INS stochastic error detection during kinematic tests and impacts on INS/GNSS performance

Inertial Navigation System (INS) and Global Navigation Satellite System (GNSS) integration requires accurate modelling of both INS deterministic and stochastic errors. The Allan Variance (AV) analysis on INS static data is one method of determining INS stochastic errors. However, it is known that INS errors can vary depending on a vehicle's motion and environment, and application of AV results from static data in kinematic operations typically results in an over-confident estimation of stochastic. In order to overcome this limitation, this paper proposes the use of Dynamic Allan Variance (DAV). The paper compares the resulting performance of the INS/GNSS integrated system by varying the stochastic coefficients obtained from the AV and DAV. The results show that the performance improved when utilizing the stochastic coefficients obtained from the DAV, applied on a kinematic dataset compared to the AV, applied on a static laboratory dataset.