A Heading Angle Estimation Approach for MEMS-INS/GNSS Integration Based on ZHVC and SAUKF

In the traditional MEMS-INS/GNSS integration, velocity and position are regarded as measurement information. However, it will lead to unobservable of the heading angle, and the heading angle error will be divergent. Eventually, the navigation accuracy will be reduced either. In this paper, a novel approach of heading angle estimation based on Zero-Heading angle-Variation-Constraint (ZHVC) and Sequential-Adaptive Unscented Kalman Filter (SAUKF) algorithms is proposed for avoiding the heading angle unobservability and unstable of the filtering. First, we inspired by the vehicle dynamics and add new information to the original measurement vectors. This new measurement information is the difference of heading angle variations from the MEMS-INS and the theoretical value. Second, we separate the measurement update of Unscented Kalman Filter (UKF) to two parts by sequential method. One is the measurement information of velocity and position, the other one is the heading angle variation. Meanwhile, adaptive UKF only estimate the heading angle variation measurement noise covariance matrix in real-time. The simulation and experiment show that ZHVC can improve the observable degree and accuracy of heading angle than the common method. The SAUKF can estimate heading angle variation measurement noise covariance matrix in real-time. And the filter results are more stable in different motions of the vehicle.