Decoupling Gyro Bias from Attitude Estimation in PAHRS: A Time-Differenced Modeling Approach for Polarization Sensor Measurement

The polarization-based attitude and heading reference system (PAHRS) is able to provide attitude and heading information in the case of global navigation satellite system signal degradation. A key issue in the PAHRS data processing is the estimation of the gyro bias and attitude misalignment angle, which could have a significant impact on the overall attitude determination performance. In the existing polarization sensor (PS)/inertial navigation system (INS) fusion models, however, the gyro bias and attitude misalignment angle are coupled. As a result, the error of gyro bias estimation and that of attitude misalignment angle estimation cannot be separated. When the attitude misalignment angle has the estimation errors (e.g., acceleration scenario), the coupling effect between the gyro bias and attitude misalignment angle will further aggravate the estimation errors, which could seriously degrade the attitude determination accuracy. To improve the accuracy and adaptability of the PAHRS, in this article, we propose a time-differenced PS (TDPS)/INS modeling approach and present a polarization units selection strategy for information extraction from PS and INS. Different from the existing PS/INS fusion models, only gyro bias appears in the TDPS measurement model, thus eliminating the impact of attitude misalignment angle errors on gyro bias estimation. With the estimated gyro bias, a compensation term is constructed for the attitude misalignment angle estimation in the TDPS/INS data fusion process. In this way, decoupling of the gyro bias from attitude misalignment angle estimation can be achieved. Moreover, a polarization units selection strategy is proposed, the reliable polarization units in the PS can be, thereby, selected adaptively to fuse with the INS, which can reduce the effect of unreliable polarization data. Finally, the effectiveness and practicality of the proposed TDPS/INS modeling and data fusion approaches are demonstrated via both numerical simulations and vehicle test.