In Mars landed missions, accurate navigation of the rover, or “roving geologist”, are critical to operations such as rover safety assessment, coordinated multidisciplinary high-precision scientific experiments, repeated sampling that requires the rover to return to the same location, and even sample return in future missions. This is particularly true during long-range traversal at landing sites.

As a follow-up to the highly successful operation of the rover Sojourner during the 1997 Mars Pathfinder (MPF) mission, the Mars Exploration Rover (MER) mission has successfully launched two larger and better-equipped rovers. Named Spirit and Opportunity through a student essay contest, the rovers will land on two different sites, Gusev Crater and Meridiani Planum, that lie half a globe apart. They will document the geology of the landing sites, gathering compositional, mineralogical, and textural information about selected Martian soils and rocks in order to examine the possibility that water has existed on Mars.

The expansion of the extent of rover traverse operations for the 2003 MER mission presents a great challenge for detailed landing site mapping using surface images and long-range rover localization. In the previous MPF mission, Sojourner achieved a rover localization accuracy of 12% within an operation area of about 10m ×10m. For the MER mission, the two rovers will traverse an extended distance of from 600m up to 1,000m.

Since 1998, the Mapping and GIS Laboratory at the OSU Department of Civil and Environmental Engineering and Geodesic Science, led by Professor Ron Li, has worked in conjunction with JPL to conduct research on a new technology for long-range rover localization. This new technology is based on a rover image network and bundle adjustment. In this technology, the rover images are linked to form an image network through the use of tie points, which are corresponding feature points and landmarks that are identified in the stereo rover images. Bundle adjustment of the image network provides high precision camera positions and attitudes (0.1%) of the images as well as 3D ground positions of the tie points. Orbital satellite images can also be linked to the network to suppress error accumulation in the image traverse. Bundle adjustment results are used to update rover positions and attitudes sol-by-sol and to support landing site map production.

As a Participating Scientist of the MER Mission, Dr. Li and his team (Co-I Dr. Kaichang Di with PhD students Fengliang Xu and Jue Wang) are working on a project called “Surface Image-based High-precision Near Real-time Landing Site Mapping and Long-range Rover Localization”. Charles Serafy, a student in Mapping and GIS, will also be involved in the experiments for his MS thesis.

Dr. Li and his team have developed a “MarsMapper” bundle adjustment and mapping software program. MarsMapper has been tested using field test data acquired in 1999 and 2000 at a test site at Silver Lake, CA, along with Mars Pathfinder lander and rover imagery plus FIDO (Field Integrated Design and Operations) rover data acquired by the Athena Science team in August 2002. The achieved accuracy is 1% - 0.1%, which is a significant improvement over the nominal accuracy of 10%. This software program provides efficient and highly accurate localization for the Mars rovers, enabling them to traverse the Martian surface and gather scientific samples during the anticipated mission operational period of January to April 2004. In addition, the team will produce very accurate maps (such as digital terrain models, orthophoto maps, and rover traverse maps of the landing site) in near real-time to support science and engineering operations.

Starting Sol 1 after the landing of the two rovers, Dr. Li will be working at JPL in Pasadena as his team works here at OSU. They will be processing the images taken each sol (Martian day) and downlinked each “evening”. Each Martian night, new rover positions and attitudes will be calculated to update the rover location information. The data will also be used to produce topographic maps in near real-time (sol by sol). These maps will be provided to MER scientists to help them in their scientific investigation tasks. In addition, the team will also establish a web-based GIS for rover tracking and image and map visualizing.