Abstract: To reduce the negative impact existing in the ultra-tightly coupled structure based on single-modulating loop, an ultra-tightly coupled structure based on double-modulating loop is designed. Aiming at overcoming the nonlinear filter problem of the tracking loop in the designed structure, an ultra-tightly coupled tracking method using a robust adaptive cubature Kalman filter (RACKF) algorithm is proposed to cover measurement outliers and kinematic model errors. The proposed RACKF algorithm adopts the robust M estimation to adjust cubature Kalman filter (CKF) algorithm so as to detect and reduce the influence "degree" of measurement outliers. At the same time, RACKF adopts an adaptive factor to dispose the influence introduced by dynamic disturbance errors. Experiment results show that the proposed method can effectively resist errors aroused by inaccurate model, and that the high-accuracy and stably tracking for GPS satellite signal are achieved preferably. As revealed by comparative studies, the proposed method is superior to tracking method based on single-modulating loop in tracking performance. Moreover, the proposed method has a higher tracking performance than tracking methods based on unscented Kalman filter (UKF) and CKF, and thus it improves the applicability of the navigation system under the circumstance of high dynamics.