Descent and Landing Phase: Vision-Based Robust Spacecraft Motion Estimation and Elevation Modeling

Publication from Digital

Gerhard Paar, Wolfgang Polzleitner

CNES, Missions, Technologies, and Design of Planetary Mobile Vehicles, Toulouse, France , 1/1992


A technique for automatic elevation model computation, which uses a binocular camera system and an algorithm termed 'hierarchical feature vector matching' to derive an elevation model (or range map), as well as to compute the interframe correspondences for tracking, is introduced. It is robust in terms of geometric distortions between the stereo partners and relies on hierarchical image representations usable for the realtime demand during descent and landing. It is argued that this algorithm unifies the procedures of range estimation (stereo correspondence, tracking (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 six geometrical degrees of freedom. This system provides a valuable tool to generate realistic test imagery for the mockup of a spacecraft approaching the landing site. The trajectory of the spacecraft can be predefined and is controlled by a computer interfaced to six motorized positioners. Several experiments that were defined to estimate the accuracy of the computer vision system are reported.