Adaptive Dual-Bias Estimation for Hybrid Inertial Navigation with Single-Axis Cold Atom Interferometric Gyroscope

Abstract

The errors of an inertial navigation system accumulate over time due to its operating principles and sensor limitations. Numerous studies have investigated hybrid inertial navigation schemes combining a cold atom interferometric gyroscope (CAIG) and a conventional inertial measurement unit. This paper constructs a dual-bias joint state model to realize decoupling of gyroscope biases between the conventional gyroscope and the CAIG, eliminating modeling error introduced by the drift-free reference sensor assumption. Furthermore, an adaptive Kalman filtering algorithm based on real-time residual autocorrelation identification is proposed to cope with the temporally correlated time-varying noise characteristics. A series of simulations including noise-characteristic verification and bias-decoupling comparison, as well as static physical experiments, are carried out to validate the bias-decoupling capability and overall navigation performance. Taking advantage of the low bias-drift property of the CAIG and the high sampling rate of conventional gyroscopes, static experimental results show that the hybrid navigation system with a single-axis CAIG reduces positioning errors by more than 85% after compensating for inertial sensor biases, and achieves a positioning accuracy of 5 nautical miles over a one-hour test. Although the current validation is limited to stationary conditions, this work provides a feasible solution for CAIG-aided hybrid inertial navigation.