Tesla secures grant for vision-based model aggregating static objects in autonomous driving.
The filing
Tesla received a US grant on November 10, 2026, for a system that processes vehicle sensor images to detect static objects in a birds-eye view. The invention uses a machine learning model to project features into a vector space and store them in a queue that updates based on vehicle speed and location, allowing the system to aggregate prior image features. The granted patent covers a method, system, and storage media where the model uses multiple heads to process these features, with specific claims for a hydra network architecture that identifies road edges, travel directions, and lane connections.
The specification of record has no background section.