Global Navigation Satellite System-Aided Post-Simultaneous Localization and Mapping Correction of Lidar Maps: Modular Pipeline for Enhanced Global Accuracy

Lidar-based simultaneous localization and mapping (SLAM) enables the generation of detailed 3D maps for such applications as autonomous navigation and infrastructure monitoring. However, SLAM systems are prone to drift accumulation, especially in the absence of a tightly integrated global navigation satellite system (GNSS) or inertial measurement unit, leading to degraded global accuracy. This paper presents a modular, post-processing correction pipeline that leverages high-accuracy GNSS real-time kinematic (RTK) data to correct lidar SLAM outputs in high-drift scenarios. The pipeline operates in three stages: (1) segment-based SLAM trajectory correction through alignment with time-synchronized GNSS RTK points using a singular value decomposition and Iterative Closest Point (SVD-ICP) method; (2) propagation of these corrections to the aggregated point cloud through timestamp-aligned delta application, generating a non-rigid intermediate reference map; and (3) conditional map refinement using either the SVD-ICP method for globally rigid alignment in low-drift scenarios or a Random Sample Consensus (RANSAC)-progressive Iterative Closest Point (ICP) method for robust registration in cases with significant residual drift, outliers, or local inconsistencies. The final output is a globally corrected high-fidelity point cloud in LAS file format. Experimental results demonstrate sub-meter global accuracy and strong visual consistency with satellite basemaps and GNSS control points, confirming the pipeline’s effectiveness for post-SLAM correction in environments with limited sensor integration.