Insects/birds can use polarization information and geomagnetic to navigate. The bionic polarization navigation using the sky polarized light field as the information source has the advantages of good real-time and no cumulative error, which can improve the robustness of AHRS. The hardware and software integration of polarized light sensor, IMU, magnetometer are implemented to realize the integration and high performance of the system. And addressing the issue of polarization information being susceptible to interference based on Kalman filter and confidence estimation, the "polarized light/geomagnetic/inertial" fusion algorithm is designed. Finally, the performance of the system is tested. The RMSE of attitude and heading in static condition are 0.24°, 0.25° and 0.25°. Under dynamic condition, the attitude RMSE is 0.49° and 0.66°; under non-interference condition, the heading angle RMSE is 0.34°. Under light interference, the heading angle RMSE is 0.39°, which is 40% lower than before fusion; under magnetic interference, the RMSE is 0.38°, which is 86.1% lower than MTI AHRS. The results show that the system has high attitude accuracy and good robustness.