AbstractThe presented work aims to give an overview of different calibration methods for magnetic field sensors, which are used for attitude determination. These methods are applicable in the field without any additional equipment. However, sometimes they require simplification assumptions. The paper addresses the validity of these assumptions, the accuracy and efficiency of the methods and the influence of the calibration error on the orientation estimation. Both simulations and measurements are used for evaluation. The measurements are performed using a GNSS multi-antenna system, providing an orientation reference (roll, pitch, yaw) without unknown external magnetic disturbances and with a sufficient accuracy (about 0.5 degrees). It is shown in simulations, that a full calibration of the sensor (including soft and hard iron disturbances by nearby materials) is possible without any additional equipment. However, experiments show, that some parts of the full calibration procedure are sensitive to an accurate execution of the necessary movements, which may lead to calibration errors in the same order of magnitude as a simplified method, which ignores the presence of soft iron disturbances.