Musk endeavours

[odysseus2000]

VW putting 60billion Euro into gasoline car development:

https://uk.motor1.com/news/722598/volks ... n-engines/

Regards,

[odysseus2000]

Share holder meeting tomorrow, so just today and tomorrow to sell pre the vote on Elon's pay package and company move to Texas.

Shares were under pressure yesterday but did not dip significantly. Possibly enough folk now believe that Elon's deal will pass and that this will make shorts unwilling to press.

For what its worth I liquidated a small amount, just in case there is a big sell off, but otherwise I still like the company and have it has the largest position in my portfolio.

It will be interesting to see what the share price does post the meeting and vote and then going forwards. We are now closer to the Robo taxi event on the 8th August, so shorts will likely have to hit the shares before then or risk being caught in a squeeze if the event is well received.

Regards,

[odysseus2000]

Tesla cuts prices in most major markets:

https://www.reuters.com/business/autos- ... 024-04-21/

All the commentary is that these price cuts are bearish for Tesla, but the bearishness will be most acute in legacy auto & legacy leasing companies.

This move is the exact play book of Henry T Ford, one that made his investors rich & which seriously hurt his competitors. Eventually the Dodge brothers sued him for unfair pricing, but his sales were profitable.

Regards,

[odysseus2000]

Both Tesla pay package & move to Texas both currently passing by large margins:

https://x.com/elonmusk/status/180108478 ... DCpgdbFBxg

Shares up over 6% in pre market, likely short squeeze happening.

Regards,

[odysseus2000]

The Tesla shareholder meeting saw the passing of Elon’s pay proposal & the move to Texas.

Once the formalities were over there was a super interesting & inspiring presentation by Elon that ended with a series of questions from attendees who were all intensely focused on supporting the current endeavours & leadership.

I am not going to attempt to précis it as there is way too much to consider & anyone interested in the auto business, AI & robotics needs to listen to the entire presentation which can be found here:

https://youtu.be/remZ1KMR_Z4

Or accessed from the Tesla investor relations page.

One caveat is that if you are aligned with how things now are & the way things have been done historically you will likely struggle with the implications of what is said & what it means going forwards. An open mind to new ideas is needed to make sense of all that is forecast including the prediction that there will be (2-3)x more biped robots than humans & the implications for investments & the human species.

Regards,

[Howard]

BYD release a video of their assisted driving capability at night through busy Chinese city roads. To a layman this looks impressive. It is at x 1.5 speed.

I wonder how this compares with a Tesla's capability in a similar Chinese situation? Has Tesla released a comparable video drive in a congested city with narrow streets, stray pedestrians, motorbikes squeezing past in the dark. It would be interesting to see a comparison.

https://weibo.com/1746221281/OiLYFF0ng

regards

Howard

[odysseus2000]

BYD release a video of their assisted driving capability at night through busy Chinese city roads. To a layman this looks impressive. It is at x 1.5 speed.

I wonder how this compares with a Tesla's capability in a similar Chinese situation? Has Tesla released a comparable video drive in a congested city with narrow streets, stray pedestrians, motorbikes squeezing past in the dark. It would be interesting to see a comparison.

https://weibo.com/1746221281/OiLYFF0ng

regards

Howard

Interesting how the BYD video shows the driver hands very close to the steering wheel, seemingly ready to grab it should it do something dangerous.

By contrast in this video the driver is involved in a no hands telephone call and keeps his hands well back from the steering wheel and the Tesla when told to go initially waits as another car pulls out behind him and then the Tesla goes:

https://youtu.be/aXwSwNLaPJ4?si=sNSOq9tegedsNqgY

According to Musk, the time between interventions has now reached about 10,000 miles, so one of the limits to training is that there are now very few interventions to train on. One way around this is to use shadow driving from none FSD cars to see the difference between what the human does and what the car would have done if it was driving.

Regards,

[ReformedCharacter]

According to Musk, the time between interventions has now reached about 10,000 miles, so one of the limits to training is that there are now very few interventions to train on. One way around this is to use shadow driving from none FSD cars to see the difference between what the human does and what the car would have done if it was driving.

Regards,

I got the impression from Tesla AI 2021 that a lot of training is performed with synthetic data:

Training a full self-driving A.I. system requires data, a lot of data… If there is anything Tesla has, it’s data, a lot of data collected directly from the thousands of cars they have sold and are cruising the roads every day. However they couldn’t use these raw images directly to train their AI models, so they developed more AI models to auto-tag the raw images. But still, that wasn’t enough, comes simulation to create realistic faithful synthetic images to complement the real-world data.

https://anyverse.ai/news/tesla-ai-day-physically-correct-synthetic-data/

Tesla AI 2021:

https://www.youtube.com/watch?v=j0z4FweCy4M

RC

[Howard]

BYD release a video of their assisted driving capability at night through busy Chinese city roads. To a layman this looks impressive. It is at x 1.5 speed.

I wonder how this compares with a Tesla's capability in a similar Chinese situation? Has Tesla released a comparable video drive in a congested city with narrow streets, stray pedestrians, motorbikes squeezing past in the dark. It would be interesting to see a comparison.

https://weibo.com/1746221281/OiLYFF0ng

regards

Howard

Interesting how the BYD video shows the driver hands very close to the steering wheel, seemingly ready to grab it should it do something dangerous.

By contrast in this video the driver is involved in a no hands telephone call and keeps his hands well back from the steering wheel and the Tesla when told to go initially waits as another car pulls out behind him and then the Tesla goes:

https://youtu.be/aXwSwNLaPJ4?si=sNSOq9tegedsNqgY

According to Musk, the time between interventions has now reached about 10,000 miles, so one of the limits to training is that there are now very few interventions to train on. One way around this is to use shadow driving from none FSD cars to see the difference between what the human does and what the car would have done if it was driving.

Regards,

This is an apples and oranges comparison. The route taken through San Francisco involved wide streets, dual carriageways and disciplined pedestrians and cars. Not a valid comparison. We need to see a proper test of a Tesla in a crowded less disciplined environment.

I've driven a lot in the USA in cities and even in New York on the main roads like the Tesla drive, it's much more disciplined than the Chinese route which was more like the back streets of Paris or Milan with many more unpredictable hazards like the motorbike squeezing through a gap.

And what happened at the end of the Tesla video? A lot of manual use of the steering wheel. Looking at the screen, it appeared to completely miss the left hand turn and require the driver to reverse out of a difficult situation.

We need to see a proper comparison with a Tesla driven in Europe or China.

The Chinese authorities are still to give Tesla approval for tests like the BYD drive. I guess they will want to see how a Tesla can deal with a real life Chinese situation.

Challenge for you Ody - show us a video of a proper Tesla test in a similar environment to the BYD drive - can you do it?

regards

Howard

[odysseus2000]

According to Musk, the time between interventions has now reached about 10,000 miles, so one of the limits to training is that there are now very few interventions to train on. One way around this is to use shadow driving from none FSD cars to see the difference between what the human does and what the car would have done if it was driving.

Regards,

I got the impression from Tesla AI 2021 that a lot of training is performed with synthetic data:

Training a full self-driving A.I. system requires data, a lot of data… If there is anything Tesla has, it’s data, a lot of data collected directly from the thousands of cars they have sold and are cruising the roads every day. However they couldn’t use these raw images directly to train their AI models, so they developed more AI models to auto-tag the raw images. But still, that wasn’t enough, comes simulation to create realistic faithful synthetic images to complement the real-world data.

https://anyverse.ai/news/tesla-ai-day-physically-correct-synthetic-data/

Tesla AI 2021:

https://www.youtube.com/watch?v=j0z4FweCy4M

RC

Things have changed greatly in the last 3 years.

Tesla say they are not compute constrained & are now constrained by the lack of interventions as i noted before. There was no discussion of synthetic driving training other than shadow mode where they take the difference between non FSD car predictions of what it would do & what the human does.

Regards,

[odysseus2000]

Interesting how the BYD video shows the driver hands very close to the steering wheel, seemingly ready to grab it should it do something dangerous.

By contrast in this video the driver is involved in a no hands telephone call and keeps his hands well back from the steering wheel and the Tesla when told to go initially waits as another car pulls out behind him and then the Tesla goes:

https://youtu.be/aXwSwNLaPJ4?si=sNSOq9tegedsNqgY

According to Musk, the time between interventions has now reached about 10,000 miles, so one of the limits to training is that there are now very few interventions to train on. One way around this is to use shadow driving from none FSD cars to see the difference between what the human does and what the car would have done if it was driving.

Regards,

This is an apples and oranges comparison. The route taken through San Francisco involved wide streets, dual carriageways and disciplined pedestrians and cars. Not a valid comparison. We need to see a proper test of a Tesla in a crowded less disciplined environment.

I've driven a lot in the USA in cities and even in New York on the main roads like the Tesla drive, it's much more disciplined than the Chinese route which was more like the back streets of Paris or Milan with many more unpredictable hazards like the motorbike squeezing through a gap.

And what happened at the end of the Tesla video? A lot of manual use of the steering wheel. Looking at the screen, it appeared to completely miss the left hand turn and require the driver to reverse out of a difficult situation.

We need to see a proper comparison with a Tesla driven in Europe or China.

The Chinese authorities are still to give Tesla approval for tests like the BYD drive. I guess they will want to see how a Tesla can deal with a real life Chinese situation.

Challenge for you Ody - show us a video of a proper Tesla test in a similar environment to the BYD drive - can you do it?

regards

Howard

Since Tesla supervised FSD only operates in the US one can’t find examples of other nations street systems & FSD.

Similarly as far as I know byd do not have authorization to robo drive in the US so there are no side by side comparisons there either.

Regards,

[Howard]

This is an apples and oranges comparison. The route taken through San Francisco involved wide streets, dual carriageways and disciplined pedestrians and cars. Not a valid comparison. We need to see a proper test of a Tesla in a crowded less disciplined environment.

I've driven a lot in the USA in cities and even in New York on the main roads like the Tesla drive, it's much more disciplined than the Chinese route which was more like the back streets of Paris or Milan with many more unpredictable hazards like the motorbike squeezing through a gap.

And what happened at the end of the Tesla video? A lot of manual use of the steering wheel. Looking at the screen, it appeared to completely miss the left hand turn and require the driver to reverse out of a difficult situation.

We need to see a proper comparison with a Tesla driven in Europe or China.

The Chinese authorities are still to give Tesla approval for tests like the BYD drive. I guess they will want to see how a Tesla can deal with a real life Chinese situation.

Challenge for you Ody - show us a video of a proper Tesla test in a similar environment to the BYD drive - can you do it?

regards

Howard

Since Tesla supervised FSD only operates in the US one can’t find examples of other nations street systems & FSD.

Similarly as far as I know byd do not have authorization to robo drive in the US so there are no side by side comparisons there either.

Regards,

This is a pretty critical issue. Tesla fans have no idea how the cars will perform in world markets outside the USA.

As an engineer you surely must accept that Tesla is working to tolerances of feet whilst BYD's controls work to tolerances of inches as demonstrated by the two videos.

I've driven a Tesla and it's not very confident when on UK roads dealing with a few inch wide passing room. The videos of Tesla's failures to self park in UK situations have been posted on this thread several times.

Your video suggests that FSD may be severely tested in China, the largest EV market in the world. Its performance there will almost certainly affect the share price in the future.

regards

Howard

[odysseus2000]

Since Tesla supervised FSD only operates in the US one can’t find examples of other nations street systems & FSD.

Similarly as far as I know byd do not have authorization to robo drive in the US so there are no side by side comparisons there either.

Regards,

This is a pretty critical issue. Tesla fans have no idea how the cars will perform in world markets outside the USA.

As an engineer you surely must accept that Tesla is working to tolerances of feet whilst BYD's controls work to tolerances of inches as demonstrated by the two videos.

I've driven a Tesla and it's not very confident when on UK roads dealing with a few inch wide passing room. The videos of Tesla's failures to self park in UK situations have been posted on this thread several times.

Your video suggests that FSD may be severely tested in China, the largest EV market in the world. Its performance there will almost certainly affect the share price in the future.

regards

Howard

The fundamental difference is that I chart my course based on my understanding as a professional physicist of the underlying technology.

I am not influenced by eye or ear candy.

Sure the byd looks impressive, but how independent was the tester, how many takes were needed, why does he have his hands so close to the wheel. What is the level of output of the lidar & phased arrays etc & at what frequency? We know lidar, phased array, flir etc work well in military situations such as cruise missiles & bombs being steered into fire escapes etc & the sniper pods of the f16 & other military jets through to the B2, but those lasers are powerful enough to blind people & have very tight usage cases. What do we know about byd & other Chinese LiDAR etc? What do we know about the confusion to neural nets when the system gets inputs from many systems rather than just vision , some cases radar too as in the Tesla case? What do we know about the effects of atmospheric conditions on these systems at the power levels being used.

These are just some elementary questions. For the byd system to be taken seriously we would need independent reviews, tear downs & extensive testing. There was a series of tests performed in China a few years back comparing Tesla to several local makers. Tesla was so superior the others were not safe, hitting pedestrians & a whole host of other mistakes.

Regards,

[Howard]

This is a pretty critical issue. Tesla fans have no idea how the cars will perform in world markets outside the USA.

As an engineer you surely must accept that Tesla is working to tolerances of feet whilst BYD's controls work to tolerances of inches as demonstrated by the two videos.

I've driven a Tesla and it's not very confident when on UK roads dealing with a few inch wide passing room. The videos of Tesla's failures to self park in UK situations have been posted on this thread several times.

Your video suggests that FSD may be severely tested in China, the largest EV market in the world. Its performance there will almost certainly affect the share price in the future.

regards

Howard

The fundamental difference is that I chart my course based on my understanding as a professional physicist of the underlying technology.

I am not influenced by eye or ear candy.

Sure the byd looks impressive, but how independent was the tester, how many takes were needed, why does he have his hands so close to the wheel. What is the level of output of the lidar & phased arrays etc & at what frequency? We know lidar, phased array, flir etc work well in military situations such as cruise missiles & bombs being steered into fire escapes etc & the sniper pods of the f16 & other military jets through to the B2, but those lasers are powerful enough to blind people & have very tight usage cases. What do we know about byd & other Chinese LiDAR etc? What do we know about the confusion to neural nets when the system gets inputs from many systems rather than just vision , some cases radar too as in the Tesla case? What do we know about the effects of atmospheric conditions on these systems at the power levels being used.

These are just some elementary questions. For the byd system to be taken seriously we would need independent reviews, tear downs & extensive testing. There was a series of tests performed in China a few years back comparing Tesla to several local makers. Tesla was so superior the others were not safe, hitting pedestrians & a whole host of other mistakes.

Regards,

The critical issue for investors is not techno-babble at the moment. Surely it's whether Tesla gets approval from the Chinese authorities to join autonomous testing on Chinese roads in a real world environment.

One can waffle about theoretical issues. But if one has never driven a Tesla and a number of its rivals and experienced the shortcomings it's all hypothetical.

Several of Tesla's competitors have been authorised for testing but Tesla hasn't yet. Let's wait and see if it gets approval to do real life testing in the biggest EV market in the world. This is important because Tesla's market share is dropping in China and it needs approval to keep up with faster growing competitors.

regards

Howard

[odysseus2000]

The fundamental difference is that I chart my course based on my understanding as a professional physicist of the underlying technology.

I am not influenced by eye or ear candy.

Sure the byd looks impressive, but how independent was the tester, how many takes were needed, why does he have his hands so close to the wheel. What is the level of output of the lidar & phased arrays etc & at what frequency? We know lidar, phased array, flir etc work well in military situations such as cruise missiles & bombs being steered into fire escapes etc & the sniper pods of the f16 & other military jets through to the B2, but those lasers are powerful enough to blind people & have very tight usage cases. What do we know about byd & other Chinese LiDAR etc? What do we know about the confusion to neural nets when the system gets inputs from many systems rather than just vision , some cases radar too as in the Tesla case? What do we know about the effects of atmospheric conditions on these systems at the power levels being used.

These are just some elementary questions. For the byd system to be taken seriously we would need independent reviews, tear downs & extensive testing. There was a series of tests performed in China a few years back comparing Tesla to several local makers. Tesla was so superior the others were not safe, hitting pedestrians & a whole host of other mistakes.

Regards,

The critical issue for investors is not techno-babble at the moment. Surely it's whether Tesla gets approval from the Chinese authorities to join autonomous testing on Chinese roads in a real world environment.

One can waffle about theoretical issues. But if one has never driven a Tesla and a number of its rivals and experienced the shortcomings it's all hypothetical.

Several of Tesla's competitors have been authorised for testing but Tesla hasn't yet. Let's wait and see if it gets approval to do real life testing in the biggest EV market in the world. This is important because Tesla's market share is dropping in China and it needs approval to keep up with faster growing competitors.

regards

Howard

It is purely a probability game.

Who has the best chance of getting to market with a practical & affordable robot system that is many times safer than a human driven car.

Getting to market fast with a system that has flaws will be murderous to reputations & expensive in settling legal claims.

Regards,

[Howard]

The problem for Tesla is if the Chinese authorities look at videos like the one linked below they will see that the latest model 2024 Highland model 3 has serious flaws in the sensors. The car won't stop for a pedestrian at low speeds. It can't see objects in front of it. The wheelie bin test is hilarious because the car is distracted by an imaginary bin and touches a pedestrian.

Even stranger, an older Tesla with parking sensors performs better than the 2024 model at spotting a smaller obstruction.

Watch from 11 minutes.

https://www.youtube.com/watch?v=-t2QykN4hsk

Another challenge for you Ody.

This isn't just a software problem, it suggests a hardware problem which can't be solved by OTA updates.

Explain why the 2024 car's sensors are so flaky at low speeds and explain why this might not suggest problems at higher speeds. What is the problem with the technology?

This problem should be of concern to investors who are being told that autonomous driving is possible with existing cars?

Regards

Howard

[Howard]

We may see how FSD performs in China fairly soon as today it was reported that:

"Shanghai's Nanhui New Town is facilitating 10 Tesla vehicles to carry out the pilot use of FSD (Full Self-Driving), local media SH Observer said today, citing Lu Sen, director of the data department of the Shanghai Free Trade Zone Lingang New Area Administrative Committee.

The report did not provide more details on FSD or when the 10 Tesla vehicles will be ready to begin pilot use of FSD."

https://cnevpost.com/2024/06/14/shangha ... fsd-pilot/

Hopefully the cars' sensors work ok.

Regards

Howard

[odysseus2000]

The problem for Tesla is if the Chinese authorities look at videos like the one linked below they will see that the latest model 2024 Highland model 3 has serious flaws in the sensors. The car won't stop for a pedestrian at low speeds. It can't see objects in front of it. The wheelie bin test is hilarious because the car is distracted by an imaginary bin and touches a pedestrian.

Even stranger, an older Tesla with parking sensors performs better than the 2024 model at spotting a smaller obstruction.

Watch from 11 minutes.

https://www.youtube.com/watch?v=-t2QykN4hsk

Another challenge for you Ody.

This isn't just a software problem, it suggests a hardware problem which can't be solved by OTA updates.

Explain why the 2024 car's sensors are so flaky at low speeds and explain why this might not suggest problems at higher speeds. What is the problem with the technology?

This problem should be of concern to investors who are being told that autonomous driving is possible with existing cars?

Regards

Howard

This is all the old system: 11.n, not the new version 12.4, so the whole thing is obsolete & tells us nothing about the latest software.

Regards,

[odysseus2000]

Tesla has just got permission to test Tesla FSD in China with ten cars:

https://carnewschina.com/2024/06/14/bre ... cial-says/

Regards,

[Howard]

The problem for Tesla is if the Chinese authorities look at videos like the one linked below they will see that the latest model 2024 Highland model 3 has serious flaws in the sensors. The car won't stop for a pedestrian at low speeds. It can't see objects in front of it. The wheelie bin test is hilarious because the car is distracted by an imaginary bin and touches a pedestrian.

Even stranger, an older Tesla with parking sensors performs better than the 2024 model at spotting a smaller obstruction.

Watch from 11 minutes.

https://www.youtube.com/watch?v=-t2QykN4hsk

Another challenge for you Ody.

This isn't just a software problem, it suggests a hardware problem which can't be solved by OTA updates.

Explain why the 2024 car's sensors are so flaky at low speeds and explain why this might not suggest problems at higher speeds. What is the problem with the technology?

This problem should be of concern to investors who are being told that autonomous driving is possible with existing cars?

Regards

Howard

This is all the old system: 11.n, not the new version 12.4, so the whole thing is obsolete & tells us nothing about the latest software.

Regards,

This is the system which Tesla fans thought was so wonderful until only a couple of weeks ago. The new release uses the same sensors and cameras fitted to the brand new 2024 model which failed the low speed tests in the video. Release 12.4 is yet to be tested in real life situations.

If you look at RS Symons initial test of 12.4 you will see it thinks it is on a US road and it doesn't look very comfortable in UK conditions.

However, the key test for the software is in China and the comparison with rival brands. Let's hope it works.

regards

Howard