Autonomous Horizon-based Asteroid Navigation With Observability-constrained Maneuvers

TL;DR

This paper introduces an observability-constrained Lyapunov controller for autonomous asteroid navigation that maintains optical navigation reliability by avoiding poor observability regions, significantly improving orbit attainment success rates.

Contribution

It develops a novel control framework that integrates observability constraints into asteroid orbit control, enhancing autonomous navigation robustness in low gravity environments.

Findings

Up to 94% improvement in target orbit attainment without observability violations.

Enhanced robustness over conventional control methods.

Effective in both spherical and ellipsoidal asteroid scenarios.

Abstract

Small body exploration is a pertinent challenge due to low gravity environments and strong sensitivity to perturbations like Solar Radiation Pressure (SRP). Thus, autonomous methods are being developed to enable safe navigation and control around small bodies. These methods often involve using Optical Navigation (OpNav) to determine the spacecraft's location. Ensuring OpNav reliability would allow the spacecraft to maintain an accurate state estimate throughout its mission. This research presents an observability-constrained Lyapunov controller that steers a spacecraft to a desired target orbit while guaranteeing continuous OpNav observability. We design observability path constraints to avoid regions where horizon-based OpNav methods exhibit poor performance, ensuring control input that maintains good observability. This controller is implemented with a framework that simulates small body dynamics, synthetic image generation, edge detection, horizon-based OpNav, and filtering. We evaluate the approach in two representative scenarios, orbit maintenance and approach with circularization, around spherical and ellipsoidal target bodies. In Monte Carlo simulations, the proposed approach improves the rate of attaining target orbits without observability violations by up to 94% compared to an unconstrained Lyapunov baseline, demonstrating improved robustness over conventional methods.