Robust In-Motion Alignment of Low-Cost SINS/GNSS for Autonomous Vehicles Using IT2 Fuzzy Logic

• Published in: IEEE internet of things journal Vol. 12; no. 8; pp. 9996 - 10011
• Main Authors: Lyu, Weiwei, Meng, Fanlong, Jin, Shuanggen, Zeng, Qingjun, Wang, Yingli, Wang, Jinling
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 15.04.2025; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Accelerometers; Accuracy; Alignment; Automation; Bias; Coarse alignment; Covariance matrix; Data analysis; Errors; Estimation; Fuzzy logic; Global navigation satellite system; global navigation satellite system (GNSS); inertial navigation system (INS); Inertial platforms; Inertial sensing devices; Inertial sensors; Internet of Things; interval type-2 (IT2) fuzzy logic; Kalman filters; Low cost; Matrix decomposition; Navigation; Noise; Noise measurement; Outliers (statistics); Position measurement; reconstructed observation vectors; Robustness (mathematics); State space models; Vectors
• ISSN: 2327-4662; 2327-4662
• DOI: 10.1109/JIOT.2024.3508716

Low-cost strapdown inertial sensors have been increasingly applied to navigation and positioning in the Internet of Things (IoT) devices in recent years. The initial alignment accuracy of low-cost inertial sensors is critical for subsequent navigation performance. However, they face significant alignment challenges, particularly in complex environments. This article proposes a robust in-motion alignment method, which takes the bias of the gyroscope, the bias of the accelerometer, and the lever arm into account, thus avoiding the continuous accumulation of inertial measurement unit (IMU) bias errors during alignment. The accuracy of vector construction is improved by developing a novel state-space model for online calibration and compensation of unknown IMU errors. The reconstructed observation vectors achieved through the residual term enable the detection and isolation of outliers within the aided velocity information. The proposed dual-input interval type-2 (IT2) fuzzy inference system enables accurate adjustment of the measurement noise covariance matrix, even when the system is subject to outlier interference. Simulation and experimental results show that the proposed alignment method significantly enhances alignment accuracy with an improvement of over 63% when compared to the HIMCA method and more than 37% when compared to the IICA method. In particular, for the heading angle, the proposed method demonstrates outstanding performance with errors stabilizing within 1.28° after 350 s, while the other four methods struggle to achieve convergence. The proposed alignment method exhibits stronger anti-interference capability and can significantly improve the alignment accuracy.