A Bayesian framework for calibration and real-time localization of magnetometers using a controllable passive permanent magnet

Magnetic localization systems based on passive permanent magnets (PM) are of great interest due to its ability to provide non-contact sensing and lack of a power requirement of the PM. One sub-problem of particular interest is accurately localizing, in real-time, a single magnetometer with unknown position and orientation, using a passive PM with controllable position and orientation. This is a challenging problem due to the presence of measurement noises and biases, inaccuracy of the magnetic field model, and possible low observability. Bayesian statistical signal processing is a promising approach for this problem, due to its strong mathematical foundation, robustness and suitability for real-time processing. In this work, we develop a Bayesian framework for the individual sensor localization problem which is composed of three parts: magnetic field modeling, sensor calibration, and real-time sensor localization. The effectiveness of the framework is demonstrated using a experimental setup that emulates a possible Transcranial Magnetic Stimulation (TMS) application.