A Unified Gaussian Filtering Fusion Approach for Multi-Sensor Uncertain Systems

This article studies the Gaussian filtering fusion problem for multi-sensor uncertain systems. The measurements are classified as the normal and the abnormal measurements by the hypothesis tests, and a unified fusion framework of optimal estimation is proposed based on the Bayesian filtering theory to integrate the classified measurements. Under the unified fusion framework, the measurements are treated with different fusion strategies, thus the process and the measurement uncertainties are compensated by the internal interactions among the local estimators. Moreover, instead of solving the adaptive factors, the measurement uncertainties are compensated by controlling the steps of the progressive measurement update. Finally, the effectiveness of the proposed unified fusion method is verified through numerous simulations.