Generalized Angles-Only Navigation Architecture for Autonomous Distributed Space Systems
J.L. Sullivan, Adam W. Koenig, Justin Kruger, Simone D’Amico
Abstract
This paper addresses the development and verification of an estimation architecture for autonomous relative navigation in multi-satellite systems using angles-only measurements. Unlike traditional angles-only navigation approaches, which are generally hindered by poor handling of observability constraints, this prototype solution is designed to be flexible and applicable to a multitude of relevant missions. The estimator converges accurately and robustly in varying orbital environments without requiring maneuvers by exploiting system nonlinearities using an unscented Kalman filter that is streamlined by formulating the measurement and dynamics models using relative orbital elements. While the majority of prior work has focused on only solving for the relative motion of a single target, the approach here enables estimation of several additional parameters. Key design trades are evaluated for estimating additive sensor biases, target ballistic properties in environments with strong nonconservative perturbations, adaptive process noise statistics, and the absolute orbit of the observing satellite. The architecture is generalized for relative navigation with multiple targets, and for decentralized estimation using multiple coordinated observers. These filter strategies are used to inform the development of a novel on-orbit demonstration of angles-only navigation known as the Starling Formation-flying Optical eXperiment (StarFOX) in partnership with NASA. Key hardware-in-the-loop algorithm verification results for StarFOX-specific scenarios are provided.