Modeling irregular small bodies gravity field via extreme learning machines and Bayesian optimization

Close proximity operations around small bodies are extremely challenging due to their uncertain dynamical environment. Autonomous guidance and navigation require fast accurate modeling of the gravitational field for potential on-board computation. In this paper, we investigate a model-based, data-driven approach compute predict acceleration irregular bodies. More specifically, employ Extreme Learning Machine (ELM) theories design, train validate Single-Layer Feedforward Networks (SLFN) capable learning relationship between spacecraft position acceleration. ELM-base neural networks trained without iterative tuning therefore dramatically reducing training time. Analysis performance in constant density models asteroid 25143 Itokawa comet 67/P Churyumov-Gerasimenko show that ELM-based SLFN able learn desired functional both globally selected localized areas near surface. The latter results robust algorithm on-board, real-time calculation gravity needed control close-proximity