Anonymous Feature Processing for Efficient Onboard Navigation
James S. McCabe, Kyle J. DeMars · AIAA Scitech 2020 Forum · 2020
Recent developments using “anonymous features” permit feature-based navigation techniques, such as lidar retroreflector tracking or vision-based terrain relative navigation, to be performed without the need of a step that explicitly identifies the observed features, thereby eliminating sensitivity to misidentification errors and reducing the reliance upon potentially computationally burdensome identification techniques. This paper seeks to further advance this “anonymous” approach to feature-based navigation by preparing a formulation tailored for flight applications, i.e. focusing upon computational efficiency and numerical stability, to further assess and improve its applicability to real flight applications. In particular, this work explores some of the key details necessary to implement the new anonymous feature processing techniques in square-root and UDU factorized filter formulations to widen its applicability to systems constructed around factorized covariance representations.