In order to improve the precision of MEMS inertial navigation in attitude solving

a modified square root cubature Kalman filtering(MSRCKF)algorithm is proposed.The algorithm takes the quaternion and gyro drift as the system states and the accelerometer data as the system observations.Based on the square root cubature Kalman filtering (SRCKF)

the residual vector is used to construct the robust weighting matrix to adjust the measurement noise matrix in real time

and reduce the influence of the abnormal observations on the attitude solving

in order to improve the anti-jamming capability of the system.Different experimental scenarios are designed and compared with extended Kalman filtering(EKF) and cubature Kalman filtering(CKF).The results show that the root-mean-square error of the attitude measured using the MSRCKF method is less than 0.3° in static experiments

and the root-mean-square error is less than 0.5° in dynamic experiments.This method can significantly improve the precision of attitude solving compared to EKF and CKF.