Planetary Mapping Using Deep Learning: A Method to Evaluate Feature Identification Confidence Applied to Habitats in Mars-Analog Terrain.

The goals of Mars exploration are evolving beyond describing environmental habitability at global and regional scales to targeting specific locations for biosignature detection, sample return, and eventual human exploration. An increase in the specificity of scientific goals-from follow the water to find the biosignatures-requires parallel developments in strategies that translate terrestrial Mars-analog research into confident identification of rover-explorable targets on Mars. Precisely how to integrate terrestrial, ground-based analyses with orbital data sets and transfer those lessons into rover-relevant search strategies for biosignatures on Mars remains an open challenge. Here, leveraging small Unmanned Aerial System (sUAS) technology and state-of-the-art fully convolutional neural networks for pixel-wise classification, we present an end-to-end methodology that applies Deep Learning to map geomorphologic units and quantify feature identification confidence. We used this method to assess the identification confidence of rover-explorable habitats in the Mars-analog Salar de Pajonales over a range of spatial resolutions and found that spatial resolutions two times better than are available from Mars would be necessary to identify habitats in this study at the 1-σ (85%) confidence level. The approach we present could be used to compare the identifiability of habitats across Mars-analog environments and focus Mars exploration from the scale of regional habitability to the scale of specific habitats. Our methods could also be adapted to map dome- and ridge-like features on the surface of Mars to further understand their origin and astrobiological potential.

Spatial Spectroscopic Models for Remote Exploration.

Ancient hydrothermal systems are a high-priority target for a future Mars sample return mission because they contain energy sources for microbes and can preserve organic materials (Farmer, 2000 ; MEPAG Next Decade Science Analysis Group, 2008 ; McLennan et al., 2012 ; Michalski et al., 2017 ). Characterizing these large, heterogeneous systems with a remote explorer is difficult due to communications bandwidth and latency; such a mission will require significant advances in spacecraft autonomy. Science autonomy uses intelligent sensor platforms that analyze data in real-time, setting measurement and downlink priorities to provide the best information toward investigation goals. Such automation must relate abstract science hypotheses to the measurable quantities available to the robot. This study captures these relationships by formalizing traditional "science traceability matrices" into probabilistic models. This permits experimental design techniques to optimize future measurements and maximize information value toward the investigation objectives, directing remote explorers that respond appropriately to new data. Such models are a rich new language for commanding informed robotic decision making in physically grounded terms. We apply these models to quantify the information content of different rover traverses providing profiling spectroscopy of Cuprite Hills, Nevada. We also develop two methods of representing spatial correlations using human-defined maps and remote sensing data. Model unit classifications are broadly consistent with prior maps of the site's alteration mineralogy, indicating that the model has successfully represented critical spatial and mineralogical relationships at Cuprite. Key Words: Autonomous science-Imaging spectroscopy-Alteration mineralogy-Field geology-Cuprite-AVIRIS-NG-Robotic exploration. Astrobiology 18, 934-954.