Incorporating delayed and multi-rate measurements in navigation filter for autonomous space rendezvous

Abstract

In the scope of space missions involving rendezvous between a chaser and a target, vision based navigation relies on the use of optical sensors coupled with image processing and computer vision algorithms to obtain a measurement of the target relative pose. These algorithms usually have high latency time, implying that the chaser navigation filter has to fuse delayed and multi-rate measurements. This article has two main contributions: it provides a detailed modelization of the relative dynamics within the estimation filter, and it proposes a comparison of two delay management techniques suitable for this application. The selected methods are the Filter Recalculation method -which always provides an optimal estimation at the expense of a high computational load- and the Larsen’s method -which provides a faster solution whose optimality lies on stronger requirements. The application of these techniques to the space rendezvous problem is discussed and formalized. Finally, the current article proposes a comparison of the methods based on a Monte-Carlo campaign, aimed at demonstrating whether the loss of performance of Larsen’s method due to its sub-optimality still enables target state robust tracking.