MEMS IMU Array System-Level Backtracking Calibration Method Based on Angular Velocity Constraint

Abstract

To improve the measurement accuracy of a microelectromechanical system (MEMS) inertial measurement unit (IMU) array, this article proposes an angular-velocity-constrained left-invariant system-level backtracking calibration method. A simplified three-position calibration path is designed using a dual-axis turntable to excite the systematic errors of the MEMS IMU array. The primary innovation is the construction of a calibration model that introduces the turntable angular velocity as an additional measurement, while backtracking navigation reuses the stored data to accelerate parameter convergence. Numerical observability results demonstrate that the angular velocity constraint (AVC) mainly improves the observability of systematic errors related to and coupled with the gyro. Ablation results further verify the respective contributions of the AVC, backtracking procedure, and left-invariant error model. Simulations with different array sizes and operating conditions show that the proposed method maintains stable calibration accuracy and achieves a relative root-mean-square error (RRMSE) below 3 % under all tested initial-error, sensor-noise, and rotation-rate conditions. For a 16-MEMS IMU array, the proposed method estimates 432 systematic-error parameters within 250 s. Turntable and navigation experiments demonstrate that the proposed method reduces calibration time and improves the navigation performance of the MEMS IMU array compared with the filter-based comparison methods.