Elon Musk

AI Is taking Tesla’s Full-Service Driving System To The Next Level

As automakers continue gearing up for the future, artificial intelligence (AI) and self-driving modes are becoming an increased point of focus. One such automaker that has heavily touted its vehicles’ full-self driving (FSD) capabilities is Tesla. FSD is one of three autonomous driving programs they offer customers, with the other two being Autopilot and Enhanced Autopilot driver assistance.

According to a report from Autoweek, the latest update for FSD represents Tesla’s most advanced self-driving program to date.

What Is Full-Service Driving?

In its beta v12.1.2 version, FSD uses neural networks for traffic navigation and comes with what Tesla refers to as Traffic and Stop Sign Control and Autosteer on City Streets. The automaker highlighted what makes this latest version different from their previous FSD versions in their release notes, “FSD Beta v12 upgrades the city-streets driving stack to a single end-to-end neural network trained on millions of video clips, replacing over 300k lines of explicit C++ code.”

That means the machine learning technology Tesla is using to improve its FSD program gets training from viewing real-life driving footage instead of lines of code. It marks a new direction for the world’s most successful electric vehicle (EV) maker and a different approach than what the auto industry has usually used to train their autonomous vehicles so far. Usually, autonomous tech developers employ a combination of radar, lidar, and vision sensors. Tesla first announced they would use a vision-only approach for their autopilot functions in the fall of 2022.

What’s Tesla’s Long-Term Goal With FSD?

It will take some time before it becomes clear how extensive the improvements to this beta version will be, though this also raises questions about the automaker’s plan for FSD in the long run.

Are they hoping FSD will eventually have Society of Automotive Engineers (SAE) Level 3 autonomous driving capabilities? Under its most recent update, FSD still requires paying attention to what the vehicle is doing and for human drivers to be ready to take control of it at any moment. A car with SAE Level 3 autonomous driving capabilities can fully drive itself while the human takes a nap, watches videos, reads, or writes.

Of this potential development, Tesla said, “Full autonomy will be dependent on achieving reliability far in excess of human drivers as demonstrated by billions of miles of experience, as well as regulatory approval, which may take longer in some jurisdictions.” It’s important to note that the automaker did not specifically mention an SAE Level 3 system in their statement. Tesla has clarified to regulators that FSD will not surpass SAE Level 2 driving capabilities. If it did, the automaker would have some significant regulatory blocks that they would have to deal with.

The only vehicles that are SAE Level 3 certified with permission to be driven on roads in the United States right now are Mercedes S-Class sedans and EQS vehicles, and they can only operate their SAE Level 3 driving modes in California and Nevada under certain circumstances. So, if Tesla owners could sit back, relax, and enjoy a fully autonomous ride, the automaker would have much to explain to California regulators.

The automaker also found itself in hot water late last year over numerous accidents involving the application of their autopilot software that they eventually fixed with an over-the-air update. At the time of the update, Tesla said they would implement new car controls to limit what the autopilot feature was capable of and that if the vehicle senses the driver has taken their hands off the steering wheel, the driver will get reminders to have their hands on the controls at all times. Hopefully, the Tesla drivers using this latest beta v12.1.2 version of FSD will understand that their vehicle is not fully autonomous and always keep their hands on the steering wheel.

Author: Jarret Hendrickson

Similar Posts