This invention describes an autonomous robot designed to navigate by analyzing images of its surroundings. It uses different specialized neural network models, each uniquely associated with a specific "region" of its environment. As the robot moves from one region to another, it switches to the corresponding neural network model to guide its movement. If an obstacle blocks its path, the robot can request a new route from a remote server, which provides new regions and their respective neural network models.
Why it matters: Filed before significant advancements in neural network efficiency and edge AI hardware. Training and deploying specialized neural networks for specific environmental regions on a robot has become substantially more practical and performant since 2019.
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