Nonlinearity-Aware ZUPT-Aided Pedestrian Inertial Navigation Based on Cubature Kalman Filter in Urban Canyons

• Published in: IEEE transactions on instrumentation and measurement Vol. 73; pp. 1 - 15
• Main Authors: Xu, Ruijie, Chen, Shichao, Bai, Shiyu, Wen, Weisong
• Format: Journal Article
• Language: English
• Published: New York IEEE 2024; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Accuracy; Cubature Kalman filter (CKF); Datasets; Detectors; Drift; Extended Kalman filter; Global navigation satellite system; High rise buildings; Inertial coordinates; Inertial navigation; Inertial platforms; Inertial sensing devices; inertial sensor; Kalman filters; Nonlinearity; pedestrian navigation; Pedestrians; Pins; Real time; State estimation; Street canyons; zero-velocity (ZV) detection
• DOI: 10.1109/TIM.2024.3451578
• ISSN: 0018-9456

Urban pedestrian navigation is a challenging issue as the most popular positioning source, global navigation satellite systems (GNSSs), is severely affected by signal reflections or blockages from high-rise buildings. Unlike the GNSS, the inertial measurement unit (IMU) is less sensitive to environmental conditions but is, unfortunately, subject to drift over time. Applying the motion constraints, such as the zero-velocity update (ZUPT), is a promising solution for mitigating the drift. However, existing zero-velocity (ZV) detections can cause false results in urban scenarios involving more complex pedestrian motions. Meanwhile, the IMU-based model's nonlinearity further reduces the accuracy of the state estimation. This article proposes a nonlinearity-aware ZUPT-aided pedestrian inertial navigation in urban canyons to fill this gap. Our method begins with a gait interval (GI)-aided dual-threshold ZV detection scheme to prevent false or missed detections in complex pedestrian motions. A ZUPT-aided inertial navigation based on cubature Kalman filter (CKF) is formed to mitigate the impact of nonlinearity. Several datasets are collected to validate the effectiveness of the proposed method. The experimental results demonstrate that the proposed method can detect ZV more accurately and estimate pedestrian location more precisely with the CKF than extended Kalman filter (EKF) and unscented Kalman filter (UKF)-based methods. Meanwhile, the time consumption of the proposed method is essentially on par with the UKF-based method. It achieves a balance between computational efficiency and accuracy, which provides a low-drift self-contained real-time inertial navigation for pedestrians when external positioning data are unavailable. To benefit the research community, we open-source our dataset via GitHub: https://github.com/RuijieXu0408/PINS-datasets-based-on-Xsens-IMU .