Initial Results of Rover Localization and Topographic Mapping for the 2003 Mars Exploration Rover Mission

Abstract

This paper presents the initial results of lander and rover localization and topographic mapping of the MER 2003 mission (by Sol 225 for Spirit and Sol 206 for Opportunity). The Spirit rover has traversed a distance of 3.2 km (actual distance traveled instead of odometry) and Opportunity at 1.2 km. We localized the landers in the Gusev Crater and on the Meridiani Planum using two-way Doppler radio positioning technology and cartographic triangulations through landmarks visible in both orbital and ground images. Additional high-resolution orbital images were taken to verify the determined lander positions. Visual odometry and bundleadjustment technologies were applied to overcome wheel slippages, azimuthal angle drift and other navigation errors (as large as 21 percent). We generated timely topographic products including 68 orthophoto maps and 3D Digital Terrain Models, eight horizontal rover traverse maps, vertical traverse profiles up to Sol 214 for Spirit and Sol 62 for

Mars Exploration Rover Geologic traverse by the Spirit rover in the Plains of Gusev Crater, Mars

“…Summed traverse distances across the plains from the rim of Bonneville crater on sol 68 (the point of initial arrival at the rim) to arrival within a few meters of the geologic contact between the plains material and underlying materials of the Columbia Hills on sol 156 was 2.59 km from point to point (map distance), 3.39 km from wheel turns (odometery), and 2.71 km in summed straight line segments between localization stations Li et al, 2005), the higher value for odometry reflecting slippage and back-tracking. Maximum relief as measured through acquisition of rover tilt, wheel turns, and rover image correlation data (Li et al, 2005) was associated with the rims of Bonneville (Ah = 6.4 m) , and Missoula (Ah ~ 4 m) and Lahontan (Ah ~ 2.5 m) craters (Li et al, 2005). Numerous small impact craters probably account for most of the rolling relief on the plains, which created local relief on the order of 1 m over tens of meters horizontal distance in the vicinity of several small impact craters ("hollows") between Lahontan crater and the base of the Columbia Hills.…”

Position estimation by registration to planetary terrain

“…The MERs were first localized at their landers using radiometric tracking from the Mars Odyssey orbiter. Cartographic triangulations combining imagery from lander descent, ground panoramas, and orbital imagery served as a resource for human updates to optimally determine absolute lander position [11].…”

“…Visual odometry is used intermittently to detect slip and provide accurate position estimation. Bundle Adjustment is performed incrementally, sol-by-sol, on downlinked rover imagery to further improve position estimates [11].…”

“…It is assumed that rover yaw is known from a star tracker or sun sensor [11] and that local vertical is known from accelerometer measurements of the gravity vector.…”

Long‐range rover localization by matching LIDAR scans to orbital elevation maps

“…Wheel odometry is extremely vulnerable to sensor noise and mechanical disturbances such as wheel slippage and azimuthal angle drift. For example, during the MER mission, the Opportunity rover's wheel odometry once underestimated a 19m, three day traverse by over 8% (Li et al, 2005).…”