News
Tesla is patenting a clever way to train Autopilot with augmented camera images
Tesla is currently tackling what could only be described as its biggest challenge to date. In his Master Plan, Part Deux, CEO Elon Musk envisioned a fleet of zero-emissions vehicles that are capable of driving on their own. Tesla has made steps towards this goal with improvements and refinements to its Autopilot and Full Self-Driving suites, but a lot of work remains to be done.
As noted by Tesla during its Autonomy Day presentation last year, attaining Full Self-Driving is largely a matter of training the neural networks used by the company. Tesla adopts what could be described as a somewhat organic approach for autonomy, with the company using a system that is centered on cameras and artificial intelligence — the equivalent of a human primarily using the eyes and brain to drive.
Tesla’s camera-centric approach may be quite controversial due to Elon Musk’s strong stance against LiDAR, but it is gaining ground, with other autonomous vehicle companies such as MobilEye developing FSD systems that rely primarily on visual data and a trained neural network. This approach does come with its challenges, as training neural networks requires tons of data. Tesla emphasized this point as much during its Autonomy Day presentation.
With this in mind, it is pertinent for the electric car maker to train its neural networks in a way that is as efficient as possible with zero compromises. To help accomplish this, Tesla seems to be looking into the utilization of augmented data, as described in a recently published patent titled “Systems and Methods for Training Machine Models with Augmented Data.”
Teslas are equipped with a suite of cameras that provide 360-degree visual coverage for the vehicle. In the patent’s description, Tesla noted that images used for neural network training are usually captured by various sensors, which, at times, have different characteristics. An example of this may lie in a Tesla’s three forward-facing cameras, each of which has a different field of view and range as the other two.
Tesla’s recent patent describes a system that allows the company to process these images in an optimized manner. Part of how this is done is through augmentation, which opens the doors to flexible and widespread neural network training, even when it involves vehicles equipped with differently-specced cameras. The electric car maker describes this process as such:
“Augmentation may provide generalization and greater robustness to the model prediction, particularly when images are clouded, occluded, or otherwise do not provide clear views of the detectable objects. These approaches may be particularly useful for object detection and in autonomous vehicles. This approach may also be beneficial for other situations in which the same camera configurations may be deployed to many devices. Since these devices may have a consistent set of sensors in a consistent orientation, the training data may be collected with a given configuration, a model may be trained with augmented data from the collected training data, and the trained model may be deployed to devices having the same configuration.”
Among the most notable aspects of Tesla’s recent patent is the use of “cutouts,” which allow Tesla’s neural networks to be trained using an optimized set of images. This was something that was discussed by former Tesla Autopilot engineer Eshak Mir in a Third Row Podcast interview, where he hinted at a system adopted in the electric car maker’s ongoing Autopilot rewrite that helped lay out “all the camera images” from a vehicle “into one view.” Such a process has the potential to help Tesla with 3D labeling, especially since the images used for neural network training are stitched together. Tesla’s patent seems to reference a system that is very similar to that described by the former Autopilot engineer.
“As a further example, the images may be augmented with a“cutout” function that removes a portion of the original image. The removed portion of the image may then be replaced with other image content, such as a specified color, blur, noise, or from another image. The number, size, region, and replacement content for cutouts may be varied and may be based on the label of the image (e.g., the region of interest in the image, or a bounding box for an object).”
Tesla is aiming to release a feature-complete version of its Full Self-Driving suite as soon as possible. Elon Musk remains optimistic about this, despite the company missing its initial timeline that was set at the end of 2019. That being said, Elon Musk did mention previously that Tesla is working on a foundational rewrite of Autopilot. In a tweet early last month, Musk stated that an essential part of the rewrite involves work on Autopilot’s core foundation code and 3D labeling. Once done, the CEO indicated that additional functionalities could be rolled out quickly. This recent patent, if any, seems to give a glimpse at how these improvements are being done.
Tesla Full Self-Driving v14.3.3 driver monitoring was reportedly scaled back in recent releases, but a new version that was released in the early hours of June 3 aimed to do a better job of keeping those in control of their cars honest, according to release notes.
The release notes for FSD v14.3.3, via Software Version 2026.14.6.7 added:
“Improved driver monitoring system sensitivity with better eye gaze tracking, eye wear handling, and higher accuracy in variable lighting conditions.”
However, Tesla said this was already enabled in the first rollout of FSD v14.3.3 in late May. We tested it anyway, especially as the Standard Speed Profile seemed less-than-worried about what you were doing during operation.
I decided to try out the Hurry and Mad Max Speed Profiles for this test, and it gave me results that I would have expected. Tesla has evidently ramped up driver monitoring based on the Speed Profile you are using to travel.
The more aggressive the Speed Profile, the more on the hook you will be for taking your attention away from the road. Our testing showed that Mad Max was less likely to allow you to do normal things like change music or adjust navigation without getting an on-screen warning or nag from the driver monitoring system.
Hurry Mode Results
On Hurry, the driver monitoring system on FSD v14.3.3, via Software Version 2026.14.6.7, was more restrictive than Standard but less restrictive than Mad Max. I found that I could scroll through music options for a considerable amount of time, more than 30 seconds:
Roughly :31 between first touching the center screen and getting the first nag
— TESLARATI (@Teslarati) June 3, 2026
Standard gave me about 80 seconds of phone scrolling with absolutely no nags or warnings in a previous test. It is worth noting that this was a previous branch of v14.3.3, but Standard is such a goodie-two-shoes on the road that it is my impression it would not change much.
Here’s an 80-second phone nag test on Tesla FSD v14.3.3.
No alerts, no nagging, no annoyance. https://t.co/1dxvTOw5Cn pic.twitter.com/vYViFpjfoK
— TESLARATI (@Teslarati) May 29, 2026
Mad Max Results
I spent the majority of the drive on Mad Max to see how it truly reacted to the driver having their attention elsewhere. While I did do a short phone test, I am aiming to steer away from those and use the center screen. I think it is a valid criticism that the phone test is dangerous and, not to mention, illegal in Pennsylvania. Changing the navigation and music is a more reasonable, more responsible, and safer test.
With Mad Max being the fastest and most aggressive Speed Profile, I anticipated this being the quickest mode to give me an alert that I needed to look at the road. That was the case with music:
🎥 Testing Tesla FSD v14.3.3 (via 2026.14.6.7) nags on Mad Max https://t.co/qZALU2OujY pic.twitter.com/XddOJ0D47x
— TESLARATI (@Teslarati) June 3, 2026
As well as adjusting Navigation, when I received two nags:
🎥 Testing Tesla FSD v14.3.3 (via 2026.14.6.7) nag while adjusting navigation
Two nags here https://t.co/qZALU2OujY pic.twitter.com/xa3dtaDG1L
— TESLARATI (@Teslarati) June 3, 2026
These nags were more than reasonable, and I think it’s probably good that Tesla is ramping up the driver monitoring. I do believe that it should be relatively strict across all of the Speed Profiles, especially with phone use. When using the center screen, the nag intervals should be based on the speed profile you are utilizing at the time.
These driver monitoring adjustments are a great thing to have while FSD is still under its “Supervised” moniker, but I expect Tesla to continue pushing the limits on what it will allow, especially considering CEO Elon Musk has hinted that phone use is capable with the more recent versions.
You can watch the full drive on YouTube below:
Tesla has responded to the skeptics of its Robotaxi program by launching a massive expansion of the unsupervised program in its initial rollout city of Austin.
The company’s geofence, the enabled area of operation for rides, now covers the entire Austin Metropolitan area, an incredible move just days after media headlines attempted to discredit the ride-hailing service.
Those who have access to the Tesla Robotaxi app on their smartphones can now request a ride in any portion of the Austin Metro area. The company confirmed this on the social media platform X:
Unsupervised Robotaxi now in the entire Austin Metro area https://t.co/eXNBdarvVS
— Tesla Robotaxi (@robotaxi) June 3, 2026
This is Tesla’s fifth expansion of the geofence, with the others occurring in July, early August, late August, and late October 2025. It has remained at that size since October 26, but Tesla has now more than doubled that size.
It is now covering the entire area, including suburbs like Pflugerville and Manor, as well as I-35 highways, Gigafactory Texas, and the Austin-Bergstrom Airport.
The move comes just days after various media outlets highlighted the small fleet size of Tesla’s Robotaxi fleet in Austin, something that is a reasonable criticism but an understandable move on the company’s part to prioritize safety.
Tesla has expanded its Robotaxi geofence many times, but its fleet has remained at a relatively conservative size as the company continues to push safety as its most crucial metric.
The latest expansion is a key indicator of Tesla’s comfort level to expand the ride-hailing service. The move shows Tesla is scaling unsupervised autonomy, as it demonstrates that the company’s Full Self-Driving system has reached sufficient reliability for a broader real-world deployment, which is something the company has worked on extensively.
It also shows Tesla is game for a competition with its rivals in the autonomous ride-hailing sector. Tesla has often matched or exceeded competitors like Waymo in coverage area, despite its smaller fleet. This step highlights Tesla’s iterative, data-driven progress toward a high-margin, app-based Robotaxi network.
It’s not the absolute largest area expansion ever, but achieving full unsupervised operations across a major metro is a key moment in the Robotaxi story. It shifts the program from limited pilot/testing toward a more mature commercial service, while gathering the miles needed for faster growth.
Tesla has improved Dashcam playback with an awesome new addition, as the company has launched a web-based version that is potentially easier to navigate and operate.
The tool is available at dashcam.tesla.com and will be enabled as your vehicle receives the 2026.20 Software Version. Clips that are captured by your Tesla will be available on the Online Dashcam Clip Viewer once the files on your car’s storage drive are encrypted.
Not a Tesla App first noticed the new feature, and states that once your Tesla updates to 2026.20, the car will automatically protect the clips with an encryption key that is uniquely tied to your owner account.
Tesla Launches New Web-Based Dashcam Viewer https://t.co/AlJKXYxujJ pic.twitter.com/4igicYpvkX
— Not a Tesla App (@NotATeslaApp) June 2, 2026
The web-based viewer should be easier to operate for most. All you will do is head over to dashcam.tesla.com and log in using your account credentials.
Ensure your vehicle is updated to 2026.20 in order for the web-based viewer tool to fetch your vehicle’s saved dashcam clips.
Currently, only a small percentage of owners are updated to this, so it may be a couple of weeks until a majority of owners in the fleet are able to access this feature.
Watching Dashcam clips on the Tesla smartphone app is quick and convenient, as they can also be easily downloaded and stored right on your smartphone.
However, the clips are sometimes tougher to navigate, and in order to get details like self-driving activation, speed, and turn signals, owners have to screen record the Tesla app and crop out the rest of the screen.
It could also be a massive storage saver as you’ll be able to download the Dashcam clips from the online viewer and save them to your laptop, desktop, a flash drive, or even an external hard drive. This will keep all your clips in one place.
Tesla Full Self-Driving v14.3.3 driver monitoring: We tested it
Tesla responds to Robotaxi skeptics with a massive move in Austin
