基于深度学习的非合作航天器姿态估计

Aiming at the problem that the attitude measurement of non-cooperative spacecraft is greatly affected by illumination and Earth background, an cnd-to-end attitude estimation method based on Convolutional Neural Network(CNN) is proposed for attitude measurement of non-cooperative targets. In this method, AlexNet and RcsNct arc uscd as the backbone network. Firstly, the füll connection layer at the end of the backbone network is removed, and three füll connection layers are connected in parallel. The attitude angle is estimated by three branch networks. Second-ly, the loss funetion which combines the Classification problem and the regression problem is de-signed. The attitude estimation is limited to a certain ränge by the Classification method, and then the regression method is uscd to further finc tunc the attitude. The attitude Classification loss funetion determines the datum point of the attitude angle, and the attitude regression loss funetion adjusts the estimated angle. Compared with the regression method, this method can effectively rc-duce the average absolute error, Standard deviation and maximum error of attitude estimation. The measurement aecuraey of different backbone networks is compared, the average absolute error ranges from 0. 376 ° to 0. 746 ° and the optimal Standard deviation is 0. 474°.