Deep-space optical navigation is among the most promising techniques to autonomously estimate the position of a spacecraft in deep space. The method relies on the acquisition of the line-of-sight directions to a number of navigation beacons. The position knowledge depends upon the tracked objects. This paper elaborates on the impact of the observation geometry to the overall performances of the method. A covariance analysis is carried out considering beacons geometry as well as pointing and input errors. A performance index is formulated, and criteria for an optimal beacons selection are derived in a scenario involving two measurements. A test case introducing ten available beacons pairs is used to prove the effectiveness of the developed strategy in selecting the optimal pair, which leads to the smallest achievable error.

深空光学导航是自主估计航天器在深空中位置的最有前途的技术之一。该方法依赖于对多个导航信标的视线方向的获取。位置知识取决于被跟踪的对象。本文详细阐述了观测几何形状对方法整体性能的影响。考虑信标几何以及指向和输入错误进行协方差分析。在涉及两次测量的情况下，制定了性能指标，并得出了最佳信标选择的标准。一个测试案例介绍了十个可用的信标对，以证明所开发策略在选择最佳对时的有效性，这导致最小的可实现误差。