Elon tweeted that v9 of the FSD beta would remove its reliance on radar completely and instead determine decisions based purely on vision. Humans don’t have radar after all, so it seems like a logical solution and tells us Tesla is feeling much more confident in their vision AI.
Radar and vision each have their advantages, but radar has thus far been much more reliable in detecting objects and determining speed. If you’ve ever noticed your Tesla being able to detect two vehicles in front of you when you can only see the one directly ahead of you, that’s radar at work.
In this situation the radio waves from the radar sensor are bouncing underneath the car in front of you and are able to continue traveling and detect that there is another object ahead even though it could never “see” it.
It really is one of those wow moments where you can feel the future and the ability for AI-powered cars to drive better than humans one day. It’s baby steps and slowly we’ll see more and more of these situations where the vehicle simply sees or does something we could never do.
There’s no doubting that more sensors could provide a more reliable and accurate interpretation of the real world as they each have their own advantages. In an ideal world a vehicle with radar, lidar, vision, ultrasonic sensors and even audio processing would provide the best solution. However, more sensors and systems come at a price, resulting in increased vehicle cost and system complexity.
After all humans are relatively safe drivers with two “cameras” and vision alone. If Tesla can completely solve vision, they’ll easily be able to achieve superhuman driving capabilities. Teslas have eight cameras, facing in all directions. They’re able to analyze all of them concurrently and make much more accurate interpretations then we ever could in the same amount of time.
Tristan on Twitter recently had some great insight into Tesla vision AI and how they’re going to replace radar. Here’s what Tristan had to say:
"We recently got some insight into how Tesla is going to replace radar in the recent firmware updates + some nifty ML model techniques
From the binaries we can see that they've added velocity and acceleration outputs. These predictions in addition to the existing xyz outputs give much of the same information that radar traditionally provides
(distance + velocity + acceleration).
For autosteer on city streets, you need to know the velocity and acceleration of cars in all directions but radar is only pointing forward. If it's accurate enough to make a left turn, radar is probably unnecessary for the most part.
How can a neural network figure out velocity and acceleration from static images you ask?
They can't!
They've recently switched to something that appears to be styled on an Recurrent Neural Network.
Net structure is unknown (LSTM?) but they're providing the net with a queue of the 15 most recent hidden states. Seems quite a bit easier to train than normal RNNs which need to learn to encode historical data and can have issues like vanishing gradients for longer time windows.
The velocity and acceleration predictions is new, by giving the last 15 frames (~1s) of data I'd expect you can train a highly accurate net to predict velocity + acceleration based off of the learned time series.
They've already been using these queue based RNNs with the normal position nets for a few months presumably to improve stability of the predictions.
This matches with the recent public statements from Tesla about new models training on video instead of static images.
To evaluate the performance compared to radar, I bet Tesla has run some feature importance techniques on the models and radar importance has probably dropped quite a bit with the new nets. See tools like https://captum.ai for more info.
I still think that radar is going to stick around for quite a while for highway usage since the current camera performance in rain and snow isn't great.
NoA often disables in mild rain. City streets might behave better since the relative rain speed is lower.
One other nifty trick they've recently added is a task to rectify the images before feeding them into the neural nets.
This is a common in classical CV applications so surprised it only popped up in the last couple of months.
This makes a lot of sense since it means that the nets don't need to learn the lens distortion. It also likely makes it a lot easier for the nets to correlate objects across multiple cameras since the movement is now much more linear.
For more background on LSTMs (Long Short-Term Memory) see https://towardsdatascience.com/illustrated-guide-to-lstms-and-gru-s-a-step-by-step-explanation-44e9eb85bf21
They're tricky to train because they need to encode history which is fed into future runs. The more times you pass the state, the more the earlier frames is diluted hence "vanishing gradients".
Tesla’s FSD beta v9 will be a big improvement forward from what FSD beta users have been using where the system was still relying on radar. And it’ll be an even bigger leap from what non-beta testers currently have access to. We can’t wait. Now where’s that button?
Almost ready with FSD Beta V9.0. Step change improvement is massive, especially for weird corner cases & bad weather. Pure vision, no radar.
Tesla has pioneered the use of single-piece castings for the front and rear sections of their vehicles, thanks to its innovative Gigapress process. Many automakers are now following suit, as this approach allows the crash structure to be integrated directly into the casting.
This makes the castings not only safer but also easier to manufacture in a single step, reducing costs and improving repairability. For example, replacing the entire rear frame of a Cybertruck is estimated to cost under $10,000 USD, with most of the expense coming from labor, according to estimates shared on X after high-speed rear collisions.
These insights come from Sandy Munro’s interview (posted below) with Lars Moravy, Tesla’s VP of Vehicle Engineering, highlighting how these advancements contribute to the improvements in Tesla’s latest vehicles, including the New Model Y.
However, with the new Model Y, Tesla has decided to go a different route and eliminated the front gigacast.
No Front Casting
Tesla’s factories aren’t equipped to produce both front and rear castings for the Model Y. Only Giga Texas and Giga Berlin used structural battery packs, but these were quickly phased out due to the underwhelming performance of the first-generation 4680 battery.
Tesla has gone back to building a common body across the globe, increasing part interchangeability and reducing supply chain complexity across the four factories that produce the Model Y. They’ve instead improved and reduced the number of unique parts up front to help simplify assembly and repair.
There is still potential for Tesla to switch back to using a front and rear casting - especially with their innovative unboxed assembly method. However, that will also require Tesla to begin using a structural battery pack again, which could potentially happen in the future with new battery technology.
Rear Casting Improvements
The rear casting has been completely redesigned, shedding 7 kg (15.4 lbs) and cutting machining time in half. Originally weighing around 67 kg (147 lbs), the new casting is now approximately 60 kg (132 lbs).
This 15% weight reduction improves both vehicle dynamics and range while also increasing the rear structure’s stiffness, reducing body flex during maneuvers.
Tesla leveraged its in-house fluid dynamics software to optimize the design, resulting in castings that resemble organic structures in some areas and flowing river patterns in others. Additionally, manufacturing efficiency has dramatically improved—the casting process, which originally took 180 seconds per part, has been reduced to just 75 seconds, a nearly 60% time reduction per unit.
Advancements in die-casting machines and cooling systems have allowed @Tesla to dramatically reduce cycle times and improve dimensional stability. pic.twitter.com/WB5ji67rvV
Tesla’s new casting method incorporates conformal cooling, which cools the die directly within the gigapress. Tesla has been refining the die-casting machines and collaborating with manufacturers to improve the gigapress process.
In 2023, Tesla patented a thermal control unit for the casting process. This system uses real-time temperature analysis and precise mixing of metal streams to optimize casting quality. SETI Park, which covers Tesla’s manufacturing patents on X, offers a great series for those interested in learning more.
The new system allows Tesla to control the flow of cooling liquid, precisely directing water to different parts of the die, cooling them at varying rates. This enables faster material flow and quicker cooling, improving both dimensional stability and the speed of removing the part from the press for the next stage.
With these new process improvements, Tesla now rolls out a new Model Y at Giga Berlin, Giga Texas, and Fremont every 43 seconds—an astounding achievement in auto manufacturing. Meanwhile, Giga Shanghai operates two Model Y lines, delivering a completed vehicle every 35 seconds.
Having the ability to open your trunk hands-free can be incredibly useful when your hands are full, especially in a busy parking lot.
Tesla vehicles now support opening the vehicle’s trunk or frunk completely hands-free — no foot waving required.
What is Hands-Free Frunk and Trunk?
Tesla implemented its hands-free feature by leveraging your phone’s position in relation to the vehicle. When you stand still behind your vehicle, the trunk will automagically open for you.
While this functionality isn’t available on every vehicle, it’s available on every vehicle Tesla manufactures today, including the new Model Y, the Cybertruck and other recent models.
With a compatible device and a supported vehicle, you can now open your Tesla’s trunk hands-free.
How It Works
Tesla’s hands-free feature requires the use of ultrawide-band (UWB) in the vehicle and on your phone. Apple and Samsung have supported ultra-wideband for a number of years and most flagship Android devices also support the low-energy feature.
Ultra-wideband allows another device to precisely detect its relative location. In this case, the vehicle is tracking where the driver’s phone is in relation to the vehicle. Since the vehicle is able to more precisely track the phone’s location, ultra-wideband also improves Tesla’s phone key feature.
Since the vehicle depends on your phone, you’ll need to have your phone on you in order to activate the hands-free feature. Simply stand within 2.5 to 3 feet from the front or rear of your vehicle for the frunk or trunk to open. You’ll then hear a couple of chimes. If you continue to stand still, then your frunk or trunk will open automatically.
The chimes serve as a warning that the trunk will open if you don’t move, which helps reduce accidental openings.
Hands-Free Trunk in Action
The video below shows how Tesla’s hands-free trunk feature works.
Supported Models
Since Tesla uses ultra-wideband to power the hands-free feature, only vehicles with the needed hardware are supported. The list of supported vehicles includes:
2021 Model S and later
2021 Model X and later
2024 Model 3 (Highland) and later
2026 Model Y (Juniper) and later
All Cybertrucks
Supported Phones
Your phone will also need to support UWB. Luckily, most manufacturers have included UWB in their devices for several years.
Apple: All Apple devices since the iPhone 11 have included UWB, except for the iPhone SE (2nd and 3rd generation). The iPhone 16e also has UWB.
Android: Most Android phones - especially flagship devices - already support and use UWB for other uses, but it’s not available on all phones. If you have a Google Pixel 6 or higher, Samsung Fold 2 or higher, Samsung S21+, or other recent Android phone, then your phone already supports ultra wideband.
Which Models Support Hands-Free Frunk
Unfortunately, not every supported model supports the hands-free frunk and trunk feature. The hands-free frunk feature is only supported on the Model S, Model X, and the Cybertruck. In addition, the Cyebrtruck is the only vehicle with a powered frunk, so while the Model S and Model X will unlock the frunk for you, you’ll still need to lift it and close it manually. The Cybertruck will open the frunk for you, much like the trunk on another Tesla.
Which Models Support Hands-Free Trunk
While most supported Tesla vehicles can use the hands-free trunk, it excludes the Cybertruck, which doesn’t have a powered trunk.
Enable Hands-Free Trunk / Frunk
If you plan to use your vehicle’s hands-free trunk feature, you’ll need to enable it in settings, as it’s off by default. Simply open Controls by tapping the vehicle icon in the bottom left corner, then navigate to the Locks section.
Within the Hands-Free section, you’ll find a few options, depending on your model. You’ll be able to choose whether to enable the hands-free frunk or trunk and whether you’d like to disable the feature at home.
Preventing Accidental Opening - Exclude Home
Although the hands-free feature requires you to stay still in front or behind your vehicle for a couple of seconds, it can still be triggered accidentally if you’re working around your garage. To prevent accidental opening of the frunk or trunk, Tesla allows you to disable the feature while your vehicle is parked at home.
Tesla determines your home location by the address that’s set in your vehicle. However, it also adds a buffer, meaning that your hands-free trunk feature will also not work in your driveway or at your neighbor’s home. The exclude home feature is located in the same spot as other hands-free trunk features, Controls > Locks > Hands-Free > Exclude Home.
If you have a recent Tesla that’s supported, go ahead and give the feature a try.
![Tesla Adds HEPA Filter to New Model 3, Will it Get Bioweapon Defense Mode? [Update: Tesla Edits Post]](https://www.notateslaapp.com/images/news/2023/new-model-3-hepa-filter_300w.jpg)
