Using Observability Measures in a Strapdown INS Auto-Calibration

Auto-calibration is a technique generally used to improve the long-term performance of the pure inertial mode of an inertial navigation system equipped with a GPS receiver. Techniques of this sort use weighted estimates for instrumental error parameters, which were obtained in previous flight series, to compensate navigation errors in the current flight. Weights of the estimates should reflect their accuracy, and the way they are derived does highly affect the performance of the auto-calibration algorithm. One may define the weights either empirically based on the properties of the flight in which the estimates were obtained, or take the actual covariance estimates from the estimation procedure. We consider a stochastic observability measure derived from Kalman filtering to be used for calculating weights in the auto-calibration algorithm.