In large-scale road environments, constructing and maintaining high-precision maps is challenging, while GNSS-denied conditions exacerbate accumulated drift due to the lack of global references. Additionally, existing methods largely rely on LiDAR data but inadequately preprocess the data, which often leads to degraded accuracy and instability. To address these issues, this study proposes large-scale drift-resilient localization via multi-sensor fusion and topological map matching. The method leverages digital maps to extract topological road networks, eliminating the need for high-precision map construction. Accumulated drift is corrected by matching the odometry trajectory with the topological map, while localization accuracy and stability are further improved through precise ground point filtering and the integration of wheel odometry into a LiDAR-inertial odometry. Experiments on two campus datasets and KITTI 05 demonstrate the high accuracy and generalization of the proposed method in large-scale localization. Notably, on the longer School Dataset (3645 m), the mean error drops by 48.1% relative to LIO-SAM and 44.2% relative to FAST-LIO2. Repeated ablation trials further confirm the stability of the proposed method. These results demonstrate accurate and stable large-scale localization without high-precision prior maps.
Yang et al. (Mon,) studied this question.