【摘要】 传统的未知地标导航方法是通过对同一地标的多次观测来估计航天器的状态。相较于使用已知地标，该方法无需进行地标识别与地标库建立，但由于地标状态未知，该方法收敛速度较慢，导航精度也较差。针对上述问题，提出一种同时估计航天器位置、速度及地标位置的多未知地标导航方法，在模拟的遥感图像序列中，使用Lucas-Kanada稀疏光流法跟踪多个未知地标，将地标视线矢量作为观测量，利用扩展卡尔曼滤波器对航天器和地标的状态进行估计。以轨道高度300 km的卫星进行仿真验证，仿真结果表明，所提方法可实现300 s内快速收敛，滤波收敛后卫星位置和速度误差分别为402.67 m、0.481 m/s，地标位置误差为679.88 m，在提高收敛速度的基础上，获得了较高导航精度。更多还原

【Abstract】 Traditional navigation method for unknown landmarks relies on multiple observations of the same landmark to estimate the state of the spacecraft. Compared with the navigation method using known landmarks, the method does not require image recognition of landmarks and establishment of landmark library. However, since the state of the landmark is unknown, the method converges slowly and has poor navigation accuracy. In response to the above problems, a multi-unknown landmark navigation method that simultaneously estimates the position and velocity of the spacecraft and the landmark position is proposed. In the simulated remote sensing image sequence, the Lucas□Kanada optical flow method is used to track multiple unknown landmarks, the landmark sight line vector is used as the observations, and the states of the spacecraft and landmark are estimated by the extended Kalman filter. A satellite with an orbital height of 300 km is used for simulation verification. The simulation results show that the proposed method can achieve rapid convergence within 300 s. After the filtering converges, the position and velocity estimation errors of the satellite are 402.67 m and 0.481 m/s respectively, and the estimated error of the landmark location is 665.2 m, which not only improves the convergence speed but also obtains higher navigation accuracy.更多还原