Gary Zhou 
Since Tesla released its Full Self-Driving Beta last week, the fortunate group of people who have shared the capabilities of the new feature are demonstrating that Autopilot has improved significantly. Thanks to a 4D perception of the world, which is a hint of what is to come with Tesla’s forthcoming “Dojo” Supercomputer, previously difficult tasks for the 2.5-dimension Autopilot models are no longer a tough challenge.
Tesla owners who have had FSD for some time are aware that the self-driving suite’s capabilities have been somewhat limited. All who bought FSD realized it was a work in progress, and with the aid of Tesla’s Neural Network, it became more advanced by driving with the capability enabled. The Tesla Artificial Intelligence team, however, understood what had to be done: the amount of information that could be processed needed to be greater, and it needed to be more complicated for the vehicle to understand its environment. As a result, Tesla is building Dojo.
Dojo is the Neural Network training program of Tesla that aims to start breaking down data in 4D instead of “~2.5D,” which is what the Autopilot of the automaker used earlier.
Musk detailed the need for a more complex autonomy system during the Q2 2020 Earnings Call:
“Well, the actual major milestone that’s happening right now is really a transition of the autonomy system or the cars, like AI, if you will, from thinking about things in — like two-and-a-half feet. It’s like think — things like isolated pictures and doing image recognition on pictures that are harshly correlated in time but not very well and transitioning to kind of a 4D, where it’s like — which is video essentially.”
The problem with previous constructions of FSD and Autopilot was that not enough information was transmitted via photographs. There had to be timestamps and more precision with an awareness of the world that was increasingly fluid. Transitioning from pictures, or 2D, as Musk called it, to video, or 4D, was the answer.
“So what we’ve been doing, thus far, has really just been like 2D — mostly 2D, and like I said, well correlated in time. So just hard to convey just how much better a fully 4D system would work — does work. It’s capable of things that if you just look — looking at things as individual pictures as opposed to video — basically, like you could go from like individual pictures to surround video, so it’s fundamental. So the car will seem to have just like a giant improvement.”
Previous models of Autopilot had trouble going around roundabouts, and without human interference, they only rarely managed to get through one. An example can be seen in a July 2019 video from YouTuber Dirty Tesla, who showed his Model 3 trying to go through the tricky stretch of roadway. You can see the Model 3 does not do a great job of getting it through at the 3:25 mark of the video, and the driver is forced to take over.
Tesla’s FSD Beta indicates that an improvement in awareness is just what Tesla Autopilot needed to operate more precisely. A video from fellow Tesla Model 3 owner James Locke, who obtained the FSD Beta, illustrates with relative ease the navigation around a roundabout. Even Locke was impressed and claimed that no interference from him was needed for the maneuver, and Autopilot independently took care of the entire operation.
In combination with the latest FSD Beta, Dojo’s upcoming release could prove to be the solution to all the problems that Tesla previously had. Autopilot is more reliable than ever with a modern, more complex device that takes in more details about terrain, surroundings, and obstacles. The increase in capability is being demonstrated daily as new videos of the FSD Beta are being rolled out on a regular basis.
Want to buy a Tesla Model 3, Model Y, Model S, or Model X? Feel free to use my referral code to get some free Supercharging miles with your purchase: http://ts.la/guanyu3423
You can also get a $100 discount on Tesla Solar with that code. Let’s help accelerate the advent of a sustainable future.
