Simultaneous localization and mapping based on indoor magnetic anomalies

This paper presents a simultaneous localization and mapping (SLAM) algorithm that utilizes the local spatial anomalies of the ambient magnetic field present in many indoor environments. In order to increase the positioning accuracy and reduce the amount of calculation, we improved particle filter algorithm, according to the characteristics of the magnetic field sensor measuring geomagnetic component on the three orthogonal directions and different weights calculation. In the localization stage, we use the improved particle filter to estimate the pose distribution of the robot. During the period, the convergence rate of the algorithms each iteration speeds up about 0.5s and positioning error reduces about 3.5 m. And in the mapping stage, Kriging interpolation method is more flexible than other interpolation algorithms, when it used to update the fluctuant magnetic field map. The interpolated map improves the positioning accuracy of the robot. The feasibility of the proposed approach is validated by MATLAB simulations, which demonstrate that the approach can quickly and accurately locate the robot and construct the consistent map using only odometric data and measurements obtained from the ambient magnetic field.