Novel GNSS-INS Integration Scheme With UKF-Based Dynamic IMU Calibration and Dual-layer Design for Reliable Navigation

In this work, a robust and improved loosely-coupled (LC) Global Navigation Satellite System (GNSS)- Inertial Navigation System (INS) integration scheme incorporating three important features, like dynamic Inertial Measurement Unit (IMU) calibration, Mahalanobis distance-based outlier rejection (MDOR) mechanism, and innovation-based adaptive estimation (IAE) technique, is presented for reliable and accurate navigation. Unscented Kalman Filter (UKF) based dynamic calibration of IMU is adapted here because it accurately transmits statistical distributions without linearization, which is better at managing non-linear INS dynamics than the Extended Kalman Filter. Further, MDOR mechanism is proposed to identify and exclude erroneous GNSS measurements before the filter update, whereas, IAE technique is proposed to dynamically tune the filter's noise covariance. A hardware setup is developed using a GNSS receiver, IMU sensor, and microcontroller to capture data for real vehicular trajectories. The proposed framework has been implemented in MATLAB and further experimentally demonstrated with real trajectory data. The navigation accuracy of the proposed method exhibits upto 75% improvement with respect to conventional LC integration. The contribution lies on careful integration and validation of dual-layer architecture with inter-layer feedback mechanism and nested-architecture for known UKF-based IMU calibration. The proposed framework can provide a precise navigation solution to improve resilience even in partial GNSS challenging areas.