Tesla Path to a General Physical World AI for Driving and Humanoid Robots

Dr Know it All points out that Ashok Elluswamy, Tesla’s Director of Autopilot software, said Tesla is building Foundational World Models for its AI. John Gibbs (Dr Know It All) believes this is Tesla converting the Full Self Driving software to have a large language model style view of the physical world. Instead of training to predict the next word, John believes Tesla is taking many short video clips of driving and blocking off a few seconds and having the system predict those few seconds.

recently gave a talk at CVPR where he mostly rehashed what we already knew from AI Day 2 last year…. However toward the end of his presentation, he got into building Foundational World Models, and this is where things got interesting indeed. If Tesla can make this happen, the whole universe of autonomous driving is about to change!

In a recent CVPR talk, Phil Duan, Senior staff software engineer at Tesla, provided more clues about where Tesla is headed with their current Occupancy Networks and future Foundation World models. Along with Ashok Elluswamy’s talk and other recent videos (see links below) this talk helps us understand where Tesla is headed in the next 6 months. Tesla is making an physical world “everything bagel” that helps them build a true AI FSD empire.

This general foundation AI will be a base understanding of our physical work for driving, walking and interacting.