A Switched Systems Approach to Image-Based Feature Tracking for Autonomous Satellite Inspection

TL;DR

The paper tackles autonomous inspection of a chief satellite by a deputy spacecraft under illumination and field-of-view constraints without GPS. It couples a memory-based regression observer (MRE) to learn the chief’s feature structure and a switched, information-maximizing controller to guide the deputy toward illuminated feature clusters. Safety is guaranteed using observer-robust control barrier functions within a finite-horizon constrained optimization, ensuring convergence and collision avoidance. A k-means planner provides illumination-aware goal directions, while sequential feature tracking and dwell-time analysis enable robust relative localization. Simulation results validate stability, observability improvement, and safe, information-rich inspection trajectories over time.

Abstract

This paper presents an information-based guidance and control architecture for an autonomous deputy spacecraft tasked with inspecting a chief satellite in orbit. The primary objective is for the deputy spacecraft to maximize information gain while tracking features on the chief satellite. The deputy spacecraft needs to respect various constraints such as illumination, field-of-view (FOV), fuel limitations, and avoidance regions. Additionally, the absence of GPS information poses a significant challenge for relative localization within the space environment. To learn the structure of the chief satellite while achieving relative self-localization and maximizing the information gain, this paper integrates a memory regression extension (MRE)-based distance observer with an information-maximizing adaptive controller. The distance observer utilizes feedback from a camera. A switched systems approach is used to determine the minimum dwell time required for a feature to remain within the FOV of the camera to ensure accurate estimation. A k-means clustering algorithm acts as a high-level planner to intermittently generate goal locations that guide the deputy spacecraft toward the nearest cluster of uninspected points on the chief satellite, subject to illumination and FOV constraints. A Lyapunov-based stability analysis is conducted to analyze the developed architecture, and simulation results validate the theoretical results of the paper.

Figures (12)

• Figure 1: Schematic of the relative position of the deputy spacecraft $\mathcal{O}_{B}$, chief satellite $\mathcal{O}_{H}$, and the goal location $\mathcal{O}_{G}$ is shown in Hill’s reference Frame.
• Figure 2: Visualization of the light emanating from the sun on the points on the surface of the chief satellite: Yellow dots represent inspection points that are illuminated by the sun, while blue dots indicate points that are not illuminated.
• Figure 3: Sequential feature-based localization with feature switching upon FOV exit after satisfying dwell time.
• Figure 4: Avoidance Regions
• Figure 5: Trajectory of the deputy spacecraft over 1000 seconds while inspecting points on the chief satellite; inspected points are shown in green, and uninspected points in red

Theorems & Definitions (19)

• Remark 1
• Remark 2
• Remark 3
• Remark 4
• Remark 5
• Remark 6
• Theorem 1
• proof
• Corollary 1
• proof