Magneto-Inertial Dead-Reckoning in inhomogeneous field and indoor applications

This thesis is about complementing strapdown inertial navigation techniques with the use of local magnetic inhomogeneity, and the application thereof to indoor dead-reckoning with low-cost micro-electromechanical magneto-inertial sensors. This method provides an indirect velocity measurement of the system independently of its wearer’s movements and works without neither mapping of the magnetic ield beforehand, nor dedicated infrastructure. Modelization of the navigation problem and our sensors are studied, together with the efect of measurement uncertainty on movement estimation accuracy and the limits of this technique. Navigation algorithms based on extended Kalman iltering are implemented and evaluated in experiments. Lastly, two magnetic gradiometers calibration techniques are introduced and tested, to ease its realization both in production and during the system’s lifetime.