Calibration and sensitivity enhancement of AI rate integrator through EKF and impact on raman spectroscopy optimal wavelengths.

Developments in atomic rate sensors have opened new opportunities for achieving high-precision rate integration. This study investigates the measurement principles and calibration parameters of an atomic interferometry (AI) rotation-rate sensor, modeled through Raman beam interactions and validated using real trajectory data. The primary objective is to identify and estimate the sensor parameters in order to improve orientation accuracy. To this end, real test data were collected from GPS-aided inertial systems (AIDIS 16407 and Vitans) along a predefined route. These data were incorporated into a software-in-the-loop AI gyroscope model, in which Raman laser beams acted as the excitation signal. A two-stage cascaded Extended Kalman Filter (EKF) was developed to simultaneously estimate sensor bias and calibration parameters. This approach enables parameter identification without requiring a physical AI gyroscope, thereby providing a cost-effective calibration framework. The experimental results demonstrate a significant reduction in bias uncertainty from 10-6 to 10-10 rad/s, along with a 35% improvement in orientation accuracy compared to the uncalibrated case. Furthermore, the calibrated AI parameters were applied to Raman spectroscopy of Rubidium (Rb), where an optimal wavelength of 1100 nm increased the Raman peak intensity by 20%, enhancing molecular bond resolution and spectral clarity.These findings confirm that the proposed calibration approach not only improves navigation accuracy but also enhances Raman spectroscopic analysis, highlighting the interdisciplinary potential of AI-based sensors.