A stepwise decoupling self-calibration for attitude demodulation errors in dual-axis RINS

Abstract The attitude accuracy of dual-axis rotational inertial navigation systems (RINSs) is often degraded by periodic oscillations stemming from attitude demodulation errors. A primary cause, largely unaddressed by conventional static models, is the dynamic precession of the rotation axes. To address this, this paper proposes a systematic, stepwise decoupling self-calibration scheme to identify and compensate for these complex errors. The core of this method is an innovative time-varying error model that describes non-orthogonal axes as a superposition of constant and angle-dependent sinusoidal terms, accurately capturing the axis precession effects. The calibration scheme first resolves grating measurement errors (both functional and sampling asynchrony) and then uses single-axis rotations to isolate the non-orthogonal axis errors. These are estimated using a recursive least squares algorithm that fuses multi-source data from gyroscopes and accelerometers. Simulation and prototype experiment results demonstrate that the proposed method effectively suppresses the harmonic oscillations missed by traditional techniques. For instance, experiments showed a reduction in peak-to-peak pitch oscillation from 8 to 2 arcseconds during outer gimbal rotation, restoring attitude accuracy to the theoretical level determined by sensor noise. This work provides both a practical compensation strategy and a diagnostic tool for evaluating the mechanical quality of RINS indexing mechanisms.