ESA ExoMars: Pre-launch PanCam Geometric Modeling and Accuracy Assessment

D. Li, R Li, Alper Yılmaz2014 · DOI

Abstract

The ExoMars mission, a flagship of the ESA Aurora Programme, relies on its Panoramic Camera (PanCam) system for high-precision localization and topographic mapping. As the rover traverses kilometers of Martian terrain, understanding the impact of hardware specifications—such as stereo baseline, focal length, and pixel size—on mapping accuracy is vital. This paper presents a theoretical framework based on uncertainty propagation theory to quantify these errors pre-launch, enabling design optimization and informed mission planning.

Introduction: The Challenge of Martian Navigation

Navigating the Martian surface is a high-stakes task where precise self-localization is mandatory for safe path planning. Even minor errors in landmark perception can accumulate into significant positioning failures over long distances. To mitigate this, researchers have developed a quantitative error analysis for the ExoMars PanCam. By modeling how hardware parameters influence accuracy, engineers can optimize the system to achieve the highest possible precision for 3D mapping and rover localization.

Methodology: Modeling Error Propagation

The researchers developed a mathematical model grounded in photogrammetric principles and uncertainty propagation. The analysis is structured into three distinct stages:

  1. Single-Site Mapping: Modeling how feature identification errors translate into 3D coordinate uncertainties (range and azimuth).
  2. Two-Site Localization: Simplifying the "Bundle Adjustment" process into a rigid transformation model to calculate how measurement errors propagate to rover displacement when observing shared landmarks from consecutive positions.
  3. Multi-Site Traverse: Extending the model to long-distance scenarios to establish the relationship between traverse length and absolute/relative error.
Simulated distributions of 6, 9, 12, and 16 landmarks between two rover sites used to evaluate localization accuracy.

A significant component of this study is the inclusion of the High-Resolution Camera (HRC). While the Wide-Angle Cameras (WAC) provide standard stereo vision, the HRC’s long focal length allows it to resolve distant landmarks, acting as a high-precision "anchor" for the imagery network.

Key Insights: The Geometry of Accuracy

  • Landmark Distribution: Error ellipse analysis indicates that landmarks are most effective for localization when positioned roughly midway between two rover sites.
  • Quadratic vs. Linear Error: Range measurement error scales quadratically with distance from the camera. However, the relative localization error (error relative to distance traveled) increases linearly with the length of the traverse segment.
  • The HRC Advantage: The HRC can resolve distant features that appear blurred to the WAC. Utilizing these distant landmarks helps maintain heading and position accuracy over extended traverses.

Results and Performance Assessment

The theoretical model was validated using historical performance data from NASA’s Mars Exploration Rovers (MER), Spirit and Opportunity.

Relative localization error increases linearly with traverse length, with higher numbers of tie points significantly reducing the error rate.

Key findings include:

  • ExoMars WAC Performance: The ExoMars WAC system alone supports an average traverse segment of 36 meters while maintaining a 1% localization error, slightly outperforming the NASA MER Navcam (34 meters).
  • HRC Integration: Incorporating HRC data improves the average traverse segment length by approximately 10%, allowing for longer distances between navigation stops without compromising safety.
  • Long-Range Precision: For a 5-km journey, the model predicts a relative localization accuracy between 0.2% and 0.3%, assuming optimal landmark configurations are maintained.

Conclusion and Future Outlook

While this theoretical analysis provides a robust baseline for mission expectations, it assumes ideal conditions for landmark distribution. Real-world Martian terrain—ranging from featureless plains to boulder-strewn fields—will present additional challenges. Future work will focus on validating these models using PanCam prototypes in Earth-based "Mars yards" and integrating constraints related to onboard software processing speeds and terrain-induced obstacles.

Summary updated

Abstract

ExoMars is the flagship mission of the European Space Agency (ESA) Aurora Programme. The mobile scientific platform, or rover, will carry a drill and a suite of instruments dedicated to exobiology and geochemistry research. As the ExoMars rover is designed to travel kilometres over the Martian surface, high-precision rover localization and topographic mapping will be critical for traverse path planning and safe planetary surface operations. For such purposes, the ExoMars rover Panoramic Camera system (PanCam) will acquire images that are processed into an imagery network providing vision information for photogrammetric algorithms to localize the rover and generate 3-D mapping products. Since the design of the ExoMars PanCam will influence localization and mapping accuracy, quantitative error analysis of the PanCam design will improve scientists’ awareness of the achievable level of accuracy, and enable the PanCam design team to optimize its design to achieve the highest possible level of localization and mapping accuracy. Based on photogrammetric principles and uncertainty propagation theory, we have developed a method to theoretically analyze how mapping and localization accuracy would be affected by various factors, such as length of stereo hard-baseline, focal length, and pixel size, etc.
BibTeX
@inproceedings{li2014esa,
  author = {D. Li and R Li and Alper Yılmaz},
  title = {{ESA ExoMars: Pre-launch PanCam Geometric Modeling and Accuracy Assessment}},
  booktitle = {˜The},
  year = {2014},
  doi = {10.5194/isprsarchives-xl-3-177-2014},
}
Figure