Survive Against Degradations: An Insight of LiDAR-IMU SLAM Under Feature Degradations

LiDAR-based Simultaneous Localization and Mapping (SLAM) plays a crucial role in autonomous navigation, particularly in GPS-denied environments. However, the performance of LiDAR SLAM is significantly impacted by geometric feature degradation, such as in long corridors and staircases, where insufficient structural variation leads to localization drift and mapping inconsistencies. This study evaluates the robustness and accuracy of four state-of-the-art LiDAR SLAM algorithms—Fast-LIO, Faster-LIO, Point-LIO, and PV-LIO—using a public dataset recorded in feature-degraded environments, namely GEODE. Experimental results indicate that PV-LIO outperforms the other algorithms, demonstrating superior accuracy and robustness in challenging conditions. These results offer important insights into the constraints of existing LiDAR SLAM methods and serve as a guideline for selecting algorithms in environments with low features.