一种高鲁棒仿生偏振定向方法

A Highly Robust Bionic Polarization Orientation Method

面对全球卫星导航系统(GNSS)拒止条件下的自主定向需求,制备了纳米阵列式金属偏振光栅,搭建了仿生偏振定向系统,并开展了高鲁棒偏振定向方法研究。针对多云、阴天等恶劣天气引入偏振成像噪声导致定向精度严重下降的问题,采用DeamNet网络对获取的偏振图像进行降噪处理,随后利用太阳在相机坐标系与导航坐标系中的位置关系,实现对绝对航向角的高鲁棒测量。对所提出的高鲁棒仿生偏振定向方法进行了静态与车载试验验证,结果表明:在复杂天气下对比传统定向方法,本方法定向均方误差降低为原来的12.5%。在动态车载定向试验下,均方误差(RMS)优于0.98°。

In this paper,nano-array metal polarization gratings are fabricated and a bionic polarization orientation system is built,and a highly robust polarization orientation method is developed to meet the autonomous orientation requirements under GNSS rejection conditions.Aiming at the serious decline of orientation accuracy caused by introduction of polarization imaging noise in severe weather such as cloudy and overcast,DeamNet network is used to de-noise the obtained polarization images,and then the highly robust measurement of absolute heading angle is realized by using the position relationship between the sun in the camera coordinate system and the navigation coordinate system.Static and vehicle-mounted tests are carried out to verify the proposed method.The results show that the mean square error of this method is reduced to 12.5%compared with the traditional orientation method in complex weather.Under the dynamic vehicle orientation test,the root mean square(RMS)is better than 0.98°.