LIDAR-Based Relative Navigation and 3D Target Reconstruction for Close Proximity Operations With Unknown Target

This paper presents a LIDAR-based relative navigation architecture designed to deal with uncooperative and unknown targets. It foresees two sequential functional blocks. The former aims at reconstructing a geometric model of the target by recursively registering consecutive point clouds collected over a certain time frame. In this block, a correction step is applied at a frequency lower than the LIDAR update rate in order to bound to odometry-related drift in the estimated relative trajectory. The resulting reconstructed model is then used within the latter block which implements model-based registration algorithms to accurately estimate the target pose. The performance of the proposed architecture is assessed in a numerical simulation environment which realistically reproduces both the relative orbital dynamics and the measurement process of a spaceborne LIDAR system and assuming ENVISAT as target. Results of the reconstruction phase show a Chamfer distance of few square centimeters between the reference geometry and the reconstructed model. In the relative navigation phase dm-level and degree-level accuracy is achieved in the pose estimated with the reconstructed model.