Cubature rule-based distributed optimal fusion with identification and prediction of kinematic model error for integrated UAV navigation

Integrated MIMU/GNSS/CNS (micro-electro-mechanical system-based inertial measurement unit/global navigation satellite system/celestial navigation system) is a promising strategy for UAV (unmanned aerial vehicle) navigation. However, given its strong nonlinearity and involvement of kinematic model error, integrated MIMU/GNSS/CNS UAV navigation presents the difficulty in achieving optimal navigation solutions. This paper presents a method of cubature rule-based distributed optimal fusion combined with identification and prediction of kinematic model error to address the above problem in nonlinear integrated MIMU/GNSS/CNS. This method is in a distributed structure to simultaneously process observations from integrated MIMU/GNSS and MIMU/CNS subsystems for the subsequent global fusion. A theory for identification of kinematic model error is established using the concept of Mahalanobis distance. Further, the standard cubature Kalman filter is modified with the model prediction filter to serve as the local filters in integrated MIMU/GNSS and MIMU/CNS subsystems to hinder the disturbance of kinematic model error. Based on the above, an optimal fusion technique is developed to fuse the filtering results of each subsystem for achieving globally optimal state estimation in the sense of mean square error. Simulations and experimental results as well as comparison analysis demonstrate that the proposed distributed optimal fusion method can effectively identify and predict kinematic model error and further achieve globally optimal fusion results, leading to improved performance for integrated MIMU/GNSS/CNS UAV navigation.