Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register
  • Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
  • Tesla's Supercharger Team was recently laid off. We discuss what this means for the company on today's TMC Podcast streaming live at 1PM PDT. You can watch on X or on YouTube where you can participate in the live chat.

Considering getting the FSD package, not sure I'm thinking clearly

This site may earn commission on affiliate links.
I believe FSD as described on Tesla's website is 5+ years away. All the info I've seen leads me to believe Tesla has a long way to go:
-Still fleshing out machine learning/AI
-Autopilot is still a work in progress
-Current hardware most likely insufficient (need more radar sensors, LIDAR, better cameras, etc)
-Tesla doesn't appear to actually be testing FSD

Having said that...I'm seriously considering buying the FSD package.
-Price increase most likely coming (can't be extreme else Tesla will price themselves out of the market)
-I'm interested in experiencing the FSD features as they roll out. Trying something new/cutting edge is the main reason I bought the car.

So here I am, staring at my Tesla account page thinking about hitting that buy FSD button and weighing the pros and cons.

Rational side of me says don't spend $5k on something that doesn't exist, irrational side says FSD beta features will be fun.
 
I'll disagree with some of your timelines and premises. LIDAR and extra sensors aren't going to be needed. I can easily show you proof that a single, moveable stereoscopic camera is all that's needed. After all, that's all you are using.
It's also pretty obvious that most of the work with self driving isn't object recognition, it's figuring out what to do with it. If you look at some of the videos on "What Tesla's are seeing" it should be pretty obvious that they are identifying everything out there.
Also, don't look at just the full package, the incrementals are very powerful as well.

And if you purchase with the car, you can roll it into the financing.

And most importantly, you aren't buying a Model 3 to be sensible, you are getting it, to have fun.
 
I do not think that Tesla will deliver 100% FSD in the next couple of years. I did order and pay for FSD when I ordered my car. I believe they will be able to add a lot of FSD in the next couple of years. Things like driving major highways and city streets. Back country roads that I use a lot (not painted, sometimes gravel) probably not so much. Perhaps their Uber competition will work in big cities. I just do not expect them to do 100% FSD on every public road in the US (and yes gravel roads are public).
 
I'll disagree with some of your timelines and premises. LIDAR and extra sensors aren't going to be needed. I can easily show you proof that a single, moveable stereoscopic camera is all that's needed. After all, that's all you are using.

Two thoughts on this...

1) The CPU being used in your example is vastly more powerful than anything in a Tesla (including HW3) for complex problem solving, executive function, general decision making, etc...

2) We can move those cameras around, the car still has blind spots (especially low as you get close to things)

3) And even then... we honestly kind of suck at driving. And FSD isn't going to be a thing until the machine is much better than we are (IIRC Elon said at least twice as good)

None of that is to say they won't get there... but I think the idea of "8 cameras is all you need since humans get by with 2" isn't a terribly good argument.


It's also pretty obvious that most of the work with self driving isn't object recognition, it's figuring out what to do with it. If you look at some of the videos on "What Tesla's are seeing" it should be pretty obvious that they are identifying everything out there.

Agreed (except for the bits it still can't see at all)- but that's part of why I think true level 5 is a TON further off than Elon thinks at least.


Also, don't look at just the full package, the incrementals are very powerful as well.

Yup- that's why I grabbed the 3k price before delivery.

Even if all I get is level 3 or 4 on divided roads/highways that'd be a steal since that's 90% of my normal/daily driving.
 
Everyone knows full level 5 Autonomous FSD is highly unlikely in the near future.

3000-5000 price tag tells you everything.

When it is, add an extra zero or two.

It’s a cheap price for computer upgrades and incremental improvements over EAP.
 
  • Like
Reactions: Foxtrotter
If you can afford it and are planning on keeping your car for at least 3-4 years, then buy it. You’ll be as close to future-proofing your car as possible and for as cheap as possible, and if you’ll have the opportunity to experience all the new abilities as they roll out.
 
Two thoughts on this...
1) The CPU being used in your example is vastly more powerful than anything in a Tesla (including HW3) for complex problem solving, executive function, general decision making, etc...
The self-driving unit in Tesla is already more powerful than any CPU in existence (but not any GPU). Simply because it's designed to run neural networks FAST.

The HW3 might contain even more specialized hardware, like an ASIC layer for the first levels of the neural network.
 
The self-driving unit in Tesla is already more powerful than any CPU in existence (but not any GPU). Simply because it's designed to run neural networks FAST.

The HW3 might contain even more specialized hardware, like an ASIC layer for the first levels of the neural network.

So you are telling me, HW3 is going to be a bitcoin mining beast?
 
I’d probably be in my 4th or 5th Tesla (I am already on my 3rd since AP2) when FSD starts rolling out. I don’t care it’s 3k or 5k now as I won’t even have the car by then. Look at all those leasing AP2 cars since Oct 2016. Only one year left and they might never see FSD at all. Tesla will probably let them roll the FSD into their next [Tesla] car as a compensation.

Everyone working on autonomous vehicles have their cars on public roads for testing. Do you see any report on Tesla testing FSD? That probably indicates something.
 
I think Tesla has the best chance of getting FSD done, especially with their new proprietary hardware coming in 2019. If you look at Tesla's history, they've been hindered and sometimes royally screwed by 3rd party vendors. This dates back to their early days. So you notice a trend where they are wanting to do everything themselves by designing their own software and hardware. This sets Tesla apart from others. HW3 is going to be a beast and from earnings call you can tell how confident they are at going head on to compete with Uber & Lyft.

Autopilot at the moment is obviously limited by the current hardware. It's good enough for Navigate on Auto and doing onramp/offramp, and it has industry best lane keeping and autosteer. They are definitely on track for good things. If you can afford FSD it'll probably be a good idea, because I do think HW3 will cost a lot more when it's unleashed. Plus let's face it, we're driving a Model 3 because we want to be early adopters and drive a spaceship.

I think Tesla owners are a bit jaded because we're used to the tech by now. But try showing EAP to people who have never been in a Tesla, their jaws are dropped the entire time. Sometimes we forget just how cool and advanced our cars are heh.
 
Whether you buy it now comes down to a philosophical question. Are you more excited about the finished product, or are you more excited about the journey getting there?

This is an excellent way to look at it. I purchased FSD with the same thought in mind. Just like I look forward to software general updates now at some point I believe Tesla will begin sending out features that only owners of FSD will be able to take advantage of. Will it be fully self driving to begin with, of course not, but the journey will be worth it however rocky.
 
To be honest one can make a good argument either way.

I can only tell you why I bought it. I as in a person who absolutely hated the FSD option because of the lack of any real promise or regulatory path. I've never in my life seen a reputable company sell vaporware to that degree.

So yes I bought it despite detesting it with every bone in my body.

Why in the world would I do that?

Because self-driving is going to be incremental.

We still have another 6-9 months of EAP improvements without needing the FSD hardware (the new computer).

To gain anything beyond that point in time you're going to need the new computer.

It likely won't be more than $5K at that time, but it's an unknown.

What we do know today is if you buy the FSD option then you get that upgrade.

I felt comfortable getting it because I could see what I was getting. I was getting the better computer (when it's released), and it's going to be capable of much more complex and accurate neural nets.

My prediction is the EAP functionality will benefit from the new computer. Now it won't require it, but it will benefit from it.

What I'm really curious about is what Tesla is going to do when the new computer is released. My prediction is all new Tesla's at the time will have it. For me this makes it doubly important to get it as I want to match up with what's on new Tesla's at the time.
 
I wonder if Tesla is going to piece out the hardware depending on if you get EAP or EAP and FSD.

EAP and safety suite uses one set of hardware and FSD used another?

My rationale is some kinds of compute gets really expensive. Compare a G4400 Celeron - CPU and integrated graphics - 50 dollars.

Ryzen 2700 + 1080TI - $1000.

Not actually installing hardware for FSD could save Tesla money unless it’s ordered and paid for.

None of those 35,000 Model 3s are going to take on 10,000 worth of software.