Dual-antenna GNSS-aided robust MEMS IMU initial alignment under the earth-centered-earth-fixed frame

Abstract

The micro-electro-mechanical system (MEMS), with its small size, low cost and low power, has attracted widespread attentions in recent year, and the initial alignment of the inertial measurement unit (IMU) is an indispensable step before the navigation computation. In the traditional alignment methods with only the velocity measurements, the yaw angle is subject to divergence under low dynamic scenarios, and the measurements are also susceptible to the external environments. Furthermore, the navigation frame (n frame) is usually chosen as the reference frame. Although intuitive, it is more computationally complex compared with the earth-centered-earth-fixed frame (e frame). Therefore, to improve the alignment performance and computational efficiency, a robust MEMS IMU initial alignment method with yaw assistance under the e frame is proposed. In this method, to address the issues of the yaw divergency and computational inefficiency in the traditional methods, the relationship between the yaw angle, derived from dual-antenna GNSS baseline solution, and IMU error is derived and incorporated into the state space model with e frame as the reference frame. Besides, a robust extend Kalman filtering (REKF) is adopted to further improve the system’s performance of outlier resistance, whose robust factors are determined by Institute of Geodesy and Geophysics (IGG)-Ⅲ model. To validate the performance of the proposed method, two field experiments with the velocity of about 0.7 m s-1 and 1.5 m s-1 was conducted, and the collected data were processed for contrastive analysis. The results show that the effectiveness of the proposed method at suppressing yaw angle divergence and improving the alignment accuracy by resisting outliers under low dynamic scenarios compared with the traditional method. Moreover, the elapsed time is reduced by about 11% when the reference frame in data processing is switched from the n frame to the e frame.