Maximum correentropy-based robust Square-root Cubature Kalman Filter for vehicular cooperative navigation

Abstract

As the core method of cooperative navigation, relative positioning plays a key role in realizing intelligent vehicle driving and vehicle self-assembling network collaboration algorithms. However, when the contamination rate of measurement noise is high, the performance of filtering will be seriously affected. To better address the filtering performance degradation problem due to noise contamination, this paper proposes a vehicular cooperative localization method based on the Maximum Correentropy Robust Square-root Cubature Kalman Filter (MCSCKF). The algorithm not only retains the advantages of Square-root Cubature Kalman Filter (SCKF) but also has strong robustness to non-Gaussian noise. The experimental results of tightly integrated vehicular cooperative navigation show that compared with the Extended Kalman Filter (EKF) and Cubature Kalman Filter (CKF), the localization accuracy of MCSCKF is improved by 35.08% and 31.83%, respectively, which verified the effectiveness in improving the accuracy and robustness of the relative position estimation.