基于对偶四元数的航天器单目视觉导航算法

Monocular vision-based navigation algorithm for spacecraft using dual quaternion

为了估计两航天器的相对姿态与轨道运动,利用对偶数推导并给出了航天器相对动力学方程,该方程一体化描述了航天器的相对运动,考虑了姿态与轨道之间的耦合影响.在对偶代数框架内,基于目标航天器的特征线在视觉相机平面的投影建立了导航算法的测量模型.最后通过对系统动力学模型以及测量模型的线性化,应用EKF滤波算法对航天器的相对运动状态进行了估计.仿真结果表明,本文的算法能对航天器相对运动进行较高精度的估计,且收敛速度较快.

The monocular vision-based navigation algorithm for spacecraft is addressed in this paper.Based on dual number,coupled dynamics equations of relative position and attitude are derived,which can describe the coupling effect between the rotational motion and translational motion.The measurement model is proposed on the basis of the projection on the 2-D plane of the line fixed on the target spacecraft.And the Extended Kalman Filtering method is applied to the estimation problem.Numerical simulations are performed to demonstrate that the proposed approaches can provide accurate and fast estimation of relative position and attitude.