Analysis of the Accuracy of MEMS Magnetometers in Small UAS for use in State Estimation

Accurate orientation estimation is difficult with the sensors typically available on sUAS. Magnetometers can be used to improve the state estimate and improve observability in INS-GPS systems. Typical error modeling of magnetometers however ignores many potential sources of error in sUAS applications. In this paper interference from aircraft subsystems, calibration to eliminate scale factor and scalar biases, and the nature of noise distribution for a low cost magnetic sensor are examined experimentally. Motor operation is found to be a major source of magnetic interference at distances of as much as 40cm while servo operation and avionics are found to have significant influence at shorter ranges. Furthermore the assumption of Gaussian noise for these sensors is found to be invalid. Additionally the accuracy of using magnetic field models to produce reference vectors is examined and found to be a significant error source for sub-degree accuracy applications.