SYSTEM AND METHOD FOR NAVIGATION OF AUTONOMOUS MOBILE ROBOTS BASED ON A HIERARCHICAL WORLD MODEL

Abstract

The invention is a proposed autonomous mobile robot swarm navigation system based on a hierarchical world model in a dynamic environment containing many uncertainties, including: one or more autonomous mobile robots (100) forming a robot swarm; a training server (200) connected to the robot swarm via a wireless communication network; each robot (100) configured to perform decentralized navigation at the edge and exchange information with neighboring robots via a wireless communication network. The autonomous mobile robot swarm navigation method includes the following steps: (i) acquiring and encoding latent features at the edge, (ii) selecting a sub-policy for navigation, (iii) predicting future states using a hierarchical world model, (iv) simulating future trajectories through an imaginary loop, (v) determining intermediate target states for navigation, (vi) hierarchical navigation and control execution, (vii) adaptive learning and model optimization.