航天器交会视觉导航系统相对状态确定算法

对航天器交会对接最终逼近段计算机视觉系统,阐述相对位移与相对姿态确定算法。该算法先求解测距(目标航天器标志点离追踪航天器相机镜头中心的距离),然后应用四元数估算法,由测距确定相对状态参数最佳值。特别是,一种测距求解新算法被提出。此外,还推导出一种利用偏差传递法由影像点偏差直接求出相对状态偏差的方法,这种方法不包含相对状态确定算法的方法误差,可用以评估相对状态确定算法的有效性。大量模拟算例显示,相对状态确定算法的精度与偏差传递法计算结果基本一致,这表明,相对状态确定算法对各种标志点构型普遍适用,且方法误差很小,相对状态确定精度高。不同标志点构型比较表明,非共面构型优于共面构型,可适应更大的影像点坐标偏差与更远的相对距离;基准面内的标志点应尽可能均匀对称分布,间距尽可能加大;非共面构型标志点可取为3个(含1个突出点)、4个或5个(含1~2个突出点);共面构型标志点取为4个适合(其中任意3点不共线);在4点共面构型中,正方形构型相对状态确定精度最高。

Algorithm of Determining Relative States for Computer Vision System Used During Spacecraft Rendezvous and Docking

An algorithm of determining relative attitude and position between target and chase vehicles by quaternion estimation via ranges of feature points was interpreted,especially a new algorithm of solving for the ranges is presented.Also an error-transmission method was given to obtain the relative state errors from the image coordinate deviations directly,which can be used for evaluating the algorithm of determining relative state.Computational simulation results show that the precision of the algorithm provided almost reaches the theoretic values given by the error-transmission method.As to patterns of feature points,non-planar patterns(with 3 to 5 feature points) are better than co-planar patterns(with 4 feature points at least),and feature points in the base-plane should be distributed symmetrically.Among four-point co-planar patterns,square patterns are better than others.