Abstract:In the calibration of accelerometers and magnetometers for inexpensive attitude and heading reference systems, often the signal-to-noise ratio of magnetometers is much lower and the results of calibrating the sensor error parameters using only ellipsoidal fitting are not accurate enough. In order to improve the accuracy of the calibration results, an error parameter calibration and correction method under multi-position conditions is proposed, based on ellipsoid fitting constraints and dot product method. First, 15 pairs of low-noise measurements were acquired at static multiple positions and calibrated with ellipsoid fitting, and secondly, constructing ellipsoidal constraints using the data distribution characteristics of the compensated magnetometer, the angle between the geomagnetic vector and the horizontal plane is calculated by the dot product method, and the correction effect is quantified in terms of its mean square error. Correct the parameter of the magnetometer's inter-axis non-orthogonal error, which is less observable, and the inter-sensor non-alignment error. Ultimately improve the calibration results. Simulated and measured experimental results show that under the same ellipsoid fitting constraints, this method reduces the mean square error of the angle between the geomagnetic vector and the horizontal plane by more than 25% compared to the independent use of ellipsoid fitting and dot product invariant methods, and it is simple and easy to use.