RSOnet: An Image-Processing Framework for a Dual-Purpose Star Tracker as an Opportunistic Space Surveillance Sensor

Area of Science:

• Space Domain Awareness
• Astrodynamics
• Computer Vision

Background:

• The number of resident space objects (RSOs) in Earth's orbit is rapidly increasing, necessitating advanced detection and tracking capabilities.
• Current space surveillance systems face challenges in scaling to manage the growing orbital debris and satellite population.
• Star trackers, traditionally used for attitude determination, offer a cost-effective and flight-proven platform for augmenting space situational awareness.

Purpose of the Study:

• To present RSOnet, an image-processing framework designed to leverage star trackers for detecting, tracking, classifying, and characterizing resident space objects (RSOs).
• To demonstrate the feasibility of using star trackers as opportunistic sensors for space domain awareness in crowded low Earth orbit (LEO).
• To introduce a novel approach for enhancing space situational awareness with existing, cost-effective hardware.

Main Methods:

• Development of RSOnet, an image-processing framework integrating a convolutional neural network (CNN) model for RSO detection.
• Implementation of a graph-based multi-object classifier within the RSOnet framework for improved tracking and data association.
• Validation of RSOnet's performance using real-world detection data from a space-based star-tracker-like imager.

Main Results:

• Demonstrated the capability of star trackers to serve as a viable source for RSO detections in LEO.
• Successfully applied RSOnet to real space-based imagery, validating its performance in detecting and characterizing resident space objects.
• Showcased the effectiveness of the RSOnet CNN architecture and graph-based classifier in a practical space surveillance scenario.

Conclusions:

• Star trackers can be effectively repurposed as opportunistic sensors for space domain awareness, significantly enhancing space situational awareness capabilities.
• RSOnet provides a scalable and cost-effective solution for detecting and characterizing resident space objects in increasingly congested orbital environments.
• The research validates the potential of leveraging existing flight-proven hardware for advanced space surveillance missions.