I think, in short, yes.

But you got to look at where they've enabled these options and how the car drives. US, China and Aus have HUGE wide roads built for cars. Milton Keynes that had AV test vehicles are similar. While most UK roads are not designed for cars let alone the amount of cars parked on the road.

Unpopular opinion, I think best first step for UK AV (and EV adoption; and traffic flow) is to have a lot less but well defined parking on ALL roads. I personally really like idea of Japan's proof-of-parking requirement for car ownership.

Around this part of the world, I think less densely packed EU countries have a better chance of AV running smoothly. But alas, Germany (and thus their lobbying industrial partners) have huge political say in EU. They'd rather allow remote controlled cars than AV. 🤣

Removed due to Merc update linked below, it was allowed and approved by German gov body.

 

But alas, Germany (and thus their lobbying industrial partners) have huge political say in EU. They'd rather allow remote controlled cars than AV. 🤣

Not quite Mercedes has level 3 autonomous driving up to 95kmh on all the autobahn network.

https://group.mercedes-benz.com/innovations/product-innovation/autonomous-driving/drive-pilot-95-kmh.html

ok slow speeds but nevertheless they are moving in the direction of autonomous driving but given how crowded the autobahn are in many areas and the dreadful mix of Lorrie’s and in some areas unrestricted speed they need to be careful.
It really will be an issue I think for many European countries to allow autonomous driving given the nature of much of the road network though we in France have a pretty easy Route National and motorway network but as far as I know there is nothing happening yet in autonomous driving laws.
Not sure just allowing it really makes any difference to a country the owners of the products are the ones who get all the benefit so cannot see it matters whether the UK has it or not at this stage.

 

Considering that FSD currently is at an average of 1 intervention for each 400 miles, we are still very much away from the point where FSD will actually be FSD. In fact, FSD has been renamed FSD (Supervise) for a reason. IMO, Tesla will never reach full autonomy, especially if they keep going down the cul-de-sac of ´Vision Only´.

If you don´t believe me, maybe this vid will change your mind. It´s two Tesla influences doing the ´coast to coast´ trip that Musk promised would be possible in 2017 or thereabouts. Spoiler: They make it to 2.5% of the proposed distance:

We´re talking about road safety and human lives here. Relaxing the rules just to ´keep up with other countries´ sounds like a bad decision.

EDIT:
Look at the timeline of Elon's promises regarding FSD. It´s quite sobering: ELON FSD TIMELINE

 

I do hope so.

 

My thoughts exactly.

Our roads evolved from muddy ruts and wooden carts. FSD then was called 'an ox' and didn't go fast enough for errors to be too catastrophic.

What the hell would FSD do here?

 

Having been in a Waymo robotaxi in SF, I really like the idea of autonomous vehicles. But I am quite nervous about Tesla's solution. Waymo drives like an experienced, but cautious, limo/executive driver. Tesla aims for something similar, but its lack of nose camera (on most models) seems to mean it has to 'chance' it around junctions more. Tesla have also promised FSD will come to far simpler cars with older cameras and computer systems, which seems optimistic. If I were a betting man, I would not bet on Tesla succeeding before others do.

By the way, there are rumours that Waymo are launching in London next year, so I don't think we're that far behind. I wonder what the black cabs will think about that, they were not at all happy about Uber.

 

Interesting take - I get what you mean about Waymo feeling more cautious, but that's kind of the trade-off. Tesla's going for adaptability by training its neural net on millions of real-world edge cases, so even without a nose camera it's learning to handle junctions in a more human-like way. Waymo is brilliant where it's mapped, but Tesla's strength is scalability - it doesn't need high-def maps for every street before it can operate. Both approaches have merits, but if Tesla pulls it off, it'll roll out globally much faster.

 

camera it's learning to handle junctions in a more human-like way.

This is not the solution!

 

Tesla doesn't have fully working FSD regardless of all the media hype. After that you have the UK roads... How long it will take for the system to cut off because some arse in white van driving in and out of traffic, not to mention millions of roundabouts that are not present (widely) in the US for example. As good as Tesla software integration is, the roads, the road discipline exhibited by the other drivers is not under Tesla software control (thanks God for that one)

 

Totally agree - FSD isn't "fully working" yet, and UK roads will definitely test it harder than most places. White vans, endless roundabouts, and unpredictable drivers are a nightmare for any system. That said, the level of driver assistance it already offers is still miles ahead of basic or even "enhanced" Autopilot. In fact, calling those features "autopilot" probably does more harm than good because it sets people up for misuse. FSD supervised at least makes it clear you're still in charge while giving you a much stronger safety net.

 

That said, the level of driver assistance it already offers is still miles ahead of basic or even "enhanced" Autopilot. In fact, calling those features "autopilot" probably does more harm than good because it sets people up for misuse. FSD supervised at least makes it clear you're still in charge while giving you a much stronger safety net.

For someone that used to own Tesla for a few years with the enhanced autopilot..I tend to disagree ( if I understood you correctly)
My Smart #1 road driving assistance is excellent. The car hold/follow the road impeccably, without any hesitation. You need to keep a hand on the wheel of course but other than that, is as good as Tesla. I drove 58 miles on motorway and dual lane road entirely with road assist activated and the car didn't ghost braked at all. Something I cannot say about Tesla that's famous for it.
Lane/proximity warning is superb as well. Your blind spot is very well protected regardless of absence of side cameras. I have HUD and all the warnings are clearly showing on it.
In this aspect Tesla was...yes was the leader. Right now...they are not. Ford got fullself driving in the US, officially certified....Tesla are NOT

 

I was referring specifically to Tesla's LKA (lane keeping assist), which works fine when conditions are perfect but quickly falls apart when things change like in roadworks, faded lines, poor markings, or messy junctions. That's the main weakness with calling it "autopilot." It gives people the wrong idea when in reality it's just a basic lane assist system dressed up with a fancy name.

Where FSD stands apart is that it uses the neural nets trained on millions of real-world edge cases, so it doesn't just rely on painted lines to stay centred. It can interpret context, road edges, traffic flow, and junction layouts in a way LKA simply can't. If Tesla's basic Autopilot functions (like LKA) were powered by the same FSD networks, it would already be far safer and less prone to the quirks people often complain about (ghost braking, confusion in roadworks, etc.).

That's why I think "FSD supervised" is the only one worth taking seriously - it makes the driver's responsibility clear while still delivering far more capability and safety than the so-called "autopilot."

 

Tbf the software in autopilot is massively behind FSD and it's one thing Tesla could/should sort out while the rest of the world is waiting for FSD to become available.

I suspect the problem, from Teslas pov, is that if autopilot is improved there's a lot of people that will choose to just stay at that level rather than splash out on FSD.

 

AP on my Juniper is far better than any previous version of AP that I have experienced. Over 3,500 motorway miles we have only had 2 cases of phantom braking - both in the same place, where the road dips sharply under a bridge.

However, as Juniper forces you to either use AP or TACC, the car is set to AP. This means that AP gets activated in 20mph zones.. or it would, as after a few uses, it became clear that AP is not ready for non-motorway use in the UK.

Things like the road widening for a bus stop can cause the car to hunt for a lane. Zebra crossings and hatched areas don’t seem to be detected. Parking spaces in driveways not identified. Blue light vehicles not identified. Cyclists overtaking you in a dark lane at night when it is raining. Etc..

All this will come, and IMO only a few of these would be genuinely improved with a broader sensor suite.

My main concerns with Tesla Vision are:

Lack of sensor redundancy - the (supervised) tag essentially means that the driver is the redundancy.

Inability to correctly sense and respond to edge cases that it doesn’t recognise. See the recent coast to coast example.

Long distance sensing at highway speeds. There was a recent example where a car on FSD hit an another car that had been in an accident in the fast lane. It was at night and there was a lot of smoke around the accident which prevented the Tesla from seeing the obstruction. This scenario is exactly what radar is for.

 

AP on my Juniper is far better than any previous version of AP that I have experienced. Over 3,500 motorway miles we have only had 2 cases of phantom braking - both in the same place, where the road dips sharply under a bridge.

However, as Juniper forces you to either use AP or TACC, the car is set to AP. This means that AP gets activated in 20mph zones.. or it would, as after a few uses, it became clear that AP is not ready for non-motorway use in the UK.

Things like the road widening for a bus stop can cause the car to hunt for a lane. Zebra crossings and hatched areas don’t seem to be detected. Parking spaces in driveways not identified. Blue light vehicles not identified. Cyclists overtaking you in a dark lane at night when it is raining. Etc..

All this will come, and IMO only a few of these would be genuinely improved with a broader sensor suite.

My main concerns with Tesla Vision are:

Lack of sensor redundancy - the (supervised) tag essentially means that the driver is the redundancy.

Inability to correctly sense and respond to edge cases that it doesn’t recognise. See the recent coast to coast example.

Long distance sensing at highway speeds. There was a recent example where a car on FSD hit an another car that had been in an accident in the fast lane. It was at night and there was a lot of smoke around the accident which prevented the Tesla from seeing the obstruction. This scenario is exactly what radar is for.

I posted a video sometime back of the Chinese doing a test to 20 or 25 models of EV testing the system. They've done it in a real motorway (closed for the duration) in day and night conditions in all weather. Really interesting watch.
(I'm on the move right now, so it won't be easy to find the thread but will try)


Found it 🤣🤣🤣

https://www.speakev.com/threads/excellent-video-discussing-how-effective-all-the-safety-system-are-inside-the-evs.192502/?post_id=3805514#post-3805514

 

I agree. I doesn't look like the solution now but maybe in the future it will actually be able to surpass human driving by enough. If that's not the case then it's just a foool's errand and they will be forced to adapt. Whichever technology makes it there first (lidar or vision) the other one will have to follow suit.

For the time being it has to be the solution whilst AI cars have to share the road with humans, likely for the next 20-odd years.

Whats the point in spending all this money and R&D and ending up with drivers which are no better than humans and lack any self preservation. If we're going to replace people, which I'm not against, then it needs to be better not just almost as good.

 

In the UK human drivers are responsible for 1,500 deaths in addition to around 25,000 seriously injuries (that's defined as an injury that results in a stay in hospital longer than 7 days).

Basically, we're shit at driving. If trains killed 1,500 people a year there'd be an inquest to figure out why! We accept the risk of driving because it is extremely convenient to drive places. But if we could eliminate the driver from that, and got deaths down to around 10% of the current figure (or even better) then I think we should do it.

As for why it's worth the R&D? Well average driver does 9,000 miles a year in the UK. 30 million cars. If that person is willing to spend just 5p a mile on an autonomous driving system then that's £13.5bn a year in revenue. Over 20 years £270bn. That seems like a pretty good pay off even if you have to spend multiple billions to get the technology right. And the technology can be used anywhere in the world not just the UK.

 

I agree. I doesn't look like the solution now but maybe in the future it will actually be able to surpass human driving by enough. If that's not the case then it's just a foool's errand and they will be forced to adapt. Whichever technology makes it there first (lidar or vision) the other one will have to follow suit.

For the time being it has to be the solution whilst AI cars have to share the road with humans, likely for the next 20-odd years.

Whats the point in spending all this money and R&D and ending up with drivers which are no better than humans and lack any self preservation. If we're going to replace people, which I'm not against, then it needs to be better not just almost as good.

And you would be right but tell me something. Would you not be just as happy with a Tesla bot that can do your house work just as well as you could do it while you get to enjoy more time with your family doing other things? Is it not the same with FSD. If I had a car that can drive just as well as I can, that would free up a lot of my time and make some long journeys more enjoyable... That is, of course, just my humble opinion and anyone is free to think differently.

 

And you would be right but tell me something. Would you not be just as happy with a Tesla bot that can do your house work just as well as you could do it while you get to enjoy more time with your family doing other things? Is it not the same with FSD. If I had a car that can drive just as well as I can, that would free up a lot of my time and make some long journeys more enjoyable... That is, of course, just my humble opinion and anyone is free to think differently.

The difference is if dust-a-tron 5000 messes up no one dies.

I'm not really interested in making driving fun or more productive, I'm far more interested in getting distracted meat sacks away from controlling tonnes of metal. People are bad drivers and that's with their own lives at risk, if a computer is doing the driving it needs to be better, in a hundred years people will look back and wince that this many people were allowed to operate this many cars around pedestrians and each other. We have one chance to really improve the roads and if we let the auto manufacturers call it at as good as a person then the chance is gone.

 

I wouldn't be so sure about that first part. There are numerous movies out there with "tools" going rogue😂

I do however agree that in the ideal world, the autopilot/FSD or whatever they may call it, should be much better than people and I also believe that this is the aim. The same way the smartphone became smarter than it's owner. All we need is time and someone to give Tesla the attention it deserves. We need more visionaries. Elon has gone mad from politics😂

 

Theres “left behind”:

in the sense are we developing the capability. We have Wayne which seems to be one of the many credible companies working in this space

in the sense we don’t have it on our roads. Well that’s not being left behind, that’s letting others have all the accidents etc and when they’ve developed it we can have the more mature product in relatively short order.

and then there’s the question who really wants a supervised system? I don’t, and I personally question the motivation and driving ability of anyone who genuinely thinks the Tesla supervised system is better than their ability to drive. I can accept it might be of benefit to those with medical conditions, but that’s largely, and understandably because we have a relatively relaxed attitude to such people.

 

What are the benefits of being “ahead” on this (possibly futile) journey? Even if we can foresee a point at which all powered vehicles have this sort of technology available, our roads and streets will (and should) remain open to pedestrians, cyclists and even horses.

 

... our roads and streets will (and should) remain open to pedestrians, cyclists and even horses.

Even blind, stupid and drunk people, some of whom may be all three, possibly for causally connected reasons. There might even be Labour MPs walking about.

 

I can see how fully autonomous cars can make longer journeys easier but, personally, I’d prefer to be actually driving, rather than just watching meters & monitors, I’ve done that for a living. It’s so utterly boring & difficult to stay awake, never mind paying full attention. Don’t think it’ll catch on as fast as the advocates think in the same way that not everyone is ready to go completely electric, either at home or on the road, never mind getting in a car that drives itself!

But, for short journeys, if I want an autonomous drive, I’ll just get a taxi.

 

Driverless Waymo car makes illegal U-turn in California - and can't be given a ticket

Police say legislation enabling officers to issue notices directly to companies is being prepared - but, for now, they can't sanction a "robot".

news.sky.com

 

Whoa. Wait until Musk reads this. Robotaxis spreading death and terror in 3, 2, 1.....