Combining inertial measurements and distributed magnetometry for motion estimation

We address the problem of estimating the position of a rigid body moving indoors. Disturbances of the magnetic field observed in buildings are used to derive a reliable velocity estimate. The estimated velocity is expressed in the body reference frame, which imposes to simultaneously reconstruct the rotation of this frame with respect to an inertial frame of reference. For this, an inertial measurement unit (IMU) is used. To maximize the accuracy of the reconstructed motion, alignment and calibration of the inertial sensors have to be carefully treated, which minimizes projection errors. A first contribution of this paper is an alignment-calibration technique combining gyrometers and accelerometers to address the attitude estimation problem. A second contribution is an observer of the velocity, the convergence of which is proved. Finally, an experimental testbench is described and experimental results are provided.