Research Analyzer
← Back ICRA 2026

Geometry-Aided Vision-Based Localization of Future Mars Helicopters in Challenging Illumination Conditions

Dario Pisanti, Robert Hewitt, Roland Brockers, Georgios Georgakis

PDF

AI summary

Key figure (auto-extracted from paper)
Geo-LoFTR significantly improves map-based localization accuracy for Mars helicopters under extreme lighting and scale variations by fusing visual features with terrain depth cues.
Mars helicopter map-based localization geometry-aided matching deep learning illumination robustness MARTIAN simulation

Problem

Future Mars rotorcraft require drift-free global localization to navigate long distances, but traditional map-based systems fail when lighting and scale differ drastically between onboard cameras and orbital reference maps.

Approach

The authors introduce Geo-LoFTR, a transformer-based image matcher that combines visual data with depth information from digital terrain models, and train it using MARTIAN, a custom simulation tool that generates realistic Martian aerial imagery from orbital maps.

Key results

  • Up to 31.8% localization accuracy improvement at 1m under challenging illumination
  • Custom MARTIAN simulation pipeline for large-scale Martian dataset generation
  • Robust performance across varying sun angles and altitude scales
  • Validated on synthetic data and real Mars2020 descent imagery

Why it matters

Enables reliable long-range aerial navigation for future Mars missions by overcoming the lighting and scale limitations that currently restrict helicopter operational windows.

Abstract

Planetary exploration using aerial assets has the potential for unprecedented scientific discoveries on Mars. While NASA’s Mars helicopter Ingenuity proved flight in Martian atmosphere is possible, future Mars rotorcraft will require advanced navigation capabilities for long-range flights. One such critical capability is Map-based Localization (MbL) which registers an onboard image to a reference map dur- ing flight to mitigate cumulative drift from visual odometry. However, significant illumination differences between rotorcraft observations and a reference map prove challenging for tradi- tional MbL systems, restricting the operational window of the vehicle. In this work, we investigate a new MbL system and propose Geo-LoFTR, a geometry-aided deep learning model for image registration that is more robust under large illumination differences than prior models. The system is supported by a custom simulation framework that uses real orbital maps to produce large amounts of realistic images of the Martian ter- rain. Comprehensive evaluations show that our proposed system outperforms prior MbL efforts in terms of localization accuracy under significant lighting and scale variations. Furthermore, we demonstrate the validity of our approach across a simulated Martian day and on real Mars imagery. Code and datasets are available at: https://dpisanti.github.io/geo-loftr/.

Index terms

Space Robotics and Automation Localization Deep Learning for Visual Perception

Related papers