Accelerometer calibration with nonlinear scale factor based on multi-position observation

Abstract

The calibration of an inertial measurement unit (IMU) is a key technique to improve the accuracy of an inertial navigation system. Adding more parameters into the model and reducing the estimation errors is essential for improving the calibration methods. Given its advantage of not requiring high-precision equipment, the multi-position calibration method has been widely discussed and has shown great potential in recent years. In this paper, the multi-position calibration method is improved by introducing the accelerometer nonlinear scale factor. The observation equations for the improved multi-position calibration method are established based on a nonlinear accelerometer model. The particle swarm optimization algorithm is adopted to solve the complicated nonlinear equations. In addition, Allan variance is used to determine the optimal data collection time. The accuracy and the robustness of the proposed calibration method are verified by the simulation test. The laboratory and field experiment results for a navigation-grade IMU prove that the proposed method can successfully identify the accelerometer nonlinear scale factor and improve the multi-position calibration accuracy. The comparison of several other calibration methods highlights the superior performance of the proposed method without precise orientation control.

Gimbal Influence on the Stability of Exterior Orientation Parameters of UAV Acquired Images

“…They are subject to systematic influence errors due to imprecise scaling factors [27] and non-perpendicular axes, which results in a reduced accuracy of its position and direction [28]. …”

Accelerometer Triad Calibration for Pole Tilt Compensation Using Variance Based Sensitivity Analysis

“…Model Parameters Positions [23] Bias and Scale 6 [16,17,24,25] Bias, Scale and Non-orthogonalities 18 [22] Bias, Scale, Non-orthogonalities and Cross-axis sensitivities 18 and 24 [19] Bias, Scale and Non-orthogonalities 9 [26] Bias, Scale, Nonlinear Scale and Non-orthogonalities 24 [27,28] Bias, Scale and Non-orthogonalities 36-50 [21] Bias, Scale, Non-orthogonalities and Misalignment 30 In this contribution we investigate different accelerometer triad error models utilizing a 6-joints industrial robot. In order to compensate the pole tilt, additional mounting parameters of the triad w.r.t.…”

“…However, a more sophisticated calibration scheme is necessary for reliable estimates of the parameters. Reference [26] proposed a 24-position scheme consisting of eight 50 gon rotations per accelerometer axis. This scheme distributes the measured g n -vector evenly over the unit-sphere.…”

“…For this purpose we developed an Adjustment Sensitivity Algorithm inspired by VBSA. Starting with the 24 position scheme proposed by Reference [26] (from now on denoted L 0 ) the expected VCM of parameters can be computed [37] (for matrix notation and computation please refer to Appendix B or Reference [37]):…”

“…Bias only model ( These are estimated using the 24 position scheme proposed by Reference [26], see Section 3.1. For the mounting estimation, the IMU is mounted on a prism pole and the prism is tracked by a total station.…”