Evaluation of Machine Learning Methodologies for Novelty-based Target Selection in Planetary Imaging Data Sets: Examples from the Mars Science Laboratory Mission

Monday, 7 December 2020
Favour Nerrise1, Hannah Rae Kerner2, Kiri Wagstaff3, Steven Lu4, Raymond Francis3, Umaa Rebbapragada5 and James F Bell III6, (1)University of Maryland, College Park, College Park, MD, United States, (2)University of Maryland College Park, Geography, College Park, MD, United States, (3)Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, United States, (4)Jet Propulsion Laboratory, Machine Learning and Instrument Autonomy, Pasadena, CA, United States, (5)NASA Jet Propulsion Laboratory, Pasadena, CA, United States, (6)Arizona State University, Tempe, AZ, United States
In-situ novelty-based target selection of scientifically interesting (“novel”) surface features can expedite follow-up observations and new discoveries for the Mars Science Laboratory (MSL) rover and other planetary exploration missions. This study aims to identify which methods perform best for detecting novel surface features in MSL Navcam images for follow-up observations with the ChemCam instrument, as a complement to the existing Autonomous Exploration for Gathering Increased Science (AEGIS) onboard targeting system. We created a dataset of 6630 candidate targets within Navcam grayscale images acquired between sols 1343-2578 using the Rockster algorithm. These were the same target candidates considered by AEGIS, chosen to enable direct comparison to past AEGIS target selections. We employed five novelty detection methods, namely Discovery via Eigenbasis Modeling of Uninteresting Data (DEMUD), Isolation Forest, Principal Component Analysis (PCA), Reed-Xiaoli (RX) detector, and Local RX. To evaluate the algorithm selections, a member of the MSL science operations team independently identified candidate targets that represented example scenarios of novel geology that we would desire an algorithm to identify, such as layered rocks, light-toned unusual textures, and small light-toned veins. We compared these methods to selections made by AEGIS and a random baseline. Initial experiments for three scenarios showed that Local RX most frequently prioritized novel targets, followed by DEMUD and AEGIS. Our next steps in this study include evaluating input feature representations other than pixel intensities (e.g., Histogram of Oriented Gradients features), performing additional experiments to evaluate novel target prioritization performance, and selecting target candidates in Mastcam color images.