Statistics – Computation
Scientific paper
Dec 2005
adsabs.harvard.edu/cgi-bin/nph-data_query?bibcode=2005agufmin33d..08g&link_type=abstract
American Geophysical Union, Fall Meeting 2005, abstract #IN33D-08
Statistics
Computation
5464 Remote Sensing, 5494 Instruments And Techniques, 6225 Mars, 6297 Instruments And Techniques
Scientific paper
Mars rovers and orbiters currently collect far more data than can be downlinked to Earth, which reduces mission science return; this problem will be exacerbated by future rovers of enhanced capabilities and lifetimes. We are developing onboard intelligence sufficient to extract geologically meaningful data from spectrometer measurements of soil and rock samples, and thus to guide the selection, measurement and return of these data from significant targets at Mars. Here we report on techniques to construct mineral detectors capable of running on current and future rover and orbital hardware. We focus on carbonate and sulfate minerals which are of particular geologic importance because they can signal the presence of water and possibly life. Sulfates have also been discovered at the Eagle and Endurance craters in Meridiani Planum by the Mars Exploration Rover (MER) Opportunity and at other regions on Mars by the OMEGA instrument aboard Mars Express. We have developed highly accurate artificial neural network (ANN) and Support Vector Machine (SVM) based detectors capable of identifying calcite (CaCO3) and jarosite (KFe3(SO4)2(OH)6) in the visible/NIR (350-2500 nm) spectra of both laboratory specimens and rocks in Mars analogue field environments. To train the detectors, we used a generative model to create 1000s of linear mixtures of library end-member spectra in geologically realistic percentages. We have also augmented the model to include nonlinear mixing based on Hapke's models of bidirectional reflectance spectroscopy. Both detectors perform well on the spectra of real rocks that contain intimate mixtures of minerals, rocks in natural field environments, calcite covered by Mars analogue dust, and AVIRIS hyperspectral cubes. We will discuss the comparison of ANN and SVM classifiers for this task, technical challenges (weathering rinds, atmospheric compositions, and computational complexity), and plans for integration of these detectors into both the Coupled Layer Architecture for Robotic Autonomy (CLARAty) system and the Onboard Autonomous Science Investigation System (OASIS) at JPL.
Bornstein Benjamin
Castano Rebecca
Gilmore Martha S.
Greenwood J.
Merrill Mike
No associations
LandOfFree
Onboard autonomous mineral detectors for Mars rovers does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.
If you have personal experience with Onboard autonomous mineral detectors for Mars rovers, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Onboard autonomous mineral detectors for Mars rovers will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-752069