Adaptive Filtering and Temporal Alignment Based Fusion Algorithm for Navigation Systems in the Arctic Region

The navigation systems of vehicles face notable challenges in the Arctic region. To overcome the problems of single-sensor solutions, such as inertial navigation systems and global navigation satellite systems, hybrid navigation solutions with multiple complementary sensors are required. This paper proposes a new decentralized architecture to adapt the practical measurement noise of sensors and improve the robustness of the entire system. Previous studies have examined such techniques as Kalman filters that are primarily used to address stationary Gaussian noise. An approach named interacting dual model based adaptive filter (IDM-AF) is proposed in this study to overcome nonstationary noise and certain non-Gaussian factors, and a method that dynamically estimates the noise covariance matrix is derived in this paper. In accordance with practical requirements of fusing asynchronous sensors, we propose a new information fusion scheme based on the so-called degradation indicators of subsystems and a temporal alignment algorithm. Subsystem level and system level simulations and evaluations demonstrate that the proposed architecture and algorithms enhance navigational performance in the Arctic. Compared to a current method, the proposed architecture is shown to reduce the positioning error by 33.3-46.9% for different scenarios.