Multi-sensor Fusion Mapping and Localization Method for Complex Geographical Environments

In environments characterized by complex geographical features, the presence of vegetation with pronounced unstructured attributes challenges the efficacy of exclusive reliance on LiDAR for map construction, as it may result in localization inaccuracies due to imprecise matching of feature points. Furthermore, the impact of IMU’s cumulative errors on localization is intensified by the vast scale of such environments, while GPS signals are prone to instability owing to vegetation obstructions. These observations underscore the potential pitfalls of singular sensor deployment in compromising localization precision, and consequently, map accuracy. In light of this, the present study advances a novel SLAM (Simultaneous Localization and Mapping) approach that amalgamates the capabilities of LiDAR, IMU, and GPS. Through comparative analyses with single-sensor paradigms, our findings demonstrate that this integrative multi-sensor fusion strategy substantially enhances localization accuracy and facilitates the creation of high-fidelity maps.