely surprised (again, I need to do something about that) when Tesla didn't have anything ready to go for when the US made adaptive headlights legal.

Even if you disregard the fact that they could have deployed this tech in Europe as soon as the cars came with the hardware, but didn't - again, because they don't care about Europe, they could have been developing this on test cars in the States for however long they needed to be ready to launch it when the regulations were changed.

As you say it tends to suggest that there isn't the resource to do it, or the capability, that it is making no difference to their sales volume so they don't feel the need to, or - worse - Tesla (aka Musk) isn't particularly interested in it because other manufacturers like BMW, Audi, Mercedes, etc have already got X generation tech in place, so there isn't anything for Tesla to improve upon - only emulate - which isn't really Musk's bag. Musk's shadow looms large over development in the cars - case in point the complete absence of Apple CarPlay/Android Auto.

It also makes me think - as a cynical man - that the reason the only functional use of matrix headlights in any market is to project "TESLA" during a gimmick light show is exactly because it is something the other manufacturers haven't done (like the lightshow itself). It's all showmanship but with little substance behind it, where it counts, where it delivers value to actual customers rather than observers.

As it is I'm reasonably convinced the matrix headlights will see hardware iterations & upgrades before anyone realises any benefit from them. As it is they came on cars in late 2020 I think, maybe early 2021, and nothing has happened with them since then. Could it be another year before Tesla can be bothered with it? If sales carry on as they are, why would they care? People who own cars with the hardware could be looking at buying a replacement car before they get any benefit from it, much like me with FSD.

Do the current generation of cars have the necessary sensors to control the matrix effectively. I assume the cameras are the only option and they struggle with basic headlight dipping and windscreen wiper control ( particularly at night) so I am not sure they have capability to do much with the matrix headlights.

 

[Zilla91]

If I saw someone signalling for 30 seconds I’d assume they’d forgot to deactivate their signal

 

[Durzel]

Do the current generation of cars have the necessary sensors to control the matrix effectively. I assume the cameras are the only option and they struggle with basic headlight dipping and windscreen wiper control ( particularly at night) so I am not sure they have capability to do much with the matrix headlights.

Good question. Who knows, is probably the answer. I'm casually aware that the German marques tech, particularly Audis that I have experience of, is very advanced by this point, and is several generations old.

Having addressable individual LEDs is - I suspect - a very small part of the puzzle and I could easily believe that the logic and intelligence behind it is in its own way not trivial. Tesla would either have to build upon some open source project, tempt away engineers who have worked on those systems at the aforementioned companies, and/or develop their own systems from scratch. I'm inclined to believe Tesla/Musk is arrogant enough to believe they could do it all better themselves, much like the cameras/AI detecting rain thought process.

All of that is to say that if Tesla didn't have something ready to go for when it became legal in the States, it would tend to suggest they haven't been working on it in any meaningful way. As said, if sales are unaffected then they aren't exactly inclined to either, although they could easily be caught on the hop if more and more competitors end up with adaptive headlights, or their customer base starts to revolt.

 

[chwe]

If I saw someone signalling for 30 seconds I’d assume they’d forgot to deactivate their signal

This rather depends though on whether they would be able to safely manoeuvre. If you signal for 30 seconds but can't switch lanes because there is not a space to move into this is different from signalling for 30 seconds because you have forgotten to turn your signal off. My point was that the Highway Code says signals should be used to indicate intended actions, so if someone using autolane change wants to move lanes then I don't see that the length of time the indicator is on contravenes 'good driving practice' in the UK. Of course there is a fundamental difference in terms of whether Tesla's system implements this in a safe manner.

Edit: And for the record I don't personally stick my own indicators on and wait for a space, I do turn my indicators on shortly before actually intending to move; I'm just saying this isn't necessarily 'better' according to my interpretation of the Highway Code.

 

[randompixel]

Some more options to throw in the pot.

1. Musk stated the engineers worked on what they felt like, much like Valve (
   https://twitter.com/x/status/1413894389299023874
   ). What's to say they haven't decided that building an App Store or Steam Support is more important to them to work on. Are matrix lights that tricky thing nobody wants to touch (and get wrong!)?
2. They're finding it really hard to work around other peoples patents and they don't want to pay to use them, nor have enough of their own patents to "trade". The MobilEye set of patents looks quite comprehensive for all sorts of vision based things.

 

[MrBadger]

I disagree, Tesla's implementation of the UNECE rules make autopilot less safe in those markets.

Let me fix that for you...

Nothing wrong with UNECE regs if properly implemented and actually increase safety.

Full disclosure - this is just IMHO but I use to work for a transport research establishment that specialises in road safety and CAV (running level 4 vehicles in public areas for over 5 years for research purposes) and still keep half an eye on that side of the industry.

 

[WllXM]

Slightly deviating but would someone on here ever use a counterweight to stop the AP hands-on-wheel alerts (despite the horrendous insurance implications of course)? Not an incentive to nap behind the wheel of course, just to allow the hands to be positioned more naturally around it without having to constantly apply more force than necessary?

 

[MrBadger]

Illegal (you are not in proper control of the vehicle), highly dangerous and apparently no longer works like it use to.

 

[GeorgeSymonds]

If I saw someone signalling for 30 seconds I’d assume they’d forgot to deactivate their signal

I'd think that too.

You indicate to say what you intend to do, not what you wish to do. Roadcraft also says 1 flash for every 10mph before the manouver, so at 70mph that would be 7 flashes, I believe they have to be betweem 60 and 120 a minute, so we're probably talking about 5 seconds.

I was also taught Mirrors – Signal – Manoeuvre - you check its safe to perform the manoeuvre, you signal to say you're doing it, you do it.

 

[spooksman]

Slightly deviating but would someone on here ever use a counterweight to stop the AP hands-on-wheel alerts (despite the horrendous insurance implications of course)?

Not me for sure - seems a dangerous (and probably illegal) idea. I'm not sure what the issue is with keeping your hands on the wheel - there is no 'set' position where they have to be.

 

[WllXM]

it is illegal for sure in the event of an accident.
I’ve found that on long motorway journeys my grip is too soft and the chime starts to sound, up to the point where I actually press TOO hard on the wheel and autosteer disengages.

I’ve also been punished several times where I had kept my hands on the wheel, too softly for Tesla’s taste, and was then punished from using it for the remainder of the drive. Annoying.

Having to tap every 20 secs is not ideal either.

 

[GeorgeSymonds]

it is illegal for sure in the event of an accident.
I’ve found that on long motorway journeys my grip is too soft and the chime starts to sound, up to the point where I actually press TOO hard on the wheel and autosteer disengages.

I’ve also been punished several times where I had kept my hands on the wheel, too softly for Tesla’s taste, and was then punished from using it for the remainder of the drive. Annoying.

Having to tap every 20 secs is not ideal either.

It’s not hard or soft, it’s twist. One finger gently pushing the wheel clockwise (or anti-clockwise) is enough. I just hold the wheel and let the weight of my forearm, elbow on the door, apply the rotational force and that’s enough

 

[WllXM]

It’s not hard or soft, it’s twist. One finger gently pushing the wheel clockwise (or anti-clockwise) is enough. I just hold the wheel and let the weight of my forearm, elbow on the door, apply the rotational force and that’s enough

I must have a really bad case of coordination. Either I rest my arms too heavily on the wheel and ap disengages and I get out of my lane, or if I rest them too softly, elbow on the door as well, it doesn’t register.
The sensitivity must be off for me, and constantly focusing on that is sometimes worse than if I were to steer the car myself

 

[goRt]

I'd think that too.

You indicate to say what you intend to do, not what you wish to do. Roadcraft also says 1 flash for every 10mph before the manouver, so at 70mph that would be 7 flashes, I believe they have to be betweem 60 and 120 a minute, so we're probably talking about 5 seconds.

I was also taught Mirrors – Signal – Manoeuvre - you check its safe to perform the manoeuvre, you signal to say you're doing it, you do it.

Roadcraft teaches position - speed - gear

 

[KennethS]

>>Until Tesla do something in this area they’ll never be given level 3.<<

I don't think the US regulatory system is as independent as it perhaps is elsewhere and politics comes into it to a much greater extent.
Might is right: Musk is worth on paper billions. Hubris reigns.

He’s not exactly in favour with the current administration….

 

[Durzel]

As has been said above it’s torque the car is looking for, not pressure. It won’t help to squeeze the wheel, it has to feel a slight amount of steering torque from you.

To be honest though, you get a long warning that you need to do it, with the pulsing blue screen. There is no excuse really for letting it get as far as penalising you for ignoring it.