Correction Method for UAV Pose Estimation With Dynamic Compensation and Noise Reduction Using Multi-Sensor Fusion

Pose estimation is a key feasibility issue for autonomous navigation of civilian UAVs. In response to the poor positioning accuracy caused by unknown noise in the standard Unscented Kalman Filter (UKF) algorithm for multi-sensor fusion-based pose state estimation, this paper proposes a UAV attitude estimation correction method using dynamic compensation and denoising through multi-sensor fusion. Firstly, an adaptive adjustment of the iterative transformation parameters is performed using a distance parameter adjustment strategy to optimize the distribution of Sigma sampling points. Then, dynamic estimation thresholds are used for coordinated processing of system noise and observation noise. Additionally, to address uncertain disturbances in the measurement system, a correction factor is introduced to compensate for the predicted measurement covariance. By constructing a two-dimensional matrix, outlier data points from the sensor measurements are eliminated, reducing the impact of uncertain noise on the measurement system. Simulation results demonstrate that the proposed dynamic correction and denoising UKF algorithm, compared to the standard UKF algorithm, improves the accuracy of state estimation and exhibits good precision and robustness in UAV multi-sensor fusion-based pose state estimation tests.