An Improved Filtering Approach for the Information Sharing Error of a Multisensor Mobile Platform Against Misaligned Baselines

The land-vehicular multisensor architecture is an integrative approach for the attitude determination problem. Toward a high-accuracy navigation solution, the multiantenna Global Navigation Satellite System (MAGNSS) fused with inertial navigation system (INS) becomes a widely used information sharing construction. However, due to the constraints on the baseline configuration of the MAGNSS, the time-varying deformation and external measurement disturbances caused by the active vibration and unpredictable environment change will induce the unknown information sharing error. In this article, a suitable nonlinear measurement model for the misaligned baselines and the information sharing error model are first discussed. Then, an unscented Kalman filter (UKF) solution combined with an adaptation law is implemented to evaluate the change of statistical characteristics of measurement noises and compensate for the unmodeling error. Moreover, we proposed the modified Mahalanobis distance criterion providing a selective adaptive process, which will reduce the sensitivity of the estimation performance under the external measurement disturbances and unknown statistic noise. Hence, the estimation accuracy will be maintained with the proposed adaptive UKF assisted with the modified Malahabions distance cirterion (AUKF-MMDC). In addition to the simulation evaluation, a verification platform is developed to verify the proposed solution. Moreover, the experimental results show that the proposed solution can reduce the root-mean-square errors of compensation roll and pitch effectively.