Musk endeavours

[Howard]

Frankenstein Self Driving

Sucks the financial blood out of Tesla drivers.

regards

Howard

[Hallucigenia]

Who in their right mind puts a racing body out line on an shopping trolly MEB chassis?

It's completely normal, it's just the spiritual successor to the Scirocco, which was a 2-door coupé based on the "shopping trolley" PQ35 platform used in the Mk5/6 Golf, Touran, Eos etc. Eeeks out a few more sales of the platform, and gives a bit of a halo to the badge when used for racing.

[odysseus2000]

Who in their right mind puts a racing body out line on an shopping trolly MEB chassis?

It's completely normal, it's just the spiritual successor to the Scirocco, which was a 2-door coupé based on the "shopping trolley" PQ35 platform used in the Mk5/6 Golf, Touran, Eos etc. Eeeks out a few more sales of the platform, and gives a bit of a halo to the badge when used for racing.

Yes, it used to be the way of selling more shopping trollies, but the world has moved on and no way is this comparable to Tesla in any way that matters to the Tesla target market, just mutton dressed as lamb.

Regards,

[BobbyD]

Who in their right mind puts a racing body out line on an shopping trolly MEB chassis?

It's completely normal, it's just the spiritual successor to the Scirocco, which was a 2-door coupé based on the "shopping trolley" PQ35 platform used in the Mk5/6 Golf, Touran, Eos etc. Eeeks out a few more sales of the platform, and gives a bit of a halo to the badge when used for racing.

It's a bit nippier than the Scirroco, the concept does 0-60 in 3.2s. MEB-entry is going to result in more urbancentric vehicles, smaller, lighter, cheaper city cars which are exactly what tens of millions of want to be BEV drivers need in order to become actual BEV drivers. The sort of thing which might fulfill a mission statement of accelerating BEV uptake and saving the world...

Yes, it used to be the way of selling more shopping trollies, but the world has moved on and no way is this comparable to Tesla in any way that matters to the Tesla target market, just mutton dressed as lamb.

Regards,

Tesla sell so few models that most new models will fall outside their 'target market'. That's a problem for Tesla though, not manufacturers which can handle producing more than 4 3 2 3 models simultaneaously.

[odysseus2000]

Descripion of SpaceX intend to separate the Star Ship from the Super Heavy launcher (8 mins 25):

https://youtu.be/SmM_JhyWeeE

Regards,

[odysseus2000]

BobbyD
Tesla sell so few models that most new models will fall outside their 'target market'. That's a problem for Tesla though, not manufacturers which can handle producing more than 4 3 2 3 models simultaneaously.

Abilty to make lots of ICE cars is not something any shareholder in a legacy car maker should celebrate.

Regards,

[BobbyD]

BobbyD
Tesla sell so few models that most new models will fall outside their 'target market'. That's a problem for Tesla though, not manufacturers which can handle producing more than 4 3 2 3 models simultaneaously.

Abilty to make lots of ICE cars is not something any shareholder in a legacy car maker should celebrate.

Regards,

An inability to exceed or even maintain 4 vehicles on the market at any one time is something a company with the pretension of becoming a successful car manufacturer should find deeply humbling. An organisation with its bandwidth determined by the attention span of a single man whose actually more interested in rockets isn't a viable company it's a vanity project. Tesla needs better leadership.

[odysseus2000]

BobbyD
An inability to exceed or even maintain 4 vehicles on the market at any one time is something a company with the pretension of becoming a successful car manufacturer should find deeply humbling. An organisation with its bandwidth determined by the attention span of a single man whose actually more interested in rockets isn't a viable company it's a vanity project. Tesla needs better leadership.

Tesla model 3 and model y combined are out selling all other electric cars even the super cheap Chinese models, is not something to be humble about.

Regards,

[odysseus2000]

Nio doing an offering at the market:

https://finance.yahoo.com/news/nio-inc- ... 00326.html

Regards,

[odysseus2000]

2 min 32 seconds inspirational video on 4 civilians going to space:

https://twitter.com/netflix/status/1433 ... 37126?s=20

Regards,

[BobbyD]

Germany - Selected Group BEV sales

[odysseus2000]

A video showing the view from the Plaid as it tackles the Nuremberg track with great audio. Just over 7 minutes of extremely fast driving:

https://youtu.be/Ujp3q_aryRA

Regards,

[odysseus2000]

Some interesting general points on batteries and such including that Tesla monitor a large number of labs for advances in battery technology & expect anyone who invents a better battery will come to them to license it. As of now the estimate is that batteries are improving about 8% per year giving a doubling time of approx 72/8 = 9 years, but not clear if this is cost reduction or battery capacity:

https://youtu.be/73MWQZMguHo

Regards,

[odysseus2000]

Plaid model S sets new world record:

https://twitter.com/elonmusk/status/143 ... 51394?s=20

Regards,

[odysseus2000]

One of the first reviews of FSD v10. Looks very impressive (26 mins);

https://youtu.be/at6pjzFKfzs

Regards,

[onthemove]

One of the first reviews of FSD v10. Looks very impressive (26 mins);

https://youtu.be/at6pjzFKfzs

Regards,

Thanks...

When I watched it, Youtube recommended this one, which is a bit more down to earth and less star struck with the technology...

https://www.youtube.com/watch?v=dE8XQai ... DirtyTesla

Still a very long way from reaching fully autonomous... still quite a few quirks and disengagements.

Used the wrong lane a number of times.

The visualisation still jumpy, and even briefly showed a HGV across the car's path when no vehicle was there!

Wasn't able to interpret the road closed and diversion sign.

Didn't recognise or correctly interpret the railroad crossing.

Didn't recognise its turn at a 3-way stop.

Creeping even when visibility of road being turned into is good and unobstructed.

Doesn't seem to make any attempt at all to moderate speed when the road surface conditions are rough - it doesn't look like it's even considering it.
(How about this speed bump https://youtu.be/wWhCt9BB2Iw?t=208 ouch!)

And in both videos, I don't consider that the car is really being challenged : traffic density is fairly light, and virtually no 'left field' events like random pedestrians or other road users doing unconventional or odd things that need dynamically reacting to... no road side car doors open, no pedestrians crossing from behind vehicles, cyclists, bin lorries on their collection routes, etc.

Overall, I think my skyscraper analogy still holds... I'm still not convinced their architecture is aiming 'high' enough for where they need to be.

That they still can't seem to stabilise the jumpiness after multiple updates - the road markings, bins on the edge of the road, and crucially the planned route even when the situation is relatively benign - leads me to still suspect that they're operating close to the limit of what their existing technology stack can deliver here.

It doesn't seem to be able to generate a confident, stable model of the world around it. That it's so jumpy, just leads me to feel that it isn't confident in its interpretation of what's around it which I feel is leading to it frequently change its mind - hence the jumpiness and wavering in the route plan. I mean, watching the video and the visualisation of it driving on the roads with all the bins out, I couldn't help feel that as a human, my mental model of where the bins were just from the 2d video on YouTube, felt confident and stable... yet the visualisation kept changing its mind whether there was 1 bin or 2, or even any bin at all on occasion. I didn't replay it to check, but when I glanced back at the screen, I thought I saw it identify a telegraph post as a bin at one point (?)

While the technology is impressive for what it can do so far, from these videos, I certainly won't be betting on FSD proper being released any time soon.

To release proper FSD it needs to handle safely (one way or another) all situations for the areas that its 'approved' for, and it needs to be able to do it without any disengagements or quirks not just for a 30minute youtube video (which it can't even do that), but time after time after time... way, way, way more than any individual Tesla owner would be able to judge in isolation.

There's still a long way for Tesla to go yet before they can release to the general public anything that people will truly consider proper 'self driving' - that's to say where the driver does not need to be ready to take over, and consider the vehicle occupants as being only passengers (unlike Waymo - hands off the wheel, vs Tesla keep hand on the wheel).

[odysseus2000]

Hi onthemove,

I have now watched many of the v10 drives on youtube.

I concur that this is not ready as a robot taxi and there are still problems as you point out.

However, I also think it is much better than the previous version although it has gone backwards in a few places.

Impossible imho to know if this can be made better than humans. Looking at the videos one of the issues seems to be ambiguity in what the robot sees. If the robot accesses what is there correctly it performs flawlessly, but there are occasions when it is not sure and this leads to either the wrong action, or deer in headlights phenomena of not knowing what to do. My instinct is that these issues can be trained out in much the same way that a human learns how to do things by feedback either from an instructor or from what others do.

As I see things most of the large learning has been done and it is now a question of practice and feedback to iron out remaining troubles

One encouraging fact is that this problem has become a fascination to many folk and the ones who have been chosen to test the updates are behaving responsibly as evidenced by the lack of crashes reported during testing. As things are it is not ready for '"unskilled" users and it is clearly possible to believe that it never will, but it is also possible to believe that it will. For now I am open minded on this question, being unable to quite believe that it will work and yet having seen how far it has come unable to quite believe that it is impossible.

Regards,

[BobbyD]

While the technology is impressive for what it can do so far, from these videos, I certainly won't be betting on FSD proper being released any time soon.

The metaphorical million miles away would be an incredibly generous estimate.

[Howard]

Having driven a Tesla in town (but not with the latest FSD version) I experienced a couple of situations where it will be difficult for the car to behave as safely as a human driver.

The first situation was following a car being driven slightly erratically some distance ahead. The driver ahead got in the wrong lane and suddenly moved to the next. In this situation the Tesla would still come up, in my view, too close to the car in front. As a driver, one is aware that a second erratic move is likely. Sure enough at the next roundabout the driver in front repeated the manoeuvre without indicating. The Tesla, left to its own devices, would have been too close and would have had to brake hard to avoid running into the back of the vehicle changing lanes.

Something similar can happen with pedestrians. In a line of slow moving traffic I saw a pedestrian ahead who caught my eye and indicated he was going to jog in front of me. He correctly guessed that I would slow down to allow him to do this. The Tesla software would have no indication of this almost imperceptible human interaction. I imagine It would only react to the sudden presence of a human dangerously close in front by braking sharply risking a collision from the car behind.

regards

Howard

[odysseus2000]

Having driven a Tesla in town (but not with the latest FSD version) I experienced a couple of situations where it will be difficult for the car to behave as safely as a human driver.

The first situation was following a car being driven slightly erratically some distance ahead. The driver ahead got in the wrong lane and suddenly moved to the next. In this situation the Tesla would still come up, in my view, too close to the car in front. As a driver, one is aware that a second erratic move is likely. Sure enough at the next roundabout the driver in front repeated the manoeuvre without indicating. The Tesla, left to its own devices, would have been too close and would have had to brake hard to avoid running into the back of the vehicle changing lanes.

Something similar can happen with pedestrians. In a line of slow moving traffic I saw a pedestrian ahead who caught my eye and indicated he was going to jog in front of me. He correctly guessed that I would slow down to allow him to do this. The Tesla software would have no indication of this almost imperceptible human interaction. I imagine It would only react to the sudden presence of a human dangerously close in front by braking sharply risking a collision from the car behind.

regards

Howard

Yes, this is all true for you, but how many other human drivers would be so clewed up all the time and never distracted?

I would estimate that it is some number less than 100% and for that number, what ever it may be, we have the seeds of an accident of which there are about 10 per day that lead to fatalities or serious injury.

Whether robots will ever reach the standard of say a focused formula 1 driver we do not know, but we do know that most drivers are not focused formula 1 types but some mix of lesser ability and the question is can robots perform on average better than a ensemble of humans and if so how much better?

If this turns out to be the case at what point do politicians accept that robots are allowable drivrers?

Regards,