Aas 21-336 comparing optical tracking techniques in distributed asteroid orbiter missions using ray-tracing

Missions to asteroids have traditionally relied on the Deep Space Network and Light Detection and Ranging (LIDAR) for landmark tracking and mapping. Optical tracking and multi-agent structure from motion show great potential for improving autonomy and reducing hardware requirements in such missions. However, there is a lack of testing and simulation in this area, especially because validation is difficult. In this work, keypoint descriptors are compared to a targetspecific landmark descriptor (craters) for optical tracking within a singleand multi-satellite space mission orbiting an asteroid. Additionally, a novel method to validate correlations between any landmark descriptors using ray-tracing is developed. Optical tracking methods are evaluated in simulation to identify tracking limitations. Simulated test cases are validated using the novel ray-tracing technique. The optical tracking methods are further validated using images from the Near Earth Asteroid Rendezvous (NEAR) Shoemaker mission to Eros. Results indicate that SIFT is the most accurate optical tracking technique across monolithic and distributed space system simulations. However, SURF has the most robustness to lighting conditions and produces the most accurate 3D points from stereo-vision. Craters are shown to be an inconsistent landmark detection method, and normalized cross-correlation appears to be largely unreliable for crater correlation. Despite issues with some optical tracking methods, matching ray-traced landmarks shows great potential for optical tracking validation.