Angles-Only Initial Orbit Determination via Multivariate Gaussian Process Regression

• DOI: 10.3390/electronics11040588
• Publication Date: 2022-02-16T03:43:22Z
• Authors: David Schwab; Puneet Singla; Sean O’Rourke
• Keywords: Ground-Penetrating Radar; Orbit Determination; Observer (physics); Orbit (dynamics)

Vital for Space Situational Awareness, Initial Orbit Determination (IOD) may be used to initialize object tracking and associate observations with a tracked satellite. Classical IOD algorithms provide only point solution are sensitive noisy measurements certain target-observer geometry. This work examines the ability of Multivariate GPR (MV-GPR) accurately perform quantify associated uncertainty. Given perfect test inputs, MV-GPR performs comparably simpler multitask learning algorithm classical Gauss–Gibbs in terms prediction accuracy. It significantly outperforms uncertainty quantification due direct handling output dimension correlations. A moment-matching provides an analytic input noise problem under assumptions. The is adapted formulation shown effective tool added shows that can viable quantified which robust observation traditionally challenging orbit-observer geometries.