Data fusion strategies for hazard detection and safe site selection for planetary and small body landings

Abstract

well as decision-level data (fusing hazard maps from multiple sensors onto a single image space, with a single grid orientation and spacing). The second method presented is a hybrid reasoning fusion, the HRF, in which innovative algorithms replace the decision-level functions of the previous method, by combining three different reasoning engines-a fuzzy reasoning engine, a probabilistic reasoning engine and an evidential reasoning engine-to produce safety maps. Finally, the third method presented is called Intelligent Planetary Site Selection, the IPSIS, an innovative multi-criteria, dynamic decision-level data fusion algorithm that takes into account historical information for the selection of landing sites and a piloting function with a non-exhaustive landing site search capability, i.e., capable of finding local optima by searching a reduced set of global maps. All the discussed data fusion strategies and algorithms have been integrated, verified and validated in a closed-loop simulation environment. Monte Carlo simulation campaigns were performed for the algorithms performance assessment and benchmarking. The simulations results comprise the landing phases of Mars and Phobos landing mission scenarios.

Data fusion approach for eucalyptus trees identification

“…Data fusion approaches, based on fuzzy logic techniques, are emerging as a technique for land classification to perform correct reasoning inferences (Hyder, Shahbazian, and Waltz 2012;Santos, Andre Mora, and Joao 2016). The FIF algorithm -basis of our approachis based on fuzzy logic and specialized decision-making aggregation operators and was applied to spacecraft landing with hazard avoidance (Bourdarias et al 2010;Câmara et al 2015) as well as for land cover classification (A. D. Mora et al 2015;A. D.2016;A.2017).…”

“…'low-slope', 'height'). It is also noteworthy that the FIF algorithm was derived from a hazard avoidance landing of spaceships on planets (Bourdarias et al 2010;Câmara et al 2015) and recently was partially applied to specific remote sensing problems (Mora et al 2017(Mora et al , 2015. It should also be noted that in our approach, only two steps of FIF are used: Normalization (step 1) and Aggregation (step 4) because there is no need to filter imprecision (step 2) since confidence on different bands is identical and step 3 (relative weights) because all criteria have equal weights.…”

Feature Density Based Terrain Hazard Detection for Planetary Landing

“…These studies make great contribution to the development of terrain hazard detection for planetary landing. For the complex multitypes terrain, the combination of these methods will not be suitable for the low-cost/small-size missions [20]- [23]. Hence, it is important to find a unified terrain hazard detection method, which satisfied the requirements of complex and scientific terrain landing safety.…”