Magnetic field norm SLAM using Gaussian process regression in foot-mounted sensors

Abstract

We propose an application of magnetic field norm simultaneous localisation and mapping to measurements from a foot-mounted sensor for pedestrian navigation. The algorithm is, to the best of the authors’ knowledge, the first three dimensional drift-compensating indoor navigation method using only accelerometer, gyroscope and magnetometer measurements that does not rely on assumptions about the spatial structure of the indoor environment. We use a Rao-Blackwellized particle filter to simultaneously and recursively estimate the magnetic field norm map using reduced rank Gaussian process regression, and the position and orientation of the sensor. Our experiments demonstrate that our algorithm results in a drift-free position estimate using measurements collected from a foot-mounted sensor while walking around inside a hallway.