"Our multi-modal neural networks are already in customer vehicles—these networks take in arbitrary modalities such as camera videos, maps, navigation, IMU (Inertial Measurement Unit), GPS etc.," said Tesla in an accompanying message to its video. This video has already been viewed more than 130,000 times and received thousands of likes on Twitter. Look how sophisticated is the program that runs Tesla's FSD. I particularly like the real-time bird view, which sees the traffic far ahead and looks for the opportunity to make a turn. Tesla explains that tasks such as Occupancy prediction are already quite general in what they represent—in some ways, they are ontology-free and simply predict the probability that some 3D position is occupied. Such occupancy can be used for collision avoidance by any robot. All of this is enabled by fleet scale auto-labelling. By using video data from multiple trips in the same location, Tesla says it can reconstruct the entire scene.
Tesla is building the foundation models for autonomous robots pic.twitter.com/VUES9jU3ze
— Tesla AI (@Tesla_AI) June 21, 2023
This is a rare video in which Tesla explains its electric vehicle how it improves the full self-driving to make it safer. In addition, Tesla says its building off state-of-the-art generative modeling techniques—enabling the company to predict possible outcomes given past observations, in a jointly consistent manner across multiple camera views on its electric vehicles, which are powered by the FSD software. Now what do you say about such a dangerously sophisticated AI from Tesla? 20 years from now we'll look back, and I hope we aren't handing the keys to earth over right now. Armen Hareyan is the founder and the Editor in Chief of Torque News. He founded TorqueNews.com in 2010, which since then has been publishing expert news and analysis about the automotive industry. He can be reached at Torque News Twitter, Facebok, Linkedin and Youtube.
