Tesla, TSLA & the Investment World: the Perpetual Investors' Roundtable

[Singuy]

I don't agree to vision will be worst than vision plus radar but I do agree with the hardware/software not being ready hence gimpping them. No tech company in the world sells you hardware that's worst than previous hardware but promise it'll be the same or better one day. That one day needs to be the day the product goes on sales.

So this is like ditching hw1 for hw2 with a promise..and that too was a bad move for customers as well.

There are enough model 3s and Ys out where you can take volunteers for your vision only program to complete the validation. No need to inconvenience new buyers.

 

[Christine69420]

I'd be very surprised and disappointed if new, kickass colors are not introduced on the 10th. We're overdue.

That will happen when the new factories in Austin and Berlin start to make cars. They have new modern paint systems. The paint system in Fremont is not up to the task. Switching colors is too slow and cumbersome.

 

[mickificki]

I know fsd beta is right around the corner but I’m growing increasingly skeptical of its abilities. We’re on the latest version 2021.4.18 and on a road trip a couple days ago it’s trying to perform lane changes on cars that are speeding by next to us (cringe-worthy), and it can’t stay in the lanes when the car is on the fwy in an elongated S-shape. It’s not a severe transition, just a slow rolling snake like shape, but it crosses the line into the next lane. It’s been like this for over two years now and still feels like zero progress.

I’m praying fsd beta is worlds better but if it can’t perform a simple stay-in-the-lane-lines task...it’s beyond frustrating and doesn’t give my wife any confidence in using it.

I’m not sure how Tesla can get people to buy in to fsd if it’s still like this. I get that Elon said latest version is way better, I hope he’s right...

 

[FireMedic]

That is inherent in radar, which requires sending a signal, receiving a return, then processing the results of the returns. That takes time and huge amounts of signal processing.

In vision only systems the lags are far shorter than for radar, partly because signal ambiguity and lack of specificity are not major processing issues.

Both are valuable, but radar is ill-equipped to handle highly variable signal returns. Vision, OTOH, requires less processing because the signal is inherently visual to begin with.

In the meantime within a few weeks vision only will be superior to previous vision+radar.

It didn’t occur to me until I read this post; is it possible that there is a significant energy consumption involved with using radar combine with vision (that’s an honest question, I don’t know)?

After the autonomous vehicle demo I had yesterday(not waymo)…which seem to be a very power hungry suite of sensors, on a BEV platform, I was thinking how inefficient the system was.

Google/Waymo with their big brain engineers chose ICE vehicle platforms instead of EV (which I presume was not a cost based decision)

 

[EVMeister]

That will happen when the new factories in Austin and Berlin start to make cars. They have new modern paint systems. The paint system in Fremont is not up to the task. Switching colors is too slow and cumbersome.

That's true. Although they could still show them off perhaps? To garner some extra excitement for the Berlin and Austin vehicles. I'm not expecting this at the event but it would be cool to see.

 

[Buckminster]

Not really.
There's a shortage of radar (I used to fix radar in the Navy).

It is indicative of shenanigans when Tesla pulls the radar from their cheaper models (and gimps their autosteer top speed AND retcons their blogs to remove old explanations about the benefits of radar). If the radar was obsolete, it wouldn't still be implemented in S and X.

Radar-less cars are objectively less capable than radared cars.

We have to take Elon at his word that he will restore lost functionality in a couple weeks. As a gambling man, I don't take those odds.
Disagree all you like, we'll review this in July and see.

Who are you and what have you done to AF?
I suspect the long term bull is still in there somewhere. Perhaps the Superbull thread would suit you needs better. I am looking for someone to challenge or better my $125T 2030 projection. Bullish AF you might say.

 

[Freder Reeve]

There should be no debate left for radar. With exponential growth of units on the road using radar, the number of false positives becomes far too abundant. Apparently it will end the phantom braking issue, one early symptom of too much radar on the roads already for their old fsd path.

 

[JRP3]

I

I'm saying he's too susceptible to outside forces.

Answer the implied question:
If vision is ready for primetime, why do the S and X still ship with radar?

I don't have a good answer for that, but if it's not ready for primetime, or soon to be, why do the high volume 3 and Y ship with only vision? Do you really think Tesla wants to flood the market with an inferior version of self driving, less capable than what they already have?

 

[Lessmog]

There should be no debate left for radar. With exponential growth of units on the road using radar, the number of false positives becomes far too abundant. Apparently it will end the phantom braking issue, one early symptom of too much radar on the roads already for their old fsd path.

Plus, it seems to my myopic understanding that a proliferation of lidar with invisible but perhaps increasingly powerful IR lasers could result in growing damages to optical sensors (such as human eyes)? If so, not a clever development. A few might be OK, but several on every vehicle? No thanks. JMHO.

 

[Hock1]

That will happen when the new factories in Austin and Berlin start to make cars. They have new modern paint systems. The paint system in Fremont is not up to the task. Switching colors is too slow and cumbersome.

Are you saying that refreshed Model S and Plaid will be built in Austin and Germany? I have not heard anything about that.

 

[2daMoon]

Are you saying that refreshed Model S and Plaid will be built in Austin and Germany? I have not heard anything about that.

I think you are right about where those models are and will be manufactured, but there have been rumors that when Austin and Berlin are online, Fremont might get a paint shop upgrade.

 

[Bullpup]

I know fsd beta is right around the corner but I’m growing increasingly skeptical of its abilities. We’re on the latest version 2021.4.18 and on a road trip a couple days ago it’s trying to perform lane changes on cars that are speeding by next to us (cringe-worthy), and it can’t stay in the lanes when the car is on the fwy in an elongated S-shape. It’s not a severe transition, just a slow rolling snake like shape, but it crosses the line into the next lane. It’s been like this for over two years now and still feels like zero progress.

I’m praying fsd beta is worlds better but if it can’t perform a simple stay-in-the-lane-lines task...it’s beyond frustrating and doesn’t give my wife any confidence in using it.

I’m not sure how Tesla can get people to buy in to fsd if it’s still like this. I get that Elon said latest version is way better, I hope he’s right...

I don't mean to presume, but that doesn't sound like my experience at all except for about a month when I had an undiagnosed asymmetric tread-wear and alignment problem - AP seemed crazy, s-turning and ping-ponging until it eventually left the lane and I needed to intervene. A new set of tires, an alignment, and a autopilot recalibration and I was back to dead-steady in the center of the lane.

 

[Christine69420]

Are you saying that refreshed Model S and Plaid will be built in Austin and Germany? I have not heard anything about that.

Neither have I. I have just heard what 2daMoon just said.

 

[Hock1]

I think you are right about where those models are and will be manufactured, but there have been rumors that when Austin and Berlin are online, Fremont might get a paint shop upgrade.

That rumor would make sense then, wouldn't it. It doesn't do us Model S drivers any good if the 3s and Ys made in Austin and Germany get new colors. If Ss are to get new colors, looks like it's going to have to be in Fremont. I'm thinking the June 10 event would be the perfect time to introduce new colors. JMO

 

[MP3Mike]

I'm thinking the June 10 event would be the perfect time to introduce new colors. JMO

Not really. It is a delivery event, that isn't the time to introduce new colors/features that would make people refuse delivery.

 

[Right_Said_Fred]

I know fsd beta is right around the corner but I’m growing increasingly skeptical of its abilities. We’re on the latest version 2021.4.18 and on a road trip a couple days ago it’s trying to perform lane changes on cars that are speeding by next to us (cringe-worthy), and it can’t stay in the lanes when the car is on the fwy in an elongated S-shape. It’s not a severe transition, just a slow rolling snake like shape, but it crosses the line into the next lane. It’s been like this for over two years now and still feels like zero progress.

I’m praying fsd beta is worlds better but if it can’t perform a simple stay-in-the-lane-lines task...it’s beyond frustrating and doesn’t give my wife any confidence in using it.

I’m not sure how Tesla can get people to buy in to fsd if it’s still like this. I get that Elon said latest version is way better, I hope he’s right...

Are you on FSD beta? You say you're praying FSD beta is worlds better, yet you seem to suggest that in your car it can't perform a simple task. Are you talking about Autopilot perhaps? Those are two completely different things. The performance of Autopilot and FSD beta are not necessarily related.

 

[Kenypowa]

I know fsd beta is right around the corner but I’m growing increasingly skeptical of its abilities. We’re on the latest version 2021.4.18 and on a road trip a couple days ago it’s trying to perform lane changes on cars that are speeding by next to us (cringe-worthy), and it can’t stay in the lanes when the car is on the fwy in an elongated S-shape. It’s not a severe transition, just a slow rolling snake like shape, but it crosses the line into the next lane. It’s been like this for over two years now and still feels like zero progress.

I’m praying fsd beta is worlds better but if it can’t perform a simple stay-in-the-lane-lines task...it’s beyond frustrating and doesn’t give my wife any confidence in using it.

I’m not sure how Tesla can get people to buy in to fsd if it’s still like this. I get that Elon said latest version is way better, I hope he’s right...

I know this is based on my personal experience, but there was some discussion on Reddit about 2021.4.15.12 vs 2021.4.18. Last week I was upgraded to 2021.4.15.12, which apparently incorporated vision improvement from the FSD Beta tree, where as 20214.18 doesn't have the vision improvement.

I could tell the 4.15.12 is a much more competent release as the auto wiper worked flawlessly. AP also handled various roads very well. In fact I was quite satisfied about this build.

Then my car was pushed 4.18 a couple days ago. Right off the bat I would notice the performance was not up to par, and the autowiper reverted back to its old unreliable days. Lane keeping was also degraded as the car seemed confused at times. I drove on the same roads and the 4.15.12 build navigated just like human.

The vision improvement is real and substantial. I wish I didn't update to 4.18. Finger crossed we will get the new build containing the vision improvement.

At the same time, I haven't seen such uniform distribution where most cars are on the same latest build. Something big is definitely coming.

 

[jbcarioca]

It didn’t occur to me until I read this post; is it possible that there is a significant energy consumption involved with using radar combine with vision (that’s an honest question, I don’t know)?

After the autonomous vehicle demo I had yesterday(not waymo)…which seem to be a very power hungry suite of sensors, on a BEV platform, I was thinking how inefficient the system was.

Google/Waymo with their big brain engineers chose ICE vehicle platforms instead of EV (which I presume was not a cost based decision)

I do not know that for sure. I do know that the aircraft radar units I used were heavy energy consumers. Hopefully some of us might have expertise to know the answer regarding the Tesla radar implementation.

 

[jbcarioca]

Yes. Rain detector.

Once again maybe you could do some cursory homework;

https://www.weather.gov/media/publications/front/14dec-front.pdf

As the linked brief summary shows radar depends on reflectivity. Vision systems are perfectly capable of 'seeing' anything radar can 'see'. However radar is massively dependent on reflection strength, so a very distant hugely reflective object will have radar thing the object si closer than a much closer low reflectivity object. Resolving those fundamental inherent characteristics of radar is a gigantic math problem dependent entirely on indefinite inferences.

That basic issue is why vision systems have become more efficient and accurate than have radar. Lidar has another set of issues in object recognition and signal attenuation.

Each of the technologies is very valuable for certain things. Some, like lidar and radar, are most effective in places other than road-vehicle automation. Until recently vision-based systems were very expensive and not terribly accurate. Now technology has cured those problems and also become so cheap, reliable and accurate that the others have become obsolescent.

As usual, once people decide a given solution is "right" they are highly resistant to change their opinions.

Yet another reason why Tesla is so far ahead.

 

[Pezpunk]

Finally got the call Friday afternoon asking if I’d like to switch my Plaid+ to Plaid. I wonder if they started at the end of the line and worked backwards — my order was made the instant Plaid+ was announced during the reveal.

I declined to switch. I’m in no hurry.

the caller confirmed that production would not begin until mid-2022, and didn’t bother to ask me for a reason when I said I didn’t mind waiting — he just said “and the reason is the range?” And I said primarily, sure. I didn’t get the sense that many people were switching.

As a side note, he sounded kind of flustered / frustrated. I hope Tesla is giving these people adequate breaks and other tasking. Cold-calling is absolute soul-draining work. I mean that as a concern, not a complaint.