Research on Terrain Matching Navigation Method Combining Cross-Correlation and Decay Memory

To address the navigation needs of unmanned aerial vehicles (UAVs) in complex terrain areas and resolve issues such as low positioning accuracy and poor terrain adaptability associated with single-path terrain profile matching methods, this paper proposes a terrain matching auxiliary navigation method based on real-time two-dimensional terrain elevation maps constructed using LiDAR. This method utilizes point cloud data obtained from LiDAR scans during UAV flight to construct a two-dimensional real-time terrain elevation map, enhancing the richness of the real-time terrain information collected. Additionally, to tackle the matching errors caused by cumulative errors in the inertial navigation system, a two-dimensional terrain matching algorithm combining normalized cross-correlation information and decay memory is designed. Drawing on the grayscale region-based template matching method, this approach improves the real-time performance and accuracy of matching. Simulation flight experiments were conducted in regions with rich terrain features such as mountains and hills. The test results show that compared with traditional barometric sequence terrain matching methods, the terrain matching method proposed in this paper improves longitude positioning accuracy by 76.8% and latitude positioning accuracy by 66.3%, demonstrating high practical application value.