Analysis of noise contributions in low-cost IMUs through Allan's variance

Abstract

Thanks to their enormous versatility, inertial platforms (IMUs) based on MEMS (Micro Electromechanical Systems) technology are now widely used in various fields. The strength of MEMS technology is the ability to miniaturize entire sensors within very small volumes. Inertial units play an important role in tracking an object in space and, therefore, in all the fields attached to it: self-driving vehicles, aerospace, and many others. The estimation of orientation in space is obtained using triaxial gyroscopes, accelerometers, and magnetometers. Raw data, output from inertial sensors, are commonly processed using different sensor fusion techniques and, on the base of their structure, may vary their sensitivity with respect to the acquired signals. A key aspect to consider in highly complex systems like these is noise components affecting measurements. The aim of the paper is to analyze the different noise components affecting low-cost IMU platforms to guide the selection of the particular sensor fusion technique to be employed.