航天器交会计算机视觉系统测距求解新算法

New algorithm of solving for ranges during final approach of spacecraft rendezvous

对航天器交会对接最终逼近段应用的计算机视觉系统,提出目标航天器标志点测距求解新算法.该方法根据标志点构型几何特征,建立非线性测距方程组并构造加权目标函数,按最小二乘问题,采用Gauss-Newton数值迭代法求解测距最佳值.获得测距后,即可应用四元数估算法(QUEST)确定相对姿态与相对位置.对非共面标志点构型(如3点T型与5点锥型)和共面标志点构型(如正方形、矩形、菱形),目标函数均含标志点间距比率关系项;对共面构型,目标函数还包含共面条件项.对共面标志点四边形构型,利用对角线交点的虚影像坐标确定测距求解迭代初值.大量模拟计算结果表明,提出的测距求解算法对共面与非共面各种标志点构型普遍适用,满足最终逼近段相对状态确定要求.

A new algorithm of solving for ranges during the final translation phase of spacecraft rendezvous and docking using computer vision system was presented to determine relative attitude and position between non-planar （ such rectangle and rho as three-point T-type and five-point pyramid type） and co-planar patterns （such as square, rnbus ） the objective function contained terms concerning ratios of distances between the target spots ; but for co-planar targets, the objective function had another terms which describe co-planar feature. Gauss-Newton method was applied to solve for the optimum range solutions of the least squares problem. For quadrilateral patterns, the initial values for iterations were given by coordinates of virtual image of the intersection point of two diagonal lines. Extensive simulations, conducted for various target patterns with image coordinates errors, show that the range algorithm presented is quite effective for the relative state determination.