AEGIS Automated Targeting for the MSL ChemCam Instrument

AEGIS Automated Targeting for the MSL ChemCam Instrument
Tara Estlin1, Robert C. Anderson1, Diana Blaney1, Michael Burl1, Benjamin Bornstein1, Rebecca Castaño1, Lauren De Flores1, Daniel Gaines1, Michele Judd1, David R. Thompson1, and Roger Wiens2
1Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, This research was carried out at
the Jet Propulsion Laboratory,
California Institute of Technology.
Copyright
2013
California
Institute of Technology. All Rights
Reserved.
Pasadena, CA 91109 USA
2Los Alamos National Laboratory, Los Alamos, NM 87545 USA
Poster #P51G‐1801
Contact: [email protected]
AEGIS Overview:
The Autonomous Exploration for Gathering Increased Science (AEGIS) system enables automated data collection by planetary rovers. AEGIS software was first used on the Mars Exploration Rover (MER) mission’s Opportunity rover in January
2010 and has since successfully demonstrated automated onboard targeting for a wide range of science targets, which were selected based on scientist‐specific objectives. AEGIS is now being applied for use with the Mars Science Laboratory
(MSL) mission ChemCam spectrometer. ChemCam uses a Laser Induced Breakdown Spectrometer (LIBS) to analyze the elemental composition of rocks and soil from up to seven meters away. ChemCam’s tightly‐focused laser beam (350‐550 um)
enables targeting of very fine‐scale terrain features. AEGIS is being applied in two ways to help ChemCam collect valuable science data which are described below.
AEGIS enables targeting of narrow field‐of‐view instruments without a ground communications cycle. Without an AEGIS capability, data must be collected either blindly or through the following series of steps: images are transmitted from the
rover to the operations team on Earth; scientists manually analyze the images, select geological targets for the rover’s remote‐sensing instruments, and then generate a command sequence to execute the new measurements. This process
typically takes 1 to several sols (or Martian days). With AEGIS, targeted measurements can be collected autonomously but still directed by scientist priorities. AEGIS is a significant step towards allowing a robotic vehicle on another planet to make
intelligent science choices. AEGIS software uses scientist input and onboard data analysis techniques to select high‐quality science targets without requiring communication with the ground team. This approach allows the rover to autonomously
select and sequence targeted observations in an opportunistic fashion, which is particularly applicable for narrow field‐of‐view instruments (such as the MER Panoramic camera, the MER Mini‐TES spectrometer, and the 2011 Mars Science
Laboratory (MSL) ChemCam spectrometer). We provide an overview of the AEGIS automated targeting capability and describe how it is being applied for use with the MSL ChemCam instrument.
MSL Application 2: ChemCam Pointing Refinement for Data Acquisition on Small Targets
A second application is to enable intelligent pointing refinement of ChemCam when acquiring data of small
targets, such as veins or concretions that are only a few millimeters wide. Due to backlash and other pointing
challenges, it can often require several downlink cycles for LIBS measurements to be acquired on small targets.
Often targets must first be imaged using the high resolution ChemCam Remote Micro Imager (RMI) and then
ground analysis performed to enable a fine‐tuned pointing correction on the next commanding cycle. AEGIS is
being applied to analyze RMI images onboard and automatically determine the pointing refinement necessary to
acquire LIBS data on small targets. This significantly decreases the amount of time and resources required to
acquire ChemCam data on such targets.
Work is currently in progress to adapt the AEGIS algorithm for the below applications and integrate the system with MSL flight software. Some testing has already been begun on MSL flight testbeds. Once integration and testing is complete,
AEGIS will be uploaded to the spacecraft for operational use in 2014.
AEGIS Onboard Process for Pointing Refinement:
MSL Application 1: Automated Targeting of ChemCam to Obtain New Rock Measurements
The first application is to enable automated targeting of ChemCam during or after or in the middle of long drives. The majority of ChemCam measurements are collected by allowing the science team to select specific targets in rover images.
However this requires the rover to stay in the same area while images are downlinked, analyzed for targets, and new commands uplinked. The only data that can be acquired without this communication cycle is via blind targeting, where
measurements are often of soil patches instead of more valuable targets such as rocks with specific properties. AEGIS is being applied to automatically analyze images onboard and select targets for ChemCam analysis. This approach allows the
rover to autonomously select and sequence targeted measurements in an opportunistic fashion at different points along the rover’s drive path. Rock targets can be prioritized for measurement based on various geologically relevant features,
including size, shape and albedo.
RMI image
acquisition
Algorithms quantify
key intuitive target
properties such as
brightness, size, and
eccentricity.
Advanced image
processing technique
enables reliable, rapid
identification of candidate
targets.
Target detection
Target feature
extraction
Target
prioritization
AEGIS Onboard Process for New Rock Measurements:
Target detection
Robust approach to
pointing selection
maximizes data of
target.
Target feature
extraction
Scientists
can prioritize
important
properties
for each run
Target
prioritization
Target pointing
determination
Coronation rock measured by ChemCam on sol 13. Target selected by ground. Contours
Flood fill +
Morphology ops
Top score
for large size
Robust approach to
pointing selection
maximizes data of target.
ChemCam
pointing
Top score
for large size
ChemCam data
acquisition
Raster performed on target.
(Raster pattern pre-selected by
scientists.) Multiple LIBS
spectrum acquired.
AEGIS Small Target Detection in RMI Images
Two methods area being investigated. One uses edge detection techniques that are also used to identify rocks (as shown on the left). The second uses texture channel analysis to identify areas of similar geologic texture.
ChemCam
pointing
ChemCam data
acquisition
LIBS spectrum
which indicates
physical and
chemical
composition of rock
Edge detection AEGIS Rock Property Analysis:
AEGIS Top Target Selection:
Albedo/Reflectance
• Scientists can prioritize different property values (and combinations of two values)
– Mean
– Variance
– Skew/Kurtosis
— e.g., prefer large, high albedo rocks
Size
– Inscribed circle
– Number of pixels
– Stereo true size
Sample Target Identification in MSL Navcam Image
Advanced image
processing technique
enables reliable, rapid
identification of
candidate targets.
Navcam
acquisition
Algorithms quantify
key intuitive target
properties such as
brightness, size,
and shape.
AEGIS Target Identification in Navcam Image
Scientists can
prioritize important
properties
for each run
Target pointing
determination
Light
Dark
Shape
• Priority specification is part of command sequencing
• Can be easily changed as rover enters different terrain areas
• AEGIS possible usages:
—
—
—
—
– Eccentricity
– Ellipse fit error
– Ruggedness
Rounded
Angular
Outcrop finder
Detection of meteorites
Support of cobble campaign
Soil‐only detector for MER/MSL soil‐survey
AEGIS analysis of original MSL Navcam image. Coronation rock identifed
as top target using standard profile.
Images from MER field trial
Near the
top of the
list of
“round”
rocks
Vein identification using edge detection techniques
Near the
bottom of
the list of
“round”
rocks
Vein identification using texture channel classification
References:
T. Estlin, B. Bornstein, D. Gaines, R. C. Anderson, D. Thompson, M. Burl, R. Castaño, and M. Judd, “AEGIS Automated Targeting for the MER
Opportunity Rover,” ACM Transactions on Intelligent Systems and Technology, 3(3), 2012.
D. R. Thompson, W. Abbey, A. Allwood, D. Bekker, B. Bornstein, N. A. Cabrol, R. Castaño, T. Estlin, T. Fuchs, K. L. Wagstaff, “Smart Cameras for
Remote Science Survey,” Proceedings of the International Symposium on Artificial Intelligence, Robotics and Automation in Space, Turin Italy, 2012.
R. Castaño, T. Estlin, R. C. Anderson, D. Gaines, A. Castano, B. Bornstein, C. Chouinard, M. Judd, “OASIS: Onboard Autonomous Science
Investigation System for Opportunistic Rover Science,” Journal of Field Robotics, Vol 24, No. 5, May 2007.
http://aegis.jpl.nasa.gov