This invention describes a computer system that determines its own movement and position using visual and motion sensor data. While the original idea might have included 3D scene rendering, the current claims specifically focus on using a neural network, a type of artificial intelligence, to process camera images and movement data for this purpose. It combines visual tracking with motion sensor data using different parts of the neural network.
Why it matters: Filed in 2019, this predates the significant advancements in neural network efficiency and real-time processing for visual-inertial odometry. Modern hardware and mature AI frameworks now make implementing and deploying such a system far more feasible and performant.
AI gives you a few directions you could take this. Pick one, and we check whether your version is different enough to patent, then write the filing.
Reinvent this with AI