Adaptive Heading Estimation Method for Pedestrian Dead Reckoning With Magnetic Interference

• Published in: IEEE transactions on instrumentation and measurement Vol. 74; pp. 1 - 11
• Main Authors: Shi, Fangyan, Xu, Xiangbo, Zhu, Yahui, Zhang, Yanan
• Format: Journal Article
• Language: English
• Published: New York IEEE 2025; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Accuracy; Algorithms; Damping ratio; Dead reckoning; Estimation; Extended Kalman filter; Filtering algorithms; Gyroscopic stability; Heading estimation; Interference; Likelihood ratio; Magnetic disturbances; Magnetic field measurement; magnetic interference; Magnetic separation; Magnetomechanical effects; Magnetometers; Mahony filter; pedestrian positioning; Pedestrians; Polynomials; Proportional integral; Real time; Real-time systems; Resonant frequencies; sensor fusion; Terrain
• ISSN: 0018-9456; 1557-9662
• DOI: 10.1109/TIM.2025.3558247

Heading estimation is crucial for pedestrian dead reckoning. Magnetometers correct gyroscopic drift by measuring Earth's magnetic field, but their susceptibility to localized magnetic interference compromises attitude estimation accuracy. To address this issue, an improved Mahony heading estimation method based on proportional-integral (PI) control and magnetic disturbance detection is proposed. The method constructs the system model based on a characteristic polynomial and estimates PI filter parameters using damping ratio and natural frequency. The generalized likelihood ratio test (GLRT) is incorporated into the method to detect magnetic interference and evaluates magnetometer reliability. Additionally, an adaptive adjustment mechanism is designed to dynamically optimize the filter's bandwidth and damping ratio based on disturbance conditions and sensor variance, enhancing heading estimation accuracy. Experiments were conducted on regular terrain and random terrain to validate the algorithm's effectiveness. The proposed method reduces the root mean square error by 9.43% and 35.47% compared to the Mahony algorithm, and by 70.73% and 76.49% compared to the extended Kalman filter (EKF)-based algorithm. The experimental results show that the improved Mahony filter effectively minimizes magnetic interference on heading estimation, while maintaining good real-time performance and accuracy.