Comparison of the North Finding Capability of several MEMS IMUs of Different Performance Grades

Utilizing Micro-Electro-Mechanical-System (MEMS) Inertial-Measurement-Units (IMUs) for determining north is increasingly attractive due to their improving performance, availability, and low cost in comparison to, e.g., Fiber-Optic-Gyroscopes (FOGs). This option is particularly relevant for navigation applications employing less reliable north-finding methods such as magnetometers for indoor use. This paper compares the north-finding capability and the effect of the data acquisition time of three MEMS IMUs with different performance grades. Furthermore, this paper discusses the influence of the north-finding capability of these IMUs on the precision of pedestrian navigation as a demanding application.Initially, a simplified version of the maytagging procedure executed on a precision rate table revealed the noticeable effect of the earth rotation rate on the gyroscope output. Then, north-finding tests were conducted by applying a maytagging procedure. The data sets from each position were averaged and further processed using the method of least squares. Results are presented by comparing three performance grades at different data acquisition times. In a third step, a pedestrian navigation algorithm was modified for utilizing the IMU’ north-finding capability and the impact of earth rotation on navigation accuracy was examined. The results of the study lead to the following conclusions:•The outputs of all three MEMS IMUs reflect noticeably the earth rotation rate and succeed in finding the north direction with a maytagging procedure.•Maytagging for north-finding or similar procedures seem to be still necessary for all IMUs in test since distinguishing between bias and real sensor signal is still challenging in one-shot-measurements.•The accuracy of north-finding generally benefits from long data acquisition times. The results also indicate, however, that north-finding is also feasible with data acquisition times shorter than the duration marked by the bias instability in the Allan Variance.•The implementation of the effects of the earth rotation rate leads to a general improvement in the accuracy of the pedestrian navigation algorithm. Especially, the accuracy of the yaw-angle estimation benefits most.