Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

The Information reports new stuff about Autopilot and autonomy

This site may earn commission on affiliate links.
Yes, but maybe he is thinking what I am thinking. I am not willing to enter an email address into another site just to read articles and then not know how else they are going to abuse my email address. Possibly more spam. I will just forego reading the articles :)

It doesn't use verification of the email address.

It could use some other verification means, but it never asks the emailed user to verify it.

So just enter in an email address of someone who spams you. The first article isn't worth bothering with, but the second one about 200 employees is a good article.
 
It is really very disturbing to me that Tesla has only recently started using simulation to test. How do they test their Autopilot code prior to release to ensure that it's safe in the billions of different situations it might in encounter out in the real world? Every other AV company's answer to this is (at least in part) rigorous testing in simulation. Tesla on the other hand, I'm sure does some testing on a test track, then releases to the early access program testers, then calls it ready for wide release. Staggering.

As for HD maps... it sounds like they're going in the direction of geo-fencing for FSD to me. Just like all those other AV companies that Tesla fans like to criticize for being geo-fenced...

And this is why i keep trying to tell @S4WRXTTCS that it matters whose right or wrong.
If we don't keep track then no one would know wtf was going on and ppl would bs their way in every discussion.
But with track records, someone whose every post is based on logic will always be right and someone whose posts are based on fantasy and myths will always be wrong.

Example: simply take a look at my past posts, i have been bulls-eye all the way through.
Then you take a look at others, for example Jimmy who is worshiped yet, was all goo-goo gaga over the AP2 nvidia chip and vowed it was more than enough to power level 5 FSD.

Now he is doing the same to the new AP3 chip and no one remembers what he said in the past. @S4WRXTTCS wants there to be no track record so guys like @jimmy_d can carry on this charade unfettered. But If you actually look at jimmy past posts, you will see how often he's been wrong, dozens of times, including his anti mobil-eye statements. He said Tesla was ahead of everyone combined and that there were no evidence to the contrary. His agenda is quite clear.

Likewise every single article @strangecosmos has written on seeking alpha has been dead wrong. He was one of the main guys who shoved the insane 2018/2019 FSD timeline down our throats. Every point he has laid out has been flat out wrong. Starting from his first articles in early 2017, "Tesla has immense lead in SDC", to his "Tesla Leapfrogs Self-Driving Competitors With Radar That's Better Than Lidar".

Radar better than lidar = wrong.
Tesla fleet learning will lead it to FSD in 2019 = wrong.
Tesla shadow mode will lead it to FSD in 2019 = wrong.
Tesla has collected of billions of miles which will lead it to FSD in 2019 = wrong.
Tesla will launch FSD in 2019 and Tesla network, other companies system slated for 2020 and 2021 are just vaporware = wrong.
Tesla has a current HD Map Moat, No competitor can do this. = wrong.

Turns out Tesla just started building infrastructure for HD maps, which is in its infancy. Its even proven by the fact that wk057 couldn't build an HD map from the NN lane output because they were so inconsistent compared to AP1.

I could go on and on, this is fringe logic at its finest.
I tried to warn Trent, but he just won't listen. He proclaimed Tesla FSD by 2018-2019. Now Look how it turned out.

Yes being right matters!
 
Last edited:
I consider the computer as part of the software. Obviously they each need to be designed to work with the other. I remember some interesting work done by the University of London,I believe, back in the 60’s. I don’t know what happened to it, but they were trying to duplicate brain cells and connections with transistors. The initial work was promising, almost surprising. What happened to it I don’t know.

The latest on that line of investigation is the SpiNNaker project at University of Manchester. They actually just last month announced the second version, which has 1 million cores, each theoretically capable of simulating 1000 neurons, so that would be 1 billion neurons, or roughly 1% of a human brain. It would not fit in a Tesla and would drain the battery before it had a chance to boot up.

SpiNNaker - Wikipedia

So they need to increase capacity by 2 orders of magnitude while dropping size and power consumption by another two orders of magnitude, at least before they could put a human-level brain inside a car. I say "at least" because I don't actually know the power consumption figures on this, but it's safe to say it's in the hundreds of kilowatts or more.

Of course, they probably don't need human-level intelligence to drive a car on most roads with no human supervision. But I bet they need more than HW3 can handle, perhaps by an order of magnitude. This would all be fine if Elon hadn't promised (and sold) FSD in 2017, because of course eventually the technology will get there.

And then they have the lack of redundancy problem, and no sensor cleaning, etc. Dead horse.
 
I don't care. I want Tesla to succeed at FSD even if they do get there later than other companies do.


There is no if. Why wont you just accept it? The writing is on the wall. Elon is simply trying to deceive the naive to sell more cars and pump the stock. As the autopilot engineer said "The system won’t be able to handle general city driving for years, this person estimated."

dont you think the person who writes the system daily actually knows what hes working on and their internal timeline which has nothing to do with Elon's gibberish.

Tesla will probably add something like stopping for traffic light / stop sign, taking right turns next year. It will be half baked and Elon will claim they now have FSD, just like he claims they now have full highway autonomy with NOA.
 
There is no if. Why wont you just accept it? The writing is on the wall. Elon is simply trying to deceive the naive to sell more cars and pump the stock. As the autopilot engineer said "The system won’t be able to handle general city driving for years, this person estimated."
That does not say, if some other player does it earlier.
 
Yes being right matters!
The-Right-Natural-Cure-For-Excessive-Sweating.jpg
 
There is no if. Why wont you just accept it? The writing is on the wall.

The shorts said the same thing about Model 3 production. "Look at the production problems. Tesla is losing money. The writing is on the wall. Tesla will fail." And look at Tesla now. Model 3 production has skyrocketed and Tesla has achieved a profitable quarter.

I don't "just accept it" because my own experience tells me differently. With every update, my autopilot has gotten incrementally closer to FSD. AP has made huge progress over the past couple of years. That's undeniable. And, the AP engineer you quote says that they will get to general city self-driving in X years. So FSD will happen. it will just take longer than Musk thinks. But that's not breaking news to anyone.

I don't care about Musk making crazy timeline predictions. I know his predictions are too optimistic. I care about my car getting new features and getting better AP over time which I am getting.

Tesla will probably add something like stopping for traffic light / stop sign, taking right turns next year. It will be half baked and Elon will claim they now have FSD, just like he claims they now have full highway autonomy with NOA.

I think you are too pessimistic and jaded. NOA is not full highway autonomy now but it will get there. Just as auto steer was pretty shaky when it was first released for AP2 but it is really solid now. So when Tesla does release traffic light/stop sign and making intersection turns, yes, it will probably have some issues at first, just like any new software release, but it too will get better over time. All new AP features have issues in the beginning but they improve with time. It's how software works.

I guess I am just a glass half full kinda guy. When Tesla adds stop signs and traffic lights to EAP/FSD, I will be happy that my car has a new feature that makes it better than before rather than complaining that it's not good enough.
 
  • Like
  • Love
Reactions: jimmy_d and skitown
I am not sure people appreciate how difficult an Autonomous Program is.

It's easy to boast how cheap and how good a "current" LIDAR is but the fact is we are nowhere near that kind of level: cheap and good!

These cheap and good LIDARs are only good for low speed and short range detection as in city driving.

When you talk about Autopilot cruising at 90 MPH, that's another whole different ball game!

The range of $75,000 Velodyne HDL-64 E is about 120 meters or less than 400 feet or about 1 football field length.

Newer Velodyne VLS-128 is much better with the range of 300 meters or more than 900 feet or almost 3 football field lengths.

Even with that much of a range, its speed limit is 70 MPH!

$75,000 is for old slow speed. You'll need much more money and beyond the cost of Velodyne VLS-128 if you want to match the speed of Tesla 90 MPH system.
.
I can appreciate how complicated such a program must be, that’s probably why Elon’s recurring BS is frustrating to me.

As for LIDARs, I’m out of my depth so might say something wrong. But it feels like this would be additive to other systems (not instead of), and more important in high-density/high-traffic/complicated areas, such as cities, where 90mph is not quite required. VLS-128 is said to be sold at $12k in volume. So not $200 for this Rolls, but getting in range.
 
And this is why i keep trying to tell @S4WRXTTCS that it matters whose right or wrong.
If we don't keep track then no one would know wtf was going on and ppl would bs their way in every discussion.
But with track records, someone whose every post is based on logic will always be right and someone whose posts are based on fantasy and myths will always be wrong.

You clearly misunderstood what I was getting at.

We don't have perfect information coming out from Tesla so we can't know exactly what they're doing. We don't know when they EXACTLY started simulation, and how it's involved over the years.

One of my biggest complaints about the Tesla AP is the lack of transparency from Tesla on something we pre-paid for.

So we're left to largely speculate. We can use our backgrounds to speculate on everything we've heard. We can also use what we observe in our own cars. This is especially important when it comes to how a person might like/hate any given feature.

Of course this speculation leads to biases.

I'm a hardware engineer so you'll see me saying "No, that can't work because it doesn't have enough sensors". I say that because I want a lot of hardware redundancy.

Then there is Jimmy who is an NN guy, and he's naturally going to biased in favor of NN solutions.

Regular Owners are naturally biased towards some level of optimism. You can see this in how many likes a positive speculative post will get about some AP behavior.

Going back there were quite a few people who had pretty solid predictions about FSD being way far away, and EAP being way more difficult to implement than Elon was letting on.

I myself felt like FSD was the second biggest con job I'd ever seen in my life. I felt like the world had lost it's marbles back in 2016.

Imagine if all those people who called it on EAP/FSD kept spouting on about how right they were?

Wouldn't that be really annoying?

Then there is also the idea that the whole AP program got reset with Karpathy. I myself didn't buy an AP2 car until I got a Model P3D+ a few months ago., and I got EAP+FSD with it.

I did so with the feeling like Tesla was finally starting to get traction with AP on HW2/HW2.5. Plus they were starting to offer compelling features outside of that. Things like track mode, and dash cam. Where I felt like Tesla was finally getting out of this black hole that was recovering from the split with MobileEye, and betting the company on the Model 3.

I can't speak for other owners, but I think a fair amount of us learn who to trust on here. Where we get an overall vibe of who to ask. Knowing who to ask is extremely important in a place where there is so much contradictory information.

When it comes to hardware I'll ask verygreen or lunitiks. There used to be ingineer, but I don't know what became of them. I also trust WK057, but he hasn't been around much lately.

When it comes to NN's I still think Jimmy is the one to ask. He's just trying to do his bet to interpret what's on the car. If I worked at Tesla I'd probably purposely mess with him where I put fake neural networks on it.

When it comes to historical stuff regarding the history of AP I think MP3Mike has been the most accurate.

Do any of those people talk about how right they are? No, why would they?

Trump claims he's right all the time. But, is Trump right all the time? Are you really sure you want to use a tag line like what Trump would use? This isn't political as most republicans I know would never claim they're right all the time.
 
when Bladerskb joins the thread

MilkyCorruptAngwantibo-max-1mb.gif


make sure to fact check Bladerskb’s claims because a lot of them are just complete fabrications. :rolleyes: I can’t tell if Bladerskb deliberately lies to get a reaction, or genuinely doesn’t know what’s real and what isn’t. maybe they suffer from creative recollection. but regardless of conscious intent this is troll behaviour & it happens in pretty much every thread in this forum.

I implore you: please either ignore Bladerskb or move these repetitive arguments to a different thread when they arise, to avoid derailing threads. I have Bladerskb on ignore but every thread fills up with replies to them, going over the same talking points again and again and again and again

you can’t have a reasonable conversation with someone who is fixated on proving they’re smarter than you. a reasonable conversation requires good faith

no hate to Blader. I’m sure this disruptive behaviour comes from a place of suffering. but that doesn’t mean we should allow it to continue to disrupt this forum
 
Last edited:
when Bladerskb joins the thread

MilkyCorruptAngwantibo-max-1mb.gif

I have Bladerskb on ignore but every thread fills up with replies to them, going over the same talking points again and again and again and again.

you can’t have a reasonable conversation with someone who is fixated on proving they’re smarter than you. a reasonable conversation requires good faith
Gee, I was wondering the same thing in the past, right before I put that one on ignore. It is interesting that the chosen name closely matches "Blatherskite" which is defined in the dictionary as "a person who talks at great length without making much sense."
 
Last edited:
It is really very disturbing to me that Tesla has only recently started using simulation to test. How do they test their Autopilot code prior to release to ensure that it's safe in the billions of different situations it might in encounter out in the real world? Every other AV company's answer to this is (at least in part) rigorous testing in simulation. Tesla on the other hand, I'm sure does some testing on a test track, then releases to the early access program testers, then calls it ready for wide release. Staggering.

What do Anti Virus company simulations have to do with Tesla?
 
  • Funny
Reactions: cwerdna
One of my biggest complaints about the Tesla AP is the lack of transparency from Tesla on something we pre-paid for.

So we're left to largely speculate. We can use our backgrounds to speculate on everything we've heard. We can also use what we observe in our own cars. This is especially important when it comes to how a person might like/hate any given feature.

Of course this speculation leads to biases.
Yep on the bolded part. One can also receive information/hints based upon the efforts of what others have achieved and reported in the autonomous vehicle space (e.g. CA DMV disengagement reports, Cruise Automation's efforts in SF, Waymo's efforts all around the US, Waymo's self-driving taxi service in Phoenix, what AV companies demonstrate to the press (esp. if recorded on video), other self-driving shuttle/taxi services, etc.) along with what Tesla has chosen to do/not do or disclose or not.

Tesla's a company all about hype. If their capabilities were that great, wouldn't they be showing off demonstrated success like the above or even surpassing it? So far, we haven't seen any evidence of that and we have the laughable CA figures for Tesla (vs. Waymo and Cruise) at Autonomous Vehicle Disengagement Reports 2017, Autonomous Vehicle Disengagement Reports 2016 and Autonomous Vehicle Disengagement Reports 2015. The 550 miles in 2016 for the publicity stunt video with a horrendous number (150+) disengagements for some earlier runs isn't real good.

If Tesla could miraculously achieve disengagement rates comparable to (or even only 1/2 as good as) Waymo on CA public roads given the vast body of software engineering talent being in Silicon Valley, why wouldn't they do it and demonstrate it? What do they have to hide?

How well can anything Tesla has deployed to customers handle Why testing self-driving cars in SF is challenging but necessary or what's on pages 36 to 39 of https://storage.googleapis.com/sdc-prod/v1/safety-report/Safety Report 2018.pdf?

And, it's not like Waymo nor Cruise Automation are perfect and don't have issues. There have been plenty reported in the press. It just seems like Tesla isn't even close.

And, you've got Elon spouting off all these dates when he thinks Tesla will achieve something in the self-driving space but they turn out to be just missed dates. And realists who aren't hardcore fanboys and part of the "cult of Elon" find the predictions to be just BS.
 
Last edited:
when Bladerskb joins the thread

The problem with Blader is not that he criticizes Tesla or, as he's doing in this thread, pointing out how ridiculous it is to believe in anything Elon says about AP given his past track record on AP promises/predictions. Blader's only problem is his obsession with MobileEye, but personally I'm willing to simply account for that bias and otherwise give his comments the respect they deserve. He has been right about AP (nevermind MobileEye) for quite a long time... unlike anybody who has ever taken Elon at his word about anything having to do with AP.
 
I don't care. I want Tesla to succeed at FSD even if they do get there later than other companies do.

I do too but if this article is true, the EAP/FSD team has a very tough roe to hoe and Elon's timeline to get FSD rolled out next year is pure fantasy land. The article makes it apparent that engineering rigor is just not there and the fruit if that is apparent when using EAP.

I don't want to diminish their efforts and progress they've made, I've seen improvement just in the 3 months I've owned my car. But wow, think what they could do if they had their *sugar* together and had taken a more robust approach earlier.
 
  • Like
Reactions: caligula666
I do too but if this article is true, the EAP/FSD team has a very tough roe to hoe and Elon's timeline to get FSD rolled out next year is pure fantasy land. The article makes it apparent that engineering rigor is just not there and the fruit if that is apparent when using EAP.

I don't want to diminish their efforts and progress they've made, I've seen improvement just in the 3 months I've owned my car. But wow, think what they could do if they had their *sugar* together and had taken a more robust approach earlier.

Well, we've known for awhile now that the AP team had a really rough start to say the least ever since the MobilEye breakup. They tried to write their own in-house "software 1.0" to recreate what MobilEye had done with AP1 and that was a disaster. Then, Karpathy came in and started from scratch with his NN approach. They had like 2 or 3 false starts. But it looks like things are working much better now. Karpathy's work with NN seems to be paying off as we've seen from how good AP2 has become recently. And I would say that V9 and Nav on AP is proof that Tesla's current approach is finally working.

I have no doubt that if Tesla had done things right from the start, we would probably have FSD by now. I am just glad that they are on the right track now.
 
  • Disagree
Reactions: Bladerskb