Spectral Analysis of Geomagnetic Signatures

Robot localization geomagnetic signatures signal processing

Electronic compasses have one drawback when used inside a building: they can easily be disturbed by electromagnetic sources (e.g. power lines, electromagnetic fields of robot internal components, fields of external instruments or large Ferro-magnetic structures). This makes it impossible to use electronic compasses as reliable heading devices for indoor applications. Inspiration for this investigation arose in trails to investigate error readings of the compass caused by electromagnetic sources or large ferromagnetic structures. It is found that reading errors for each particular terrain have a unique signature. In general, any distinctive-environment feature that is specific to a particular area can be used to recognise its own characteristics (natural landmark).

This article, therefore, introduces a technique for localization using an electronic compass in detection of any magnetic anomaly for vector mapping paradigm (VMP) based navigation. The advantage of the magnetic field disturbances is token by using them as recognition signatures to localize distinctive places. The study of geomagnetic signatures, presented in this article, relies mainly upon the frequency domain analysis using the discrete cosine transform (DCT). The DCT belongs to a group of algorithms (DFT, FFT, wavelet, Gabor, DST, etc.) used for the frequency transform of signals and patterns. Moreover, the DCT is used not only for coding but also for signal and pattern compression. This study has been implemented on B21-RWI robot platform (Colin & Robin - laboratory for autonomous robots, Uni. of Tuebingen).