Aiming at the long term and high-precision performance maintenance requirements of the satellite-borne four-axis redundant inertial measurement unit (IMU) in space, the satellite navigation velocity information, the attitude information provided by the star sensor and the redundant measurement output are used as references to derive the measurement model, including the star sensor installation error. Three rotation sequences of optimal observation, sub-optimal observation, and under-observation are designed and compared. The Kalman filter is used to calibrate the redundant IMU in the orbit. The numerical simulation shows that all three rotation sequences can calibrate more than 91% of the constant error, which meets the requirements of the use, and the calibration accuracy of the optimal rotation sequence is the highest. The ground repeatability calibration experiment is carried out 10 times for the optimal rotation sequence. The calibration results show that the range of all the error terms is less than its threshold, illustrating that the calibration result is stable and reliable. It is a real-time calibration method for all error terms on orbit, providing a reference method for the high-precision performance maintenance of the spacecraft integrated navigation system.

The in-flight or in-field calibration of gyro in terms of all-parameter errors and star sensor installation errors is vital for improving the hitting accuracy and quick reaction capability of the SINS/CNS Integration system for near-earth flight vehicles. This paper explores an in-flight or in-field decoupled 4-position calibration method by analyzing the degree of observability of error states based on singular values. Using this approach, an effort is made to isolate individual error sources to prevent contamination by using the least number of tests. The observability matrix is decoupled into three categories of sub-matrix: the gyro-bias observability matrix, gyro-scale factor observability matrix, and the gyro-misalignments and star sensor installation error observability matrix. In this study, it is suggested that the observability degree of gyro-bias is proportional to the integration time, while the other gyro errors and the star sensor installation errors are proportional to the angular increments. Covariance analysis with a Kalman Filter (KF) are performed, which confirm these conclusions. Theoretical simulations showed a reduction in the gyro all-parameter errors and the star sensor installation errors variance and in the time required to achieve convergence with the angular increments of the three-axis direction in a few minutes.

Aiming at the inability to achieve high-precision real-time magnetic heading angle estimation under single degree of freedom rotation, a novel attitude measurement while drilling system is designed, which consists of a single axis fiber optic gyroscope, a tri-axial MEMS accelerometer and a tri-axial magnetometer. The magnetometer bias calibration model based on uniform rotation is established and verification experiments are carried out. The uniform rotation experiment shows that six calibration results fluctuate within 5% of the mean. A real-time estimation model of the magnetic heading angle under random rotation is established, of which the theoretical error is less than 0.38°. Three real-time estimation experiments of magnetic heading angle are carried out, and the overall estimation error did not exceed 0.35°, and the average estimation error did not exceed 0.3°. Under continuous rotation of large angle, the bias estimation of the X-axis and Y-axis magnetometers can be achieved. The results show that the estimated value of the bias of the magnetometer under dynamic conditions can provide a reference for the initial magnetic heading angle measurement under static conditions. The experimental results prove the effectiveness of the proposed method, which provides a reference for the further improvement of the attitude measurement while drilling system.

MEMS inertial measurement unit (MIMU) has widely used in low-cost inertial navigation fields, such as pedestrian navigation system. However, the realization of self-alignment function and high-grade navigation performance are still uprising challenges. In present study, a novel MIMU/FOG compound system consisting of the MEMS inertial sensors and a skew fiber optic gyro (FOG) is proposed. The configuration which is based on the rotation modulation technology, realizes self-alignment function, and the skew FOG used as the redundant angular velocity measurement unit improves the navigation accuracy. During the rotation process, the projection of the fiber optic gyro sensitive axis is a direct current component, the earth rotation rate projected on the sensitive axis of the fiber optic gyro is a cosine component, and the initial phase is the yaw. Further, a cosine-fitting self-alignment method which requires no latitude information is introduced to calculate initial yaw alignment accuracy. The alignment time is adjustable by setting various rotation rate of rotary mechanism. When the rotary mechanism rate is fixed at 1°/s, the proposed method achieves better yaw alignment accuracy of 0.37° (1σ) by using the 0.1°/h (100s, 1σ) fiber optic gyro. Our work opens up a new perspective to implement self-alignment of the MIMU/FOG compound system.

The field of automatic driving has always been a research hotspot. Based on inertial and odometer, dead reckoning is used to navigate for L4 and above level autonomous vehicles with less human intervention when GPS is blocked. The navigation accuracy is mainly determined by the alignment error and the drift accumulated over time. In this paper, we proposed a novel method to achieve initial alignment of inertial navigation for L4 and above level autonomous vehicles. The alignment is accomplished with a MEMS inertial measurement unit (MIMU) and single antenna global positioning system real-time kinematic (GPS-RTK). By matching the trajectory obtained from MIMU dead reckoning with the trajectory obtained from RTK, the initial heading angle is obtained, and the heading angle accuracy is improved through multiple iterations. In the case of RTK precision decreasing, the MIMU can still achieve lane-level navigation after the initial alignment. Because the alignment is accomplished in motion, there is no need for additional static alignment time before the vehicle starts. The field experiment results show that the method could achieve fast initial alignment in a few seconds in moving state with an alignment accuracy of 0.05.

Safe and stable operation of metro requires accurate kinematic measurement and positioning of the train. Traditional metro navigation and positioning methods are realized with the help of odometer or beacons such as RFID and WIFI, which means external information is necessary. However, accurate, reliable and fast metro applications call for an innovative method that can obtain high-precision, continuous metro attitude and track without being interfered by the outside world. Compared with other navigation technology, inertial navigation system (INS) is the sole navigation system that can realize independent navigation. However, INS generally has large noise and drift accumulated with time, especially MEMS inertial measurement unit (MIMU) with low cost and small volume. In this paper, we proposed a novel method of autonomous positioning suitable for metro applications only using MIMU. The selection of zero-velocity-detection variables and the determination of criterion value are realized according to the motion characteristics of train. Then, the selection of variables is proved scientific according to the generalized likelihood ratio test (GLRT). When zero-velocity state is detected, the horizontal attitude of zero-velocity interval is corrected through coarse alignment method. In addition, a dead reckoning method based on kinematic constraint of train is adopted. In order to verify the advantages of the proposed method, tests of attitude and positioning precision are performed. The field experiment results show that our method could achieve high-precision attitude keeping with less computation. The test results are presented in terms of detection versus false-alarm rate. It can be seen from our results that the new detector outperformed the traditional detectors. Furthermore, the relative positioning error is less than 11.6%.