AGV Mapping and Localization Method Based on Multi-Focus Cloud Filtering and Map Feature Factors in Factory

Smart factories, a key part of Industry 4.0, rely on AGV (Automated Guided Vehicle) systems for automated logistics. For autonomous navigation, Simultaneous Localization and Mapping (SLAM) is crucial. Cartographer and Adaptive Monte Carlo Localization (AMCL) are popular SLAM techniques, with Cartographer providing high-precision mapping and AMCL enhancing localization accuracy. However, these algorithms struggle in complex industrial environments and cannot meet the high-precision and robustness required in smart factories. This paper proposes a new method for mapping and localization using multi-focus cloud filtering and map feature factors. In the mapping phase, voxel grid segmentation allocates point clouds into cells, and outlier points are removed through line fitting. In the localization phase, map feature factors combined with AMCL's weight model correct the odometry position. For improved relocalization, a feature-similar region replaces the traditional particle distribution area. Simulation results show that this approach outperforms traditional methods in mapping accuracy, localization precision, and relocalization, making it suitable for various 2D environments.