Filtering strategy to minimize errors in the estimation of the Z angle (YAW) of a rotating body with two independent MEMS gyroscopes

Abstract

A MEMS gyroscope is a low-cost electronic device used to determine the angular rotation rate ? of a rigid body with respect to the its x, y, and z axes, the latter known as yaw. The measurement of the angular rotation rate on each axis is perturbed by phenomena known as bias, bias instability, and angle random walk. It is desirable to accurately calculate the yaw angle; however, the phenomena of bias, bias instability, and angle random walk make this calculation impossible; several algorithms have been reported in the literature to achieve such goal, such as the Kalman and the complementary filters.
We propose an improvement in the estimation of the yaw angle of a rigid body in rotation; it is achieved through a complementary filter genetically optimized that fuses angular rotation rate readings from two independent gyroscopes. The genetic algorithm is employed to calculate the parameters of the complementary filter, operating on simulated runs of the gyroscopes.
The evaluation of the resulting yaw angle precision with the calculated filter demonstrated a reduction in the absolute error of the yaw angle estimation with respect to given reference signals to up to one order or magnitude with respect to the calculation given independently by each gyroscope.
The results obtained were proven empirically.