This paper presents a feasibility study of a LiDAR SLAM based prior map fitting pipeline that yields map-referenced position estimates in known harbour environments, independent of GNSS. To address the challenge of precise positioning in unstructured and dynamic environments, a LiDAR sensor provides accurate measurements of the surroundings, which are fitted to a prior map using rotational initial alignment and G-ICP detail registration. The initial alignment process improves the accuracy of subsequent registration by leveraging corner features of the environment to address the limitations associated with large transformations in ICP registration. The resulting position estimates enable redundant positioning estimated by an Unscented Kalman Filter sensor fusion algorithm together with GNSS positioning. Results demonstrate map-referenced positioning accuracy of 3-6 m and end-to-end system latencies of approximately 1.8 s, indicating the method’s practical feasibility for coastal and port operations and its potential role in improving the navigation safety and autonomy of maritime platforms. While likely not generalizable for all land structures, we argue the potential usefulness of the system for redundant positioning in pre-vetted, well structured harbour environments.
Semb et al. (Mon,) studied this question.