Robust spacecraft motion estimation and elevation modeling using a moving binocular head

Publication from Digital

Wolfgang Poelzleitner and Gerhard Paar

SPIE - Cooperative Intelligent Robotics in Space III, Boston, USA , 1/1992


A novel technique for automatic elevation model computation is introduced.  The technique employs a binocular camera system and an algorithm termed hierarchical feature vector matching to derive an elevation model, as well as to compute the interframe correspondences for tracking. It is argued that this algorithm unifies the procedures of range estimation (i.e., stereo correspondence), tracking (i.e., interframe correspondence), and recognition (input/model correspondence). This technique is demonstrated using a physical model of the Mars surface and a binocular camera system with seven geometrical degrees of freedom. This system provides a tool to generate realistic test imagery for the mock-up of a spacecraft approaching the landing site. The trajectory of the spacecraft can be predefined and is controlled by a computer interfaced to seven motorized positioners. Several experiments defined  to estimate the accuracy of the computer vision system are reported .