A Novel GPS-Aided Robust Calibration Method for SINS/DVL Integrated Navigation System

The strapdown inertial navigation system (SINS)/Doppler velocity log (DVL) integrated navigation system is pivotal in autonomous underwater vehicles (AUVs). In practical applications, misalignment angles of the DVL installation and nonuniform scale factors along each axis can significantly degrade positioning accuracy, thereby necessitating precise calibration before mission execution. This article redefines the DVL calibration task as a specific instance of the Procrustes optimization problem and incorporates nonisotropic scale errors to more accurately characterize the actual measurement properties. By employing an alternating optimization strategy to determine the rotation and scaling matrices, calibration parameters are obtained efficiently. To address the impact of outliers and non-Gaussian noise in complex marine environments, this study utilizes truncated least squares (TLS) and graduated nonconvexity (GNC) methods. Through residual truncation and progressive optimization transitioning from convex to nonconvex, these approaches enhance the calibration algorithm’s robustness. Furthermore, to reduce the dependence of robust strategies on precise noise variance settings, a quantile-based robust variance estimation method is adopted. The distributional fit is then verified using the Cramér-von Mises (CvM) statistic, enabling adaptive identification of noise variance characteristics. Simulation and sea trial results demonstrate that the proposed adaptive and robust calibration method achieves superior accuracy and stability under complex underwater conditions with high noise and outliers, thereby offering strong support for high-precision autonomous navigation of AUVs in practical engineering applications.