Close Range Tracking of An Uncooperative Space Target in a Sequence of PMD Images

Abstract

This paper presents a pose estimation routine for tracking attitude and position of an uncooperative tumbling spacecraft during close rendezvous phase. The key aspect is a possible use of a Photonic Mixer Device (PMD) sensor instead of existent state-of-the-art visual rendezvous sensors. We propose a hybrid navigation technique which employs an iterative closest point algorithm with a reverse calibration technique and image processing procedure based on a given set of N 2D-to-3D correspondences using only measurements from the PMD sensor. Applying this technique we gain a redundant estimation of the target's state vector in real-time without hardware redundancy. It would increase the reliability of the navigation system working under influence of different environmental factors with the PMD sensor. The capabilities of the proposed procedure are demonstrated with a real-time testbed for proximity operations and with images acquired from the PMD camera.