Berlin to get a Tesla Gigafactory

1,375 posts in this topic

I bought BIB some six years ago between $36-$46. it's now @ $72.  Had I bought NASDAQ instead:

 

 

BIB vs NASDAQ.PNG

0

Share this post


Link to post
Share on other sites

Seems big auto is pulling the strings against American companies Waymo, Tesla and Intel...

https://www.japantimes.co.jp/news/2020/06/26/business/tech/countries-agree-regulations-automated-driving/

 

Not surprisingly, this is being drafted by Japan (Toyota, Mitsubishi, Mazda,...) and Germany (Mercedes, BMW, Audi, VW).

This basically reduces ADAS to nothing, as it cripples it to the point when it becomes useless.

0

Share this post


Link to post
Share on other sites

I welcome some form of regulation on auto-driving. If tesla ( or other companies ) want to self drive in the EU - Japan etc, I guess that they have to show the system is safe before it can be deployed.- and no Muck putting out an over the air update until its shown to work

  

0

Share this post


Link to post
Share on other sites
On 6/27/2020, 6:32:03, yesterday said:

I welcome some form of regulation on auto-driving. If tesla ( or other companies ) want to self drive in the EU - Japan etc, I guess that they have to show the system is safe before it can be deployed.- and no Muck putting out an over the air update until its shown to work

  

Sure, just don't make it so stupid that it completely prevents ANY type of autopilot! This is what this legislation wants to.

BTW, there is already regulation in EU and it is too strict, to the point that it is dangerous. It limits the autopilot overtake feature in a way that it makes it more dangerous than without any restriction. This is well documented and evidence has been presented to the EU commission.

 

And this is the problem: some idiots in an office have no clue how these systems work and they just issue stupid regulations.

0

Share this post


Link to post
Share on other sites
1 hour ago, MikeMelga said:

Sure, just don't make it so stupid that it completely prevents ANY type of autopilot! This is what this legislation wants to.

BTW, there is already regulation in EU and it is too strict, to the point that it is dangerous. It limits the autopilot overtake feature in a way that it makes it more dangerous than without any restriction. This is well documented and evidence has been presented to the EU commission.

 

And this is the problem: some idiots in an office have no clue how these systems work and they just issue stupid regulations.

I think that we will soon see infrastructure changes to support autopilot.  Why do I need a autopilot features to interpret signs when I can just make signs that contain bluetooth or the like that can communicate directly with the car?  

 

I also think that the systems should be proven to be safER, not "safe".  Driving a 2 ton vehicle at autobahn speeds is an inherently dangerous activity.  You are never going to be 100% "safe".  Why would we even expect that given what a crappy job humans do of driving. 

0

Share this post


Link to post
Share on other sites
35 minutes ago, BradinBayern said:

I think that we will soon see infrastructure changes to support autopilot.  Why do I need a autopilot features to interpret signs when I can just make signs that contain bluetooth or the like that can communicate directly with the car?  

IMO that is the wrong approach. Not only it makes it sensible to spoofing, this will take 15 years to achieve good coverage. Also useless in case of road works, accidents, etc.

Tesla's new approach is extremely interesting:

  • each car on the fleet is always identifying signals, etc. Example: a stop sign was added to a specific street. Car acts upon, although confidence level might not be high, as it was not expecting a stop sign there
  • all high level data is sent to a main server. In this case, it sends to the server that a potential new stop sign is in that street
  • main server combines data from multiple cars and immediately sends back HD map updates to the fleet
  • next car that passes there not only confirms the data, but can already be more confident on it

This is about to be deployed NOW! Not in 15 years! that is why your approach is wrong, because it is not needed and it's crazy expensive.

TBH this is pure genius:

  • they avoid paying HD map companies. They use their own fleet
  • they update the map several times per day, per street, if required
  • the algorithm used to generate the map is the same used to drive it
  • Eventually they can even monetize this and sell this as a real-time HD map to other companies
1

Share this post


Link to post
Share on other sites
2 minutes ago, MikeMelga said:

IMO that is the wrong approach. Not only it makes it sensible to spoofing, this will take 15 years to achieve good coverage. Also useless in case of road works, accidents, etc.

Tesla's new approach is extremely interesting:

  • each car on the fleet is always identifying signals, etc. Example: a stop sign was added to a specific street. Car acts upon, although confidence level might not be high, as it was not expecting a stop sign there
  • all high level data is sent to a main server. In this case, it sends to the server that a potential new stop sign is in that street
  • main server combines data from multiple cars and immediately sends back HD map updates to the fleet
  • next car that passes there not only confirms the data, but can already be more confident on it

This is about to be deployed NOW! Not in 15 years! that is why your approach is wrong, because it is not needed and it's crazy expensive.

TBH this is pure genius:

  • they avoid paying HD map companies. They use their own fleet
  • they update the map several times per day, per street, if required
  • the algorithm used to generate the map is the same used to drive it
  • Eventually they can even monetize this and sell this as a real-time HD map to other companies

What if the sign is covered by snow?  What if he road is covered by leaves?  Does the autopilot take are of all of that?  Remember, it only takes a few accidents to reduce the confidence of the public so that it will not be accepted.  Why would it be anymore succeptible to spoofing than say, air travel or shipping which already depends on signaling?  There is nothing inherently "spoofable" for autopilot than any number of other technologies.  That is a technical problem that is easier to solve than the Burger King "spoof" that was already mentioned.  I can photocopy a stop sign and set it up on my street right now.  Tesla would recognize it as a stop sign and stop.  Spoof complete.  The question is, why would I want to?  What advantage would I gain?  Spoofing a digital signal (which would include some sort of encryption) would be much harder. Again, why would I want to? I think arguments about spoofing are a red herring.  The only thing that I have read has said that companies need to be aware of the risk- not that it is something that is unsolvable.

 

Why would a new stop sign that includes a signal not be added to the network of information provided to drivers by the Transportation Department (or whomever is responsible)?  Why do I need a private company to do it for me?  What if I am not driving a Tesla?  How would Tesla updating their network help me?

 

Costs?  How much does it cost now to add autopilot to each car?  6k? per car? 46 million car times 6000 euros would be what - 276 billion? That would pay for a lot of signs for just Germany alone. 

 

Granted that you can implement Tesla's solution now, but only if you can afford a Tesla.  It is just a stepping stone to a truly smart transportation network. 

0

Share this post


Link to post
Share on other sites
24 minutes ago, BradinBayern said:

What if the sign is covered by snow?  What if he road is covered by leaves?  Does the autopilot take are of all of that?

  1. Same problem with a normal driver
  2. HD maps will alert AP that a sign should be there
  3. AP actually detects things like wet road and adds it to the adaptive speed algorithm, so yes, leaves can be taken in consideration

 

24 minutes ago, BradinBayern said:

Why would it be anymore succeptible to spoofing than say, air travel or shipping which already depends on signaling?

Air travel is actually hard to spoof. Same for shipping.

 

24 minutes ago, BradinBayern said:

  There is nothing inherently "spoofable" for autopilot than any number of other technologies.  That is a technical problem that is easier to solve than the Burger King "spoof" that was already mentioned.  I can photocopy a stop sign and set it up on my street right now.

Sure, and also fool human drivers and get yourself arrested.

 

24 minutes ago, BradinBayern said:

 The question is, why would I want to?  What advantage would I gain?  Spoofing a digital signal (which would include some sort of encryption) would be much harder. Again, why would I want to? I think arguments about spoofing are a red herring.  The only thing that I have read has said that companies need to be aware of the risk- not that it is something that is unsolvable.

Ask the russians why they feed this anti-vaxx war. KGB tactics is to generate confusion in western world.

 

24 minutes ago, BradinBayern said:

Why would a new stop sign that includes a signal not be added to the network of information provided to drivers by the Transportation Department (or whomever is responsible)? 

A person holding a stop sign? A construction worker placing a stop sign just before starting to interrupt the road?

 

24 minutes ago, BradinBayern said:

Why do I need a private company to do it for me?  What if I am not driving a Tesla?  How would Tesla updating their network help me?

That's the point, Tesla might sell these maps to others.

 

24 minutes ago, BradinBayern said:

 

Costs?  How much does it cost now to add autopilot to each car?  6k? per car? 46 million car times 6000 euros would be what - 276 billion? That would pay for a lot of signs for just Germany alone. 

$750 for the computer, perhaps $200 for the sensors. But that's not the point: even if all signs are encoded, you STILL NEED THE COMPUTER! How do you see people without it??? Bad argument.

 

24 minutes ago, BradinBayern said:

Granted that you can implement Tesla's solution now, but only if you can afford a Tesla.  It is just a stepping stone to a truly smart transportation network. 

They are about to announce a $20.000 - $25.000 car.

0

Share this post


Link to post
Share on other sites
6 minutes ago, MikeMelga said:
  1. Same problem with a normal driver
  2. HD maps will alert AP that a sign should be there
  3. AP actually detects things like wet road and adds it to the adaptive speed algorithm, so yes, leaves can be taken in consideration

 

1. We were talking about a new sign

2. A signal would not be blocked by snow

3. A central system would not depend on sensors in each car to work correctly.

8 minutes ago, MikeMelga said:

Air travel is actually hard to spoof. Same for shipping.

So why would this be any different?

9 minutes ago, MikeMelga said:

Sure, and also fool human drivers and get yourself arrested.

 

But I wouldn't get arrested for spoofing a signal?

10 minutes ago, MikeMelga said:

Ask the russians why they feed this anti-vaxx war. KGB tactics is to generate confusion in western world.

 

Sounds very "conspiracy theory" to me.  Why would the Russians want to spoof my local stop sign instead of crashing airplanes?  Screwing with dams and other major infrastructure like nuclear power plants?  Crashing Wall Street? Taking out Twitter (at least for Donald Trump)?  Have we no counter measures to deal with the Russians? Are we helpless against them?

16 minutes ago, MikeMelga said:

A person holding a stop sign? A construction worker placing a stop sign just before starting to interrupt the road?

 

How would a smart sign not work here too?  You can have portable smart signs with batteries.

18 minutes ago, MikeMelga said:

That's the point, Tesla might sell these maps to others.

 

Good for Tesla, not so good for consumers paying for what should be a public good.

19 minutes ago, MikeMelga said:

$750 for the computer, perhaps $200 for the sensors. But that's not the point: even if all signs are encoded, you STILL NEED THE COMPUTER! How do you see people without it??? Bad argument.

 

Where do you get those numbers?  Who cares what Tesla is paying to build their system, they aren't going to sell it a price.  Right now, that price is 6k.

Of course you need a computer.  Every modern car already has one.  Of course there would be some additional costs for a simplified auto pilot, but you can't really argue that the Tesla system is cheap but a very simplified version of the same thing would suddenly be prohibitively expensive.  You would not need to "see" a sign nor interpret the image if you have  sign telling you where it is at and what it is.  Therefore the systems would be greatly simplified. 

 

How much would a smart sign cost?  I honestly do not know, but AI traffic signals are already going up and have proven to be effective. 

28 minutes ago, MikeMelga said:

They are about to announce a $20.000 - $25.000 car.

 

Tesla is not known for keeping timelines.  I will believe it when I see it.  Not to mention I would be willing to bet that that cheaper car will not include autopilot, and if it does, it will not be in that price range. 

 

I seems to me that you are arguing that in the future everyone will drive a Tesla.  That doesn't seem realistic to me at all. 

0

Share this post


Link to post
Share on other sites
1 hour ago, BradinBayern said:

What if the sign is covered by snow?  What if he road is covered by leaves?

 

what if a very tall and fat person is standing in front of the sign?  What if a truck carrying pickles turns over and spills pickles all over the road?  Then what?

I think you raise some important concerns that Mike just glosses over.  I think we should take your stop sign/bluetooth idea even further and include all street signs...one-way, no parking, etc.  We could power each sign with solar (brought to you by you-know-who) and force the technology to wherever in the world teslas are sold.  Tesla Cars would become minuscule compared to The Tesla Sign Corporation.

Or, we could just change "auto-pilot" to "driver-assist" since that's what it is.

0

Share this post


Link to post
Share on other sites
2 hours ago, BradinBayern said:

1. We were talking about a new sign

2. A signal would not be blocked by snow

It's simple: if the sign is covered by snow, a normal driver won't see it. If the sign is old, AP will have it on the map. In any case, always better than the human driver.

 

Quote

3. A central system would not depend on sensors in each car to work correctly.

huh?

 

Quote

So why would this be any different?

Many reasons, I will name a few: amount of transmission power, angle of spoofing, distance of spoofing...

 

Quote

But I wouldn't get arrested for spoofing a signal?

Of course, but it is nearly impossible to get you. You can just drop a box in the grass and activate it later or with a timer.

 

Quote

Sounds very "conspiracy theory" to me.  Why would the Russians want to spoof my local stop sign instead of crashing airplanes?  Screwing with dams and other major infrastructure like nuclear power plants?  Crashing Wall Street? Taking out Twitter (at least for Donald Trump)?  Have we no counter measures to deal with the Russians? Are we helpless against them?

Simple: they actually can do that, but bringing down a plane (they actually did it once) is something that draws too much attention, as only a handful of countries can and will do that. Spoofing could be done without getting caught.

 

Quote

How would a smart sign not work here too?  You can have portable smart signs with batteries.

So if there is something wrong with the batteries, the cars would not stop. Sorry, very bad system. Cameras on cars can pick up signs and they keep getting better, no need to complicate what is simple!

 

Quote

Good for Tesla, not so good for consumers paying for what should be a public good.

Did the public invest in a million cars with cameras? No.

 

Quote

Where do you get those numbers?  Who cares what Tesla is paying to build their system, they aren't going to sell it a price.  Right now, that price is 6k.

You are mixing COST and PRICE! And you are mixing SOFTWARE with HARDWARE! Jebus!

6k is PRICE. 6k is SOFTWARE!

The numbers I mentioned are public number from Tesla on the HW3 computer COST! And all Teslas come with one. And if you had to put it on 46 million cars, it would cut the price to less than a third, i.e. meaningless cost per car.

 

Quote

Of course you need a computer.  Every modern car already has one.  Of course there would be some additional costs for a simplified auto pilot, but you can't really argue that the Tesla system is cheap but a very simplified version of the same thing would suddenly be prohibitively expensive.  You would not need to "see" a sign nor interpret the image if you have  sign telling you where it is at and what it is.  Therefore the systems would be greatly simplified. 

You don't understand. A camera based system is the most simple system you will have. It is absolutely mandatory to have what Tesla have for any ADAS system, including the LIDAR from Waymo. Cameras are essential!

 

Quote

 

How much would a smart sign cost?  I honestly do not know, but AI traffic signals are already going up and have proven to be effective. 

Why make it so complicated?? This approach WORKS already. And your solution is incorrect, you don't put the emitter on each sign. What you need is a 5G dense network every on few meters of road, for triangulation and HD map. And you would still need cameras!

 

Quote

Tesla is not known for keeping timelines.  I will believe it when I see it.

Agree, they started production of Model Y 8 months ahead of schedule. They started production of the Chinese factory 3 months ahead of schedule.

The rest is what the media focus on.

 

Quote

  Not to mention I would be willing to bet that that cheaper car will not include autopilot, and if it does, it will not be in that price range. 

They will include, that has already been announced and TBH it is one more thing that shows you are clueless about this topic. Tesla does not care if they spend $1000 in the computer and sensors, because:

  • Normal autopilot (no FSD) is a huge selling point, as well as Sentry Mode
  • Computer and sensors are necessary for mandatory pedestrian collision avoidance. Hence, you HAVE to install it.
  • FSD can be purchased at any moment
  • They are about to announce a monthly subscription of FSD. This will be a cash cow!
  • The car can't be operated without it. They would need to waste millions developing a cheaper version
  • They get huge amount of data even if the user does not use the computer

 

Quote

 

I seems to me that you are arguing that in the future everyone will drive a Tesla.  That doesn't seem realistic to me at all. 

I am arguing what the pros in the business know: Tesla has an innovative approach that is the most likely to succeed. Others will have to copy it or license it.

While other brands will be waiting for those smart roads, Tesla, Waymo and Mobileye will already be driving cars.

 

BTW, Mobileye has the same camera based approach, but they lack the millions of cars to constantly upload data. And their customers, like Mercedes, don't allow them to update the software on the fly several times per month, as Tesla do. Which means it will take them many years to achieve the same data volume.

 

Waymo has very advanced machine learning, but they have even less access to data than Mobileye.

1

Share this post


Link to post
Share on other sites
12 hours ago, catjones said:

Or, we could just change "auto-pilot" to "driver-assist" since that's what it is.

 

This has always been my understanding. The driver still has to "drive" the car and can't take a nap, send texts, etc. 

0

Share this post


Link to post
Share on other sites

J.D. Power Qualitätsreport 2020

sorry to say ( for the Tesla fanboys ), Tesla, has taken last place.

 

https://www.auto-motor-und-sport.de/verkehr/j-d-power-qualitaetsranking-2020-tesla-auf-dem-letzten-platz/  

 

https://www.theverge.com/2020/6/25/21302804/tesla-ranks-last-on-influential-jd-power-quality-survey

 

Seems when you buy a Tesla, you just do not get a car, but also alot problems as well, 70,000 euro auto.

 

Still according to some buy a new battery and and new computer and all the problems will be fixed

 

 

0

Share this post


Link to post
Share on other sites
1 hour ago, yesterday said:

J.D. Power Qualitätsreport 2020

sorry to say ( for the Tesla fanboys ), Tesla, has taken last place.

 

https://www.auto-motor-und-sport.de/verkehr/j-d-power-qualitaetsranking-2020-tesla-auf-dem-letzten-platz/  

 

https://www.theverge.com/2020/6/25/21302804/tesla-ranks-last-on-influential-jd-power-quality-survey

 

Seems when you buy a Tesla, you just do not get a car, but also alot problems as well, 70,000 euro auto.

 

Still according to some buy a new battery and and new computer and all the problems will be fixed

 

 

Not so fast, that study has several issues into it. Here is a debunking. Main point is what defines a defect. Example:

they classify quality control issues on the same level as design issues. People complain about the paint quality from Tesla (and I also do) but that is a design issue created by California´s laws regarding car painting are very unique, they only allow water based paints, which means you will always get a crappy paint that you cannot use on a normal car wash. Same with lack of Android Auto.

 

 

 

0

Share this post


Link to post
Share on other sites
2 hours ago, yesterday said:

Tesla, has taken last place.

 

This is one reason I don't own a Tesla, but I do own TSLA.

1

Share this post


Link to post
Share on other sites

Tesla stocks high due to potential Q2 profit, which would open the doors for S&P index, which by itself will value the stock some 20-30% more.

Still, I think situation in US is going really bad and it is possible that Fremont closes again in a few weeks, dragging stock down.

In any case, I think US stocks will go down in a few weeks.

1

Share this post


Link to post
Share on other sites

You guys do not get it, Tesla makes perfect cars and all the "problems" are FUD.

 

Remember I mentioned misaligned trunk and misaligned panels some pages ago?  This is what happens when some panels are not aligned:

 

 

 

Tesla is not a car company, it will take them some time to reach decent QA.   The good thing is they are getting better.

0

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!


Register a new account

Sign in

Already have an account? Sign in here.


Sign In Now