Optimal vector matching fusion method for bionic compound eye polarization compass and inertial sensor integration

The compound eyes of insects can extract the heading by sensing multi-directional polarization information. Inspired by this, the integration with bionic eye polarization compass and low-precision MEMS inertial sensors yields a new choice for autonomous navigation in a GPS-denied environment. However, the occlusion environments, such as jungles and buildings, seriously affects the attitude estimation accuracy by destroying the polarization information. Considering the above problem, this paper proposes a reliable attitude estimation method based on vector matching measurement models for integration with bionic compound eye polarization compass and low-cost MEMS inertial sensors, including the polarization vector, solar vector, and gravity vector. To realize the matching and fusion among these vector models, the vector optimization selection algorithm is designed according to the number of visible polarization sensors in the occlusion environment. Particularly, the vector optimization selection factor is designed according to the error between the measured and theoretical polarization vectors. Furthermore, when using the solar vector estimation model, the degree of the polarization is applied to improve the calculation accuracy of the solar vector. The gravity vector measurement model is employed to enhance the autonomy of the integration. Finally, the effectiveness of the proposed method is verified by simulated and outdoor experiments under tree obscuration.