Stochastic approach to reducing calibration errors of MEMS orientation sensors

Abstract

The authenticity of the readings of accelerometers and magnetometers is of great importance when using sensors in orientation systems. At the same time, simple calibration methods do not always provide the accuracy required for operation. The article considers a method based on the calculation of correction coefficients using the method of least squares. The stochastic approach allows calibrating orientation MEMS sensors without precise reference to the coordinates. The error of the calibrated readings in this case is less than 1%. The results of the studies showed that the calibration method under consideration is stable and reliable.