Low-altitude UAV Positioning Fusing Pyramid Grid and Direction-finding Cross-location

To address the issue that traditional ground-based direction-finding equipment cannot acquire aircraft altitude data during low-altitude airspace surveillance, this study investigates a low-cost algorithm integrating Gaussian pyramid airspace grid models with ground-based direction-finding equipment, enabling real-time and precise 3D position (including altitude) prediction for low-altitude UAVs. The accessible airspace is discretized into a computable airspace using cubic-meter-level optimal granularity 3D grid technology, laying the computational foundation. During Gaussian filtering downsampling, a 3D Gaussian kernel function dynamically correlated with turbulence intensity is introduced, innovatively con-structing a multi-scale Gaussian pyramid airspace model. Trilinear interpolation upsampling ensures data continuity and precision. Real-time weather conditions, geographic information, and environmental factors are mapped to the airspace grid, establishing a dynamically weighted credibility matrix via a dynamic weighting function based on the variance of environmental parameters. Within the pyramid grid space, combined with direction-finding cross-location data, the algorithm traverses the airspace grid probability set to calculate latitude/longitude and altitude of the UAV, achieving 3D localization. Experimental validation is conducted by deploying two detection devices in a test area. The results demonstrate that: ① Positioning Ac-curacy: In a 3D grid with a minimum resolution of 8 m, the maximum latitude/longitude deviation is 20 m (during target turning), and the average altitude prediction deviation is 4.37 m (standard deviation: 7.87), significantly outperforming comparative methods. ② Computational Efficiency: The algorithm averages only 55 MB memory usage and 9% CPU utilization on an i9-13900H processor, markedly lower than comparative methods. ③ Applicability: It requires only low-cost ground-based direction-finding equipment without onboard devices. The proposed algorithm achieves low-cost, high-precision 3D real-time localization for low-altitude UAVs within cubic-meter-level deviations, providing an effective solution for scenarios with constrained low-altitude surveillance infrastructure deployment.