The surface of Mars exhibits vast expanses of mafic sediments and ancient sedimentary rocks that record signals of climate and environment. To decipher the paleoenvironments, the sediment sources and transport histories must be con-strained, but it is not well known how physical fractionation and aqueous alteration affect mafic sediments during glacial, eolian, and fluvial processes. Semi-Autonomous Navigation for Detrital Environments (SAND-E), a NASA Planetary Science and Technology through Analog Research (PSTAR) project, bridges this gap through studies of sediment-grain properties and mineralogy in the glacio-XRD)-derived mineralogies.

Early Lunar micro-rover missions will be short in duration and have constrained downlink capacity. To maximize the scientific return of these missions, Mission Control is developing technologies to autonomously classify geological features and detect novel features in rover camera imagery, which can be used to support intelligent decision-making for prioritizing data for downlink and instrument targeting. In a recently completed concept study, a trade-off analysis and performance evaluation were conducted for the terrain classifier and novelty detector algorithms across multiple datasets. The terrain classifier developed achieved accuracies of 77%-86% and Intersection over Union (IoU) scores of 0.667-0.680 across 10 different terrain type, on 3 distinct data sets (totalling 928 images), demonstrating the robustness of the approach to varying illumination conditions. In ongoing work, a comprehensive Lunar analogue dataset is being developed to continue prototyping, and the algorithms are being developed on an embedded processor for a flight demonstration opportunity.

Introduction: In almost every planetary surface scientific investigation, the characterization from a rover camera is a common initial step [1]. Mission Control is developing ASAS-CRATERS, a multi-mission technology, to enable automated surface characterization on planetary rover missions, which can benefit a wide range of science investigations, rover navigation, and activities like resource prospecting. It comprises algorithms for terrain classification and novelty detection using convolutional neural networks, and for data aggregation to produce relevant data products for supporting science operations. Built on cutting-edge algorithms and off-the-shelf computing components, it offers lowcost ways to speed up tactical decision-making in nextgeneration commercial lunar missions. Background and Motivation: Autonomy in Science Operations. Upcoming commercial lunar rover missions will have reduced latency, short lifetimes, and constrained bandwidth. This will result in a need for rapid tactical decision-making processes with limited data, leaving little time to analyze data, identify features of interest, and make decisions. Autonomous onboard terrain classification offers a way to downlink light-weight data products and reduce the bottleneck in scientific terrain assessment. Autonomous classification and novelty detection increase the chances of detecting novel/sparse features (ex: lunar outcrop or pyroclasts) that may otherwise be missed when driving and other mission needs are prioritized. Application to Lunar Geology. A rover’s navigation camera can document the surface morphology, morphometry and composition. High-resolution colour images and 3D data from stereo cameras provide information such as the size-frequency distribution and physical characteristics of craters and rocks, regolith properties, and outcrop features. This makes the technology versatile as a tool in support of many science missions. To provide a practical output as a science support tool in upcoming missions, a classification scheme is being developed. See Figure 1 for an example. Technology: ASAS-CRATERS consists of three algorithms implemented on an embedded processor: i) a terrain classifier that uses a deep-learning encoder-decoder style network which classifies each pixel into semantic terrain labels; ii) a novelty detector uses a semisupervised convolutional neural network architecture with an autoencoder module and a binary classifier that work in series; iii) a data aggregator will combine the classification and novel feature outputs on a map that is useable by onboard algorithms and light-weight for more efficient downlink.

ICELAND. P. Sinha, B. H. N. Horgan, R. C. Ewing, E. A. Rampe, M. G. A. Lapotre, M. Nachon, M. Thorpe, C. Bedford, K. Mason, E. Champion, P. C. Gray, E. Reid, , M. Faragalli, Purdue University (sinha37@purdue.edu), Texas A&M University College Station, NASA Johnson Space Center, Stanford University, Lunar and Planetary Institute, University of Arkansas, Duke University, Mission Control Space Services.

HAZARD DETECTION AND AVOIDANCE EXPERIMENTS. M. M. Battler , M. Cross, M. Safdar, K. McIsaac, and M. Faragalli. Dept. of Electrical and Computer Engineering and Centre for Planetary Science and Exploration (CPSX), Western University, 1151 Richmond St., London ON, N6A 3K7, Canada. mbattle@uwo.ca, Dept. of Earth Sciences, Western University, 1151 Richmond St., London ON, N6A 3K7, Canada, Dept. of Civil and Environmental Engineering, Western University, 1151 Richmond St., London ON, N6A 3K7, Canada, Mission Control Space Services, 1125 Colonel By Drive, 311 St. Patrick’s Building Ottawa, ON K1S 5B6.

Rovers are the state of the art for the exploration and detection of past habitability and life on other worlds. One of the most basic functions of a rover is terrain navigation. Information collected by the rover is used autonomously to mitigate terrain hazards such large rocks, while humans qualitatively assess hazardous geologic terrain such as soil type and degree of rock cover. Planetary scientists use the same information to select targets such as drill sites, and for basic scientific analysis such as characterization of rock outcrops. Although the data is complementary, data from terrain analysis for navigation and terrain analysis for scientific investigations are poorly integrated. The lack of integration creates science and operation inefficiencies that limit exploration of habitable environments. As new modes of exploration come online, such as unmanned aerial systems (UAS) (e.g., the Mars Helicopter Scout and Titan Dragonfly), a need exists to integrate terrain data and science analysis to improve operational and scientific outcomes during exploration. We present an overview of a project aimed at evaluating the effectiveness and capability rover and UAS-based semi-automated terrain analysis using the Automated Soil Assessment Systems (ASAS) developed by Mission Control Space Services for navigating, selecting targets for sampling, and characterizing mafic detrital sediments along glacio-fluvial-aeolian sand transport pathways in Iceland. We describe recent advances in automated terrain analysis in sandy environments and scientific uses of terrain assessment from sandy environments. We assess fluvial and aeolian terrains in Iceland and show how terrain analysis data can inform scientific characterization of these environments.