Autonomous detection and tracking of particles around small bodies with Artificial Intelligence

Navigating a spacecraft around small bodies, such as asteroids or comets, presents formidable challenges due to these bodies’ unique environmental conditions. The weak and highly irregular gravitational fields resulting from their small mass and non-uniform shapes complicate traditional navigation techniques. In this context, innovative ideas are needed to enhance mission success rates. For instance, the unexpected discovery of particles ejected from the surface of asteroid Bennu enabled scientists to analyze the gravitational field around the asteroid by reconstructing these particles’ trajectories. Using the latter, the gravity model of small bodies can be estimated, supporting safer and more accurate maneuvering in complex environments. Moreover, in Bennu’s case, the naturally ejected particles provide crucial insights into the asteroid’s composition and activity, offering information about the physical properties of the surface. A technique that reliably detects and tracks naturally formed or artificially crafted particles could be instrumental in mapping gravitational fields and assessing the surface composition of similar small bodies across the solar system, ultimately broadening our understanding and improving mission outcomes. Traditional image processing and tracking methods are often computationally intensive and struggle with the complexity of small-body environments, especially when numerous particles need to be tracked simultaneously against a complex background of lighting and shadows. In this context, AI-based image processing offers a promising solution by automating the detection processes, significantly reducing computational complexity while maintaining good accuracy. This thesis introduces an end-to-end AI-powered pipeline capable of autonomously detecting and tracking objects orbiting around small bodies. The algorithm is tested on real data, demonstrating its competitive performance and providing a scalable solution for future similar applications.