ESA GNC Conference Papers Repository

Hazard avoidance techniques allow safe landing in dangerous or insufficiently characterized areas on distant planets. Hazard mapping detects or estimates ground features such as slope, rocks and shadows. Site selection chooses a suitable landing site based on hazard mapping data, propulsion capabilities and mission constraints. Trajectory planning and guidance compute the trajectory and steer the vehicle towards the selected target. Recent developments on vision based slope estimation and site selection are presented. Shape from shading (SFS) and stereo by motion (SBM) techniques are described: they compute slope from a reconstructed 3D map of the terrain. Tests on synthetic images for different conditions (altitude, illumination, etc.) showed that SFS was more accurate when correctly tuned but lacked robustness. SBM is more robust but accuracy is poor in the region of the epipole. The use of 3D coordinates of feature points tracked by vision-based navigation is highly beneficial. The final trade-off will be made after tests on real images. A new, dynamic and adaptable site selection algorithm was implemented based on fuzzy multi criteria decision making methods. The proposed generic framework allows an easy tuning and the addition of new criteria in a flexible way. Candidate landing sites are ranked according to the degrees of satisfaction of criteria and past information obtained at previous iterations. The incorporation of past information makes the algorithm more discriminative. Landing areas of different sizes (in image pixels) are considered depending on the altitude. The validity of the concept is demonstrated on test scenarios.