Is Tesla's “Full Self-Driving” Feature Actually Completely Autonomous?
Oct 14, 2021 · 4 min read
For the past few years, self-driving cars have been a top priority of automakers.
Cruise control quickly became adaptive cruise control, which uses laser technology to pinpoint the locations of surrounding cars. However, even the most advanced cruise control can be defective, so drivers still need to keep their hands on the wheel.
Tesla's latest version of "Full-Self Driving" is even more advanced than its current offering. Is it finally completely autonomous? Here's what we know from CNET.
How will Tesla’s "Full-Self Driving" tech work?
While Tesla's "Full-Self Driving" isn't quite ready, engineers had a lot to show off at Tesla's AI day.
Instead of using radar technology like cruise control, "Full-Self Driving" relies on eight cameras to find its way around. The system uses the data from those cameras to create a virtual model of its surroundings called the "Vector Space."
The "Neural Net Planner," the AI component of "Full-Self Driving," uses the "Vector Space" while routing the trip.
As the car drives itself to its destination, the AI is constantly working to predict the actions of other drivers and pedestrians. In one demonstration, the AI was able to yield and eventually pass a car with the consent of an oncoming driver.
Tesla assures us that the "Vector Space" is highly accurate thanks to Tesla's new environmental recognition software. Its short-term memory can remember each vehicle at an intersection even when the cameras are obscured. The software is also still functional during severe weather.
Tesla also shared some details about the program's development process. The company has its own studio for AI training which optimizes the software's behavior. Its sensor system continues to receive updates to recognize millions of objects in its vicinity.
Though some drivers are concerned about autonomous vehicles, it's important to note that despite the name, "Full-Self Driving" still isn't completely autonomous. The driver still has to apply light pressure to the wheel for the program to operate.
Will Tesla deliver?
Tesla vehicles have been somewhat autonomous for a while with the "Autopilot" program.
"Autopilot" mostly works like cruise control, traveling on highways at a set speed, and making lane changes by itself. It also has self-parking capabilities and can be summoned from its parking spot through an app on the driver's smart device.
In 2019, Tesla offered some of the "Full-Self Driving" features as an upgrade to "Autopilot," with some drivers paying up to $10,000 for it.
However, the New York Times says that many drivers feel cheated with the current program. After all, the term "full self-driving" implies that the car should operate completely on its own. Joel Young of New Mexico even filed a class-action lawsuit against Tesla for false advertising.
Tesla reportedly told officials that none of its cars will be self-driving by the end of 2021. Drivers are understandably hesitant to trust anything the company says if it's already touting false promises.
Tech professionals have also criticized "Full-Self Driving" for its lack of real-time mapping. Tesla's reasoning is that humans only need their eyes to navigate their surroundings in most situations.
However, cameras can't detect their surroundings as quickly as radar sensors. The functionality of "Autopilot's" existing features, like summoning a parked car, is also spotty.
Good things about "Full-Self Driving"
Despite complaints, Tesla's "Autopilot" software has proven itself useful on multiple occasions. On a recent occasion, "Autopilot" prevented an accident by slowing down and parking itself safely after the driver passed out.
Though this event shows the great potential of Tesla's "Full-Self Driving" feature, experts still believe the company is moving too quickly. A fully autonomous future brings up liability concerns that insurance providers might consider.
Sign up with Jerry today and learn more about coverage for self-driving cars.