Disentangling Triaxial Sensor Nonorthogonalities and Installation Errors

Conventional characterization methods for triaxial inertial sensors, which rely on ordinary or weighed least squares fitting, blend the sensor nonorthogonalities with the installation errors. In contrast, field modulus constraint methods produce the sensor nonorthogonalities but typically cannot produce the installation errors. While neither approach is ideal out of the box, this letter presents a practical approach to efficiently disentangle the nonorthogonalities and installation errors from the results obtained through conventional characterization methods. By disentangling sensor nonorthogonalities from installation errors, it becomes feasible to assess their distinct contributions to the overall error. Although it relies on trusted mathematical techniques, the presented approach was not found to be referenced in any of the hundreds of published manuscripts related to the calibration and testing of inertial sensors. The approach is presented without loss of generality for inertial sensors, and it can be trivially extended to triaxial magnetometers. The proposed method seamlessly integrates into existing processes with minimal impact and demonstrates numerical efficiency on modern computing platforms, being three orders of magnitude faster than Gauss–Newton (GN).