Robust High-Precision UAV Positioning in Urban Environments via Lightweight GPS/IMU EKF Fusion

Abstract

In dense urban canyons, multipath and occlusions significantly degrade standalone GPS accuracy or even cause loss of lock, while low-cost MEMS IMUs provide high-rate continuity but suffer from drift. To balance shortterm stability and long-term accuracy, this paper presents a deployable and lightweight GPS/IMU tightly coupled fusion method based on the Extended Kalman Filter (EKF). The state vector includes position, velocity, attitude, and IMU biases. We employ IMU-driven prediction with Liealgebra small-angle updates and quaternion normalization for numerical stability, and introduce measurement consistency tests and an adaptive quality-based weighting scheme that adjusts the GPS covariance using satellite count, HDOP, and residual statistics. A MATLAB/Simulink urban-block simulation with noisy measurements, intermittent occlusions, and 30 s continuous blackouts compares GPS-only, IMU-only, and the proposed fusion. Results show over 60% reduction in position RMSE compared with single-sensor baselines; during a 30 s blackout the maximum position deviation is contained within 3 m, and after recovery the error returns to <2 m in 4–6 s. The approach requires few parameters and low computation, making it suitable for real-time deployment on resource-constrained UAVs.