Non-Uniform Tiling of Point Cloud Layer to Optimize Updates of HD Map Data

This paper introduces a novel algorithm for the efficient handling of crowd-sourced point cloud data in high-definition (HD) map updates, a critical component for enabling autonomous vehicles. The algorithm employs a non-uniform tiling technique that divides the point cloud layer of the HD map based on detected changes, optimizing data processing while preserving necessary details for accurate map updates. By incorporating caching techniques, the proposed method tackles the limitations and difficulties faced by existing tiling approaches, resulting in improved performance. Specifically, it addresses challenges such as large data volume, transmission overhead, and latency associated with crowd-sourced map updates. Experimental results demonstrate the effectiveness of the proposed approach in overcoming these challenges, reducing computation costs, and enhancing the efficiency of HD map updates. Simulation results show that there is 37.5% improvement in caching performance and 27.77% faster data processing in case of non-uniform tiling as compared to uniform tiling scheme.