Dipole Model-Based Disturbance Compensation Using Magnetometer Arrays

摘要

Magnetometers are widely used as compasses, but these sensors can provide unreliable measurements due to soft- and hard-iron effects caused by the presence of metallic objects placed in their surroundings. Compensation for the disturbances is needed to obtain usable information in such an environment. This article proposes a method that can compensate for the anomalies in the Earth’s magnetic field. The method is suitable for wheeled-robot localization applications, where the robots have to be equipped with a magnetic sensor array and a detection system (DS) technology that provides the position of the detected objects. It is assumed that the magnitude of the geomagnetic field is known, and the disturbing objects are modeled as magnetic dipoles. The method uses genetic algorithm-based optimization to estimate the equivalent dipole parameters that best fit the generated disturbances. These parameters can be used to compute the disturbances that can be subtracted from the measurements to achieve compensation. The method is validated using real measurements. Five different scenarios are measured to obtain the disturbances; they include three scenarios with only metallic objects and two scenarios with an additional nonmetallic object that can be detected. The concept is tested by adopting two solutions; in the first, the equivalent dipole is placed in the center of the object, while in the second, the dipole is free to be located by the algorithm in the best position inside a volume that englobes the object. Both methods provide a significant reduction of the disturbance, but the second performs much better, providing estimations of the Earth’s magnetic field with mean absolute errors (MAEs) lower than 1.8 and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$6~\mu $ </tex-math></inline-formula>T for the easiest and toughest scenario, respectively. Furthermore, headings are computed, and an MAE in the order of 0.25 rad is achieved by method (ii) in scenario 1.