Abstract
This paper presents an innovative template matching technique to perform the pose acquisition of an uncooperative space target, when no prior information is given about the relative position and attitude parameters. Unlike traditional approaches, the algorithm exploits the principal component analysis to restrain the pose search to a 1 Degree-of-Freedom database built on-line. Hence, both the computational effort and the amount of on-board data storage are cut down. Algorithm performance is investigated within a numerical simulation environment reproducing realistic conditions in terms of target configuration and sensor operation. Results demonstrate that the proposed algorithm is able to compute the target pose operating on extremely sparse three-dimensional point clouds (from about 50 to 500 points) provided by a LIDAR system. A success rate of about 95 % has been obtained over a wide range of randomly selected relative attitudes at close range.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have