Search results

That's not true for large cap stocks. Take AAPL as an example, it was growing about 50% YoY between 2009 and 2012 and in the same time its P/E was in the teens-20s-30s.

Where's that bolt located? I'll probably give it a try.

The same thing just started happening to me out of nowhere. 2020 MS LRP Curiously it started right after the latest FSD software update (2022.4.5.21). The first time it got stuck, a reboot fixed it, but the next day it got stuck again and the reboot didn't work anymore. I have a service appt in...

Also, where do I set the text that appears under my post? There are few options, like custom title, profile posts, signature, about you, etc... I tried a few ways without success.

Thank you! Do you know where I could find information on supported models?

See my previous post. I didn't imply anything, only that it is not true that having FSD 10.4 automatically gives you live camera view and vision based parking.

There are either multiple sub-versions of FSD Beta 10.4 or for some reason my 2020 Model S doesn’t support these features.

I could post my own set of images here to prove you wrong if I was at my car now. I triple checked it yesterday. I have the same software version but the live camera view and perpendicular vision based parking are both missing. They are not in the release notes either.

I have the most up to date version of the app and the FSD software. There is no settings to enable live camera view. I watched YouTube videos to see where the settings are supposed to be but that setting is missing for me.

I can confirm that 2021.36.8.5 FSD Beta 10.4 does NOT have live camera view available. I’m sitting in the car and the setting to enable it on the Safety & Security screen is missing. I also doubt that it has the vision based perpendicular parking because I tried to use it but it never offered me...

Also, with FSD you lose the latest non-FSD goodies, like remote camera view from the app, vision based parking, waypoints. None of which I found in the FSD software. In daily driving, on highways it’s pretty much the same old NoA, however on city streets you become a full-time baby sitter...

The only way is to look at the map.

I'm sure many people mentioned it here, but I don't have time to read all the comments, is that FSD only works when there is good GPS lock on the car. I park in a garage and overnight the GPS location often drifts away by up to a couple hundred feet, and when I drive out of the garage the car...

Turns and lane changes are different. Here we were mostly discussing lane changes for getting into faster lanes. Those could be made optional (with confirmation), like in NoA. Although the car often goes into the wrong lane even in preparation for turns too, which could also be improved by human...

This not true. I’ve tried to disable FSD automatic lane changes many times in Navigate on Autopilot settings but that had no effect on FSD. FSD automatic lane changes cannot be turned off.

Suggestions to Tesla FSD team (will also email it to them): There should be a user selectable mode in which each lane change needs to be confirmed by the driver, just like in Navigate on Autopilot. This would help two ways: 1) Drives would be less stressful, not having to be always ready for...

10.4 installing now, not expecting any major improvements. FSD is still a long way from wide release. Many people say how well it works on their drives, but most of the praise come from people who don't drive in busy cities. I do live in a middle size city's downtown and FSD is virtually...

Definitely true that the Model S doesn't feel like a real luxury car once you compare it to the Mercedes-Benz S-class or the Audi A8. Those have ridiculously high quality compared to a Tesla and are full of bling. However, compared to the Model S, they also feel very dated. The Model S feels...

What you call Tesla Vision actually uses a lot more sensors than just cameras. It also uses GPS, inertial sensors and mechanical odometry (wheel orientation and position), but possibly even more. If they found a better way to use existing radar, it would also use the radar. Actually it might...

Exactly! It said the same last year. And I assume before that and before that. What I'd be fine with is a publicly released and safer version of what is now the FSD beta.

Using IR or X-ray cameras for industrial computer vision is very common. If interpreting IR or X-ray images doesn't count as vision then we'll have to DQ a lot of systems. In fact you don't even need to detect photons directly to generate images that need to be processed. For example the image...

Not everyone lives on Tesla social media. I only started reading Tesla forums since I bought my car. I bought my Tesla without knowing much about Elon Musk. What convinced me to get FSD was the showroom representative (service advisor) who claimed (what the website also said) that there will be...

HDR has nothing to do with colors. Vision of color blind people still have several magnitudes wider dynamic range than most cameras (SDR). HDR is useful when the important details one wants to resolve on the image are either very bright or very dark compared to the rest of the image. For example...

Direct quote from Karpathy: "And of course, you could go into the radar stack and you could adjust the hyper parameters of the tracker, like why is it dropping tracks and so on, but then you are spending engineering efforts and focus on a stack that is not really barking up the right tree."...

Just to repeat myself. At no point did Karpathy say that it would be impossible for radar+vision to work well with each other. He said that the version of the radar stack that they have is not good enough to do it, but they could make it better by doing a proper hyper parameter sweep. However...

I'm not sure what you mean by this. In computer vision the term stitching is typically used for methods that combine semi-overlapping images of (spatial) regions of the same scene. Creating panoramas from multiple images is an example. However here you're referring to temporal stitching...

Working with neural networks is different from developing traditional algorithms and software. Based on what information I saw about their AI, it is indeed reasonable to expect that a small team can very efficiently manage it. Remember, there is very little coding done for neural networks...

He also said that it would have been possible to fix the issues with radar and do proper sensor fusion but they decided they'd rather focus resources on full vision.

I heard that Volvo's highway autopilot is very similar to Tesla's, with good automatic steering and OK adaptive cruise control. I was told it even works on certain roads in cities similarly to FSD, for example it stops at red lights. It requires you to have your hand on the steering wheel, just...

There are certain plastics that are transparent to IR light but are opaque to visible light. For example the Apple TV 4K, which looks completely black from outside is actually transparent to IR light and that's how it is able to sense IR signals from any direction. So the IR LED can be sitting...

Franky, I'd be fine if they released FSD in a not quite-full version, where it drives on most city streets safely, but it will make you take over when the AI is not certain it can make the right decision, like in the above mentioned unprotected left turns, cops directing traffic, etc. The...

As other have already said, it's not only the shape of the yoke is the problem but that important controls are placed on it. It is only intuitive to find the turn signal and horn buttons when you are holding the yoke at 9-and-3 and it's not rotated more than 90 degrees. When you turn the yoke...

Knight Industries already tested the yoke on the Two Thousand and Three Thousand models for several years and their test driver, Michael, had only good things to say about it.

I would think moving windows around are handled by the GPU and the CPU has very little to do with that. The GPU is plenty fast for that. The slowness of the Atom should be most felt during heavy computations and I/O intensive tasks. Like loading a webpage or starting applications.

This is a little misleading. Porsche didn't voluntarily reduce their EPA estimate with the 70% reduction factor. They only reduced it for the 2020 Turbo S from 200 to 192 (by 8 miles). They also reduced the the rating for the 2020 Taycan Turbo, but only by 1 mile. They still admit that their...

Tesla's auto lane change is pretty amazing. This is the thing that impresses me the most in the current public FSD package. I've been using it for a while and it's very safe. It happens all the time that other cars do unexpected, stupid or illegal things while I'm trying to do a lane change but...

And one update to the range debate: NAF EV Range Test The Norwegian Automotive Federation did a 21-car test of EV range. They compared their measurements to the European WLTP rating. The only Tesla being tested (2021 Model 3 LR) exceeded its rating by 6.7% (654.9 km tested, 614 km rated) in...

You are suggesting that Ford and other makers deliberately reduce their EPA estimates so that it will match up better with people's real world experience. There's almost zero chance a profit driven corporation would do that. I'm sure Porsche would kill to get a better EPA estimate, but the...

I can just repeat myself. It's EPA's rating and you can't blame Tesla for it, unless they cheat like VW did, which no one alleged yet. It's a completely different question whether the EPA test matches someone's real-world driving pattern/style. All ICE cars are also marketed based on EPA ratings...

Also, Edmunds itself admits that its estimates are not more accurate than the EPA's. They are just different.

Again, don't blame Tesla, blame the EPA. Tesla does not come up with the EPA number. It's the EPA that tests the car and determines the estimate. You may say that EPA's estimate is not as realistic as Edmund's, but that is a very different from saying that Tesla doesn't meet its EPA estimate.

A reminder: It's not Tesla's estimate but the EPA's. The only way Tesla can manipulate it if they do what VW did and their cars operate differently when they detect that they are being tested by the EPA.

The big difference is that humans have a much broader range of recognition capabilities than current digital neural networks, which are trained to do only a few things. One simple example is recognizing previously unseen obstacles on the road. Tesla trains its networks to recognize about a 100...

To be fair, the 400 miles EPA estimate is for the mixed cycle, not for 80 mph highway driving. Show me a single car (ICE or electric) that meets its EPA estimate on the highway at 80 mph.

Catching up on all the conversation... So we heard a lot of anecdotal "evidence" of how smart Elon is. I don't doubt he's smart. As I understand, he even studied some undergrad level physics. I'm sure he's very good as making decisions based on all the information and options that are presented...

Do you really think Elon designed all the products his companies make? Elon Musk is not an engineer but a businessman so we forgive him for not knowing how sensor fusion works. Steve Jobs also had no idea how hardware worked and wasn't a programmer either, yet he did pretty well.

I don't see it phantom braking as a huge problem. Frankly, I don't know why people are complaining about it. In most cases radar helps identify hazard situations when vision isn't yet able to recognize them. Like vision wouldn't be able to detect that you are driving towards a featureless wall...

This is a very naive view of sensor fusion. It's not a "which one do you choose" problem. Elon tweeted this as well and it immediately reminded me of the old guy in congress who described the Internet as a series of tubes. Sensor fusion doesn't do binary picks. It's getting simultaneous inputs...

Yes, just because the Mach-e is a Ford it doesn't mean it will suddenly have the same ease of servicing as an Escape. It's a completely different kind of beast from ICE Fords. When I went to the Lincoln dealership to test drive the Mach-e none of the employees knew anything about the car, only...

This is mostly true, but there indeed are NN based solutions for processing temporal data. And Tesla is actually using them. It doesn't mean they are processing actual video data with NNs, though. They aggregate data from still images then convert those to another format of lower dimensionality...