In this paper, we present a computational-efficient integrity risk optimization method in Light Detection And Ranging (LiDAR) navigation system using the concept of dilution of precision (DOP). LiDAR navigation system requires two main pre-estimator procedures; feature extraction (FE) and Data association (DA). FE and DA repeatedly and consistently identify landmarks in the environment. If these processes are failed (e.g., mistaking one landmark to another; incorrect association), significant navigation errors could occur (Joerger & Pervan 2009). Thus, previous studies established FE and DA risk prediction algorithm based on the Multiple-Hypothesis Innovation-based (MHI) DA process (Joerger et al. 2017, 2018, Joerger & Pervan 2019). The MHI DA process evaluates the impact on integrity risk of incorrect associations. However, this method requires high computational loads because this method considers all potential landmark permutations for the evaluation. Thus, we suggest a computational-efficient method by reducing the combination of landmarks. The method considers the geometric distribution of the extracted landmarks, using the concept of DOP. We assume that the landmark sets with poor DOP have higher integrity risk. To check this assumption, we conduct a sensitivity study of the integrity risk according to the geometry of the landmarks. Based on the results, we reduce the computational complexity by excluding the landmark sets with poor DOP. Then, we determine the optimal landmark set to derive the lowest integrity risk. Finally, Lidar-based navigation integrity performance based on the determined optimal landmark set is assessed using simulations.

Keywords: LiDAR, SLAM, autonomous car, HAV, integrity