Research on Location Method of Mobile Robot Based on Hierarchical Grid Semantic Map

To address the issue of frequent relocalization failures in similar or dynamic indoor environments using the Adaptive Monte Carlo Localization (AMCL) algorithm in existing mobile robots, this article proposes a method that utilizes semantic information to establish a hierarchical grid semantic map. This approach aims to enhance the robot's situational awareness, improve its understanding of the surrounding environment. It combines semantic information for visual pre-localization, adopts Gaussian and uniform distributions for particle initialization, improves particle weight update based on visual pre-localization, and integrates AMCL algorithm with environment map matching for precise localization. This approach enhances the efficiency and accuracy of mobile robot localization in similar and dynamic environments. Comparative experiments in various scenarios conducted in a simulated environment validate the effectiveness of the proposed method.