A multi-frequency filtering procedure for inertial navigation

This paper present a wavelet-based procedure for the filtering of IMU data both during the alignment and the navigation that automatically changes the cut-off parameter of the multi-resolution analysis. During the alignment procedure a "hard" filtering can in fact be used to evaluate the initial attitude angles, with a cut-off frequency close to 0.5 Hz, whereas a "soft" filtering during the navigation can be used to preserve the information on the accelerations and angular velocity measurements of the sensors, by setting a more elevated cut-off frequency. The proposed de-noising method was applied to a low-cost tactical grade Inertial Measurement Units (Crossbow IMU 700-CA and IMU 400-CC) and an improvement in the signal-to-noise ratio and consequently in the evaluation of the navigation parameters was obtained.