Onboard software for near-target optical navigation: a comparative analysis of conventional computer graphics and image segmentation-detection techniques

Autonomous spacecraft navigation is critical for reducing operational costs of deep-space inter-planetary missions, particularly when traditional ground-based tracking is limited. Optical navigation enables efficient spacecraft guidance using minimal resources by leveraging advanced image processing techniques. Embedded Software for Optical Navigation integrates line-of-sight (LoS) measurements of celestial beacons with fusion of star trackers, utilizing an Extended Kalman Filter (EKF) to achieve precise trajectory estimation during a hyperbolic interplanetary approach. During target approach, flyby, or landing, stars and surface targets serve as critical reference points for precise spacecraft navigation. Traditional Computer graphics based method involves locating target beacon edges and removing terminators for accurately deriving target location, and, during night phase, this approach changes to Star occultation-based position computation. Advanced image processing techniques, including morphological operations, image segmentation, and object detection, offer alternative approaches for optical navigation, improving spacecraft guidance with high accuracy and resource efficiency. The satellite position accuracy achieved is within 0.015% of satellite range in day or night phases of the target celestial body. This paper addresses both theoretical foundations and software implementation challenges of Optical Navigation encountered in resource-constrained spacecraft environments.