The beta version of Total Autonomous Driving by Tesla has achieved an important milestone: a trip from Los Angeles to San Francisco without any human intervention . The 611 kilometers journey started and ended within the city using a Model 3 .
The achievement, in addition, has the particularity that it was made with the version 9.1 beta of the total autonomous driving system, which uses only the cameras and dispenses with the radars. All the way there was only one interruption to the trip to do a recharge in the Kettleman City Supercharger midway.
Weeks ago Tesla announced that the total autonomous driving system in which they are working using artificial intelligence, would stop using data from their vehicles' radars and would only use artificial vision through integrated cameras . In fact, the Model 3 and Model Y manufactured in recent months for the North American market no longer include these components.
The announcement raised concern and for weeks it was argued a probable loss of quality of the total autonomous driving system, especially in low visibility conditions. Tesla has explained that radars end up being a drag on the machine learning system. After initial testing by a small group of people within the United States, the 9.1 beta version of the system has proven to work even better than before, using only the eight cameras in the vehicles.
Total autonomous driving from Los Angeles to San Francisco in a Tesla without human intervention
The video posted by popular Tesla enthusiast, Whole Mars Catalog , who has access to the beta of full autonomous driving shows a trip without much news. The vehicle can be seen driving in and out of the city, on highways, with and without other vehicle traffic.
The only time when human intervention was required was related to the act of loading. The vehicle was driven alone to the entrance of the Supercharger where it was manually parked at the chosen charging point. A routine that could perfectly be automated also in the future.
The approach chosen by Tesla to total autonomous driving is to try to depend on the least amount of external elements and technologies possible, to fully rely on artificial intelligence and its processing by the vehicle itself. That is to say, that is the closest thing to what a human needs to be able to drive: his eyes and a map.
Tesla's Total Autonomous Driving System would only need the cameras and the GPS map , which is stored in the vehicle and not downloaded on demand. In fact, if it runs out of internet, the car is still capable of navigating from point A to point B. The company also does not pre-plot routes, cities or geographical areas, unlike the method of Waymo ). Rather, it uses data from thousands of vehicles circulating around the world to create artificial learning models that drive as well or better than a human.
