LAUSR

With development of light detection and ranging (LiDAR) technology, solid-state LiDARs receive a lot of attention for their high reliability, low cost, and lightweight. However, compared with traditional rotating LiDARs, these solid-state LiDARs pose new challenges on simultaneous localization and mapping (SLAM) due to their small field of view (FoV) in horizontal direction and irregular scanning pattern, which arises the issue of degeneracy in indoor environments. To this end, we propose an accurate, robust, and real-time LiDAR-inertial SLAM method for solid-state LiDARs. First, a novel feature extraction based on geometry and intensity is proposed, which is the core of handling with degeneracy. To make full use of extracted features, two multi-weighting functions are designed for planar and edge points respectively in the process of pose optimization. Lastly, a map management module using an image processing method is developed not only to keep time efficiency and space efficiency but also to reduce edge intensity outliers in line map. Qualitative and quantitative evaluations on public and recorded datasets show that the proposed method exhibits similar and even better accuracy with state-of-the-art SLAM methods in well-constrained scenarios, while only the proposed method can survive in the robustness test toward degenerated indoor lab environment.