Tesla Model 3 and speed limit signs

[Dilly]

Sometimes you take a chance in life and you reap the award.

My SR+ is a year old now, has no fit/finish snags and doesn't rattle at all. We did 1,700 miles round the North coast of Scotland and back in October just before Nicola closed it down and enjoyed driving evey mile. Very steep hills are a wonder (you don't notice them) and the high regen coming down the other side reduced braking to zero and brought energy consumption to not far off driving on a flat road.

You will always read about complaints as often said here and its always possible to get a minger but the RH drive cars now coming from China should be far better built.

If you book a test drive, just be ready with the money as driving any of these cars is nothing like an ICE and to me is pure fun!

I agree with every word. It’s just like a good book. You can’t leave it alone

 

[Dasher]

TBH it is mainly the fit/finish/quality and reliability issues that are making me hesitate about spending £45k on a Tesla.

1. Bottom of Which! reliability survey ("Tesla must be doing something right to inspire such passion in its customers, but that something certainly isn't making reliable cars")
2. Very low in JD Powers "JD Power scored Tesla vehicles the worst among 32 major brands in its annual quality study")
3. Very low in US Consumers Reports ("Tesla’s overall poor performance in the annual study placed it second to last among the 26 ranked brands").

Now these are consumer lead feedback/reports so I don't think they can be dismissed as bias, which some Tesla owners seem to wish to do unless you are into conspiracy theories.

I have yet to be convinced that these surveys really reflect Tesla fairly. Having been reading thousands of forum posts since I reserved my M3 in March 2017, I get the strong impression that Tesla has, and may continue to have a poor quality record on a minority of cars on delivery, but there are relatively few cases of a Tesla actually giving real grief after resolving any initial problems.

Given that there are several hundred thousand around, I think that we get an uneven perception on the forums (fora?). It is always going to be unfortunate and a pain when your car goes wrong, and we know that getting a Tesla back on the road can drag out. However, I don't get the disastrous impression on overall reliability and quality that these surveys imply.

 

[AndrewGR]

But, to get back to the OP. Mobileye equipped cars read speed limits reliably and have done so for quite a few years. This includes gantry signs on motorways and ‘it depends whether it’s raining or you are towing a trailer’ ones in France. It is clearly an essential feature for FSD to work. When are we going to get it? It’s a feature I really miss from my last couple of cars - especially when I’m driving on the M25!

 

[Durzel]

We’re not going to get Mobileye on Teslas because it’s patented and Elon doesn’t like that, so they’ve had to come up with their own solution, costing however many man hours, using inefficient or ineffective alternative hardware instead.

See also rain sensors for auto wipers and ambient light sensors + camera logic for auto headlights. Two well and truly solved problems in vehicle dynamics, but it involves paying for sensors from third parties and that’s anathema to Tesla for some reason. Better to implement some unreliable AI based system instead.

 

[AndrewGR]

But the AI system has to work fully in order to provide FSD. Compared with other challenges in FSD knowing the speed limit is trivial. So I’m sure we will have it ‘by the end of the year’, or am I being a bit too optimistic? [Sceptic mode - OFF] ! Real question is, ‘which year, and in which car’?

 

[Neilio]

We’re not going to get Mobileye on Teslas because it’s patented and Elon doesn’t like that, so they’ve had to come up with their own solution, costing however many man hours, using inefficient or ineffective alternative hardware instead.

See also rain sensors for auto wipers and ambient light sensors + camera logic for auto headlights. Two well and truly solved problems in vehicle dynamics, but it involves paying for sensors from third parties and that’s anathema to Tesla for some reason. Better to implement some unreliable AI based system instead.

This is a worry for me about Tesla. I like what they are doing and their battery technology but. There's something not right about is having to put up with things that don't work because the owner doesn't like something. Any other car company would, and indeed have, implement the most commercially viable solution rather than some fancy solution that doesn't work.

 

[Gatsojon]

You have to do things differently if you want to promote innovation. Sometimes that works out later than we’d like and in this case, we are still on the journey.

That doesn’t stop be thinking how much better my BMW read road signs 5 years ago than my Model 3 does now. But then I remember how it didn’t even try when I bought it and now it’s learning. Keep the faith

 

[Neilio]

You have to do things differently if you want to promote innovation. Sometimes that works out later than we’d like and in this case, we are still on the journey.

That doesn’t stop be thinking how much better my BMW read road signs 5 years ago than my Model 3 does now. But then I remember how it didn’t even try when I bought it and now it’s learning. Keep the faith

I sort of agree with that. My issue is with things like traffic sign recognition and auto wipers we have systems that just work. No need for innovation: rain isn’t going to change and it’s unlikely speed signs will change enough to require a new system. I’m not against innovation but when I buy a tool I use every day I expect things to work that work on every other make of the same tool

 

[Mm2021]

H

But the AI system has to work fully in order to provide FSD. Compared with other challenges in FSD knowing the speed limit is trivial. So I’m sure we will have it ‘by the end of the year’, or am I being a bit too optimistic? [Sceptic mode - OFF] ! Real question is, ‘which year, and in which car’?

Haha Musk often promises things “by the end of year”. Such a failed metric. Should always be translated to “by the end of 2 years”. Still his giddy optimism is what keeps me excited about the future so I’ll take it.

 

[MrBadger]

Haha Musk often promises things “by the end of year”. Such a failed metric. Should always be translated to “by the end of 2 years”.

Or in the case of early FSD purchasers, "by the end of 5 years and counting and it won't be what we initially promised"

 

[Durzel]

You have to do things differently if you want to promote innovation. Sometimes that works out later than we’d like and in this case, we are still on the journey.

That doesn’t stop be thinking how much better my BMW read road signs 5 years ago than my Model 3 does now. But then I remember how it didn’t even try when I bought it and now it’s learning. Keep the faith

I don’t see the point of innovation for innovation’s sake. Things like auto wipers using rain sensors is a completely solved domain. It works effectively and reliably, and I can’t conceive of a way that reinventing it can make it better. It just works. Also, since it’s been refined to the nth degree, one assumes the cost of the parts is competitive.

How many CPU cycles are burned calculating rain droplets on the windscreen using AI? If the end result was a functionally equivalent system I’d still think it was a waste of resources (man hours and CPU cycles), but it is demonstrably worse than rain sensor based systems on cars several years older.

And auto headlights? Not fit for purpose. It beggars belief that a car known for having an abundance of cameras, compared to almost all other cars, completely fails to see ambient lighting changes, rear brake lights, etc. My last car - an Audi TTRS - was 7 years old at this point and it worked on that damn near flawlessly, again using cheap (relatively) tried and tested sensors.

 

[justinhow]

I don’t see the point of innovation for innovation’s sake. Things like auto wipers using rain sensors is a completely solved domain. It works effectively and reliably, and I can’t conceive of a way that reinventing it can make it better. It just works. Also, since it’s been refined to the nth degree, one assumes the cost of the parts is competitive.....

I totally agree - you must at time be pragmatic. I think too often Elon Musk is too fixed in his views with an unwillingness to change. He likes to do things differently, and that can lead to great innovation, however when those views are not accepted by many potential customers there is a problem. The result will be less market share (perhaps significantly) than he would otherwise get. Some of these seem to be driven (sorry!) by his autonomous driving vision, which we all know is still years away before reaching level 4/5 and in which his timeline has been woefully wrong. A few (contentious) design points:

1. The lack of any head up display option (which you can always turn off if you don't like)
2. The refusal to have a small screen directly ahead of the driver.
3. The refusal to have almost any buttons even for commonly used functions like adjusting air con or switching audio source.
   
4. Re-designing traffic sign recognition (re-creating the wheel??).
5. Adjustable damper/suspension settings as an option (Model 3 and Y).

Now I realise not all will agree that these are wrong (so lets not go down that rabbit hole) but equally a lot will think they are wrong.
Because Tesla has been so far ahead of the curve for EV's they have had huge success, but lets be clear, the competition is stepping up by the month now. The Hyundai Ioniq 5 (looks impressive), VW id range, BMW i4 soon etc. I am a potential Tesla buyer and think the choice is getting much closer as to which car to buy and some of strange design choices may just end up being a tipping point.

 

[Gatsojon]

I don’t see the point of innovation for innovation’s sake. Things like auto wipers using rain sensors is a completely solved domain. It works effectively and reliably, and I can’t conceive of a way that reinventing it can make it better. It just works. Also, since it’s been refined to the nth degree, one assumes the cost of the parts is competitive.

How many CPU cycles are burned calculating rain droplets on the windscreen using AI? If the end result was a functionally equivalent system I’d still think it was a waste of resources (man hours and CPU cycles), but it is demonstrably worse than rain sensor based systems on cars several years older.

And auto headlights? Not fit for purpose. It beggars belief that a car known for having an abundance of cameras, compared to almost all other cars, completely fails to see ambient lighting changes, rear brake lights, etc. My last car - an Audi TTRS - was 7 years old at this point and it worked on that damn near flawlessly, again using cheap (relatively) tried and tested sensors.

It's embedded in the company philosophy. You can't turn that on and off at every decision point. I totally agree that better systems currently exist to solve some problems. Those car manufacturers that are using them may also produce a decent electric car one day too perhaps. I hope so, then we will have more choice. Currently, for me, there is no viable alternate to the Tesla Model 3.

 

[GeorgeSymonds]

The auto headlights and auto wipers are all meant to be part of the "we know whats going on" inside the FSD computer. They claimed at one point that the car would slow down on AP if it was raining heavily recognising the road was more dangerous for example, maybe be more cautious through a curve if it was icy or very wet, auto headlights, how can you self drive if you don't what other cars are about.

Unfortunately all its really shown is that they're a long way off knowing such things with a high degree of accuracy. The "Deep Rain" windscreen wiper thing may have improved things a few years ago, but is still not on a par with a 20p sensor. What I find disappointing is that if they had fitted it, it could feed into the FSD model, in the same way hopefully temperature sensors do, and that sensor could be used to determine glass misting/dew points and other atmospheric issues and so on.. things which you could argue would be more beneficial

 

[GRiLLA]

I totally agree - you must at time be pragmatic. I think too often Elon Musk is too fixed in his views with an unwillingness to change. He likes to do things differently, and that can lead to great innovation, however when those views are not accepted by many potential customers there is a problem. The result will be less market share (perhaps significantly) than he would otherwise get. Some of these seem to be driven (sorry!) by his autonomous driving vision, which we all know is still years away before reaching level 4/5 and in which his timeline has been woefully wrong. A few (contentious) design points:

1. The lack of any head up display option (which you can always turn off if you don't like)
2. The refusal to have a small screen directly ahead of the driver.
3. The refusal to have almost any buttons even for commonly used functions like adjusting air con or switching audio source.
   
4. Re-designing traffic sign recognition (re-creating the wheel??).
5. Adjustable damper/suspension settings as an option (Model 3 and Y).

Now I realise not all will agree that these are wrong (so lets not go down that rabbit hole) but equally a lot will think they are wrong.
Because Tesla has been so far ahead of the curve for EV's they have had huge success, but lets be clear, the competition is stepping up by the month now. The Hyundai Ioniq 5 (looks impressive), VW id range, BMW i4 soon etc. I am a potential Tesla buyer and think the choice is getting much closer as to which car to buy and some of strange design choices may just end up being a tipping point.

Every car gets built to a vision, it's a bit weird to call anything that doesn't match your personal preferences a 'refusal'. A bit like saying Leonardo refused to paint hats as the Mona Lisa doesn't have one. The vision of the Model 3 is well executed, and owners love it.

 

[Glan gluaisne]

It's very easy to get sucked in to the "must do everything differently" mindset, though, and very hard to get out of it and start to look at things pragmatically. I know this from experience. I designed this house, I started out wanting to do everything differently, as I was absolutely convinced that the reason for UK housing stock being so poor, in terms of energy efficiency, was because of the "we've always done things this way" mindset within the building trade.

In the main this worked, we have a house with no foundations, loads of insulation, minimal use of concrete and that generates about twice as much energy per year as it uses, so we effectively have no bills. At the same time, I got carried away with wanting to measure anything and everything that could be measured, so the house has around 20 microcontrollers scattered around, with maybe 60 or 70 sensors, most of them networked to a data acquisition system. I did away with a lot of things that could have been purchased off-the-shelf, like the heating/cooling control system, car charge points, controller for our sewage treatment plant, etc, and built custom solutions, with all the code written and tested by myself.

Move forward a few years, and the pitfalls in taking this approach start to manifest themselves. Instead of off-the-shelf solutions that can be replaced easily and quickly, I have a stack of custom stuff that needs me to dig out code I wrote years ago and try and make sense of it in order to change something or replace it. I've gradually been going around replacing most of the custom stuff I put together with proven off-the-shelf solutions, mainly for reliability and repairability reasons. I can pop into Screwfix and buy a replacement light switch and fit it within a couple of hours, and it will work just fine. For me to repair/replace one of my home brew control systems might take a week or more, and may well not work as well as a simple off-the-shelf solution.

With hindsight, the one thing I wish I'd done differently was opt to use the best off-the-shelf, well-proven, solution for each task. I let my desire to do things differently, just because I could, override common sense. Tesla are doing the same in some areas. Instead of spending effort on doing the things that matter, like getting FSD to work reliably and improving the build quality and reliability of their cars, they are burning up resource re-inventing solutions for simple problems that have already been solved. I'm willing to bet that Tesla have expended far more money on trying to get the cameras to trigger auto wipers than they would have expended on just buying a very reliable $20 rain sensor for each car.

 

[Dilly]

Just to head back on topic...
I did a 50 mile round trip on the A47.
All the speed limits were seen and adjusted up or dow n automatically albeit a bit late.
The only spot that went awry was one transition from DC to two-way traffic where no signs were present and the map data came in well late

 

[AndrewGR]

Just to head back on topic...
I did a 50 mile round trip on the A47.
All the speed limits were seen and adjusted up or dow n automatically albeit a bit late.
The only spot that went awry was one transition from DC to two-way traffic where no signs were present and the map data came in well late

Is NoA available on that section of A47? My understanding is that for some reason, the car only reads speed limits where NoA is unavailable. Most of my driving is on relatively minor roads and most speed limits are read OK. But not if I get on to sections of the A34, or any motorways. If on AP the car does recognise a 30 limit in a nearby village, but doesn’t react until it has been passed at 50.

 

[AndrewGR]

... always accepting that I shouldn’t be using AP or even TACC in that location.

 

[Dilly]

Is NoA available on that section of A47? My understanding is that for some reason, the car only reads speed limits where NoA is unavailable. Most of my driving is on relatively minor roads and most speed limits are read OK. But not if I get on to sections of the A34, or any motorways. If on AP the car does recognise a 30 limit in a nearby village, but doesn’t react until it has been passed at 50.

It lost NoA on the DC at Swaffham but was ok elsewhere.