TY - JOUR
T1 - MER Spirit rover localization
T2 - Comparison of ground image- and orbital image-based methods and science applications
AU - Li, Rongxing
AU - He, Shaojun
AU - Chen, Yunhang
AU - Tang, Min
AU - Tang, Pingbo
AU - Di, Kaichang
AU - Matthies, Larry
AU - Arvidson, Raymond E.
AU - Squyres, Steven W.
AU - Crumpler, Larry S.
AU - Parker, Tim
AU - Sims, Michael
PY - 2011
Y1 - 2011
N2 - During 6 years of continuous operations on the Martian surface, the Mars Exploration Rover (MER) Spirit has covered a traverse of approximately 7 km from the landing point to its current position at "Troy" near Home Plate. Localization of Spirit (and Opportunity) has been performed using two different methods: one that employs an incremental bundle adjustment (IBA) using rover imagery, and one that compares image features common to both a rover orthoimage and an orbital orthoimage. The IBA method continuously yields the desired 3-D rover positions at a very high level of accuracy and provides a simultaneous solution for high-quality topographic mapping of neighborhoods surrounding the rover. On the other hand, high-resolution orbital imagery can verify rover positions wherever the rover track is visible. Rapid rover localization on the orbital orthoimage is often achieved by comparing a rover orthoimage to the orbital orthoimage. In this paper, we present research results from a systematic comparison of these two localization methods over the entire length of the Spirit traverse. Two orbital orthoimages were generated from High Resolution Imaging Science Experiment (HiRISE) imagery. Integration of Mars Orbiter Laser Altimeter (MOLA) data into the HiRISE digital elevation model (DEM) and orthoimage generation was performed and proved to be effective in reducing large inconsistencies between MOLA and HiRISE data. This study found an overall difference of 1.5 percent of the traversed distance between the two sets of traverse positions derived using the two different localization methods. After a geometric transformation from one traverse to the other, the remaining inconsistency then represents the local differences between them and can be reduced to a level of less than 0.15 percent. Discussions of error sources and the strength and weakness of the methods are given. Scientific applications of the localization data are also briefly introduced.
AB - During 6 years of continuous operations on the Martian surface, the Mars Exploration Rover (MER) Spirit has covered a traverse of approximately 7 km from the landing point to its current position at "Troy" near Home Plate. Localization of Spirit (and Opportunity) has been performed using two different methods: one that employs an incremental bundle adjustment (IBA) using rover imagery, and one that compares image features common to both a rover orthoimage and an orbital orthoimage. The IBA method continuously yields the desired 3-D rover positions at a very high level of accuracy and provides a simultaneous solution for high-quality topographic mapping of neighborhoods surrounding the rover. On the other hand, high-resolution orbital imagery can verify rover positions wherever the rover track is visible. Rapid rover localization on the orbital orthoimage is often achieved by comparing a rover orthoimage to the orbital orthoimage. In this paper, we present research results from a systematic comparison of these two localization methods over the entire length of the Spirit traverse. Two orbital orthoimages were generated from High Resolution Imaging Science Experiment (HiRISE) imagery. Integration of Mars Orbiter Laser Altimeter (MOLA) data into the HiRISE digital elevation model (DEM) and orthoimage generation was performed and proved to be effective in reducing large inconsistencies between MOLA and HiRISE data. This study found an overall difference of 1.5 percent of the traversed distance between the two sets of traverse positions derived using the two different localization methods. After a geometric transformation from one traverse to the other, the remaining inconsistency then represents the local differences between them and can be reduced to a level of less than 0.15 percent. Discussions of error sources and the strength and weakness of the methods are given. Scientific applications of the localization data are also briefly introduced.
UR - http://www.scopus.com/inward/record.url?scp=79951940204&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=79951940204&partnerID=8YFLogxK
U2 - 10.1029/2010JE003773
DO - 10.1029/2010JE003773
M3 - Article
AN - SCOPUS:79951940204
SN - 0148-0227
VL - 116
JO - Journal of Geophysical Research: Planets
JF - Journal of Geophysical Research: Planets
IS - 2
M1 - E00F16
ER -