Robust Semantic Constellation Matching and SLAM for Planetary Rover Global Localization

Emma Seabrook, Damien Vivet, Alberto Ricci

Keywords: Planetary localization, SLAM, global relocalization, map matching, RANSAC, semantic detection, EKF

Abstract: Planetary rover localization remains a critical challenge due to the absence of Global Navigation Satellite Systems and communication latency that prevents real-time human intervention. While traditional SLAM-based methods enable local navigation, they suffer from cumulative drift and rely on loop closure, a constraint that is often impractical given the non-repetitive trajectories typical of planetary exploration. This work presents a SLAM-based localization framework that leverages a global re-localization module based on geometric constellation matching. Utilizing a YOLOv12 model, semantic rock detections are extracted from both local rover imagery and overhead orbital maps. These landmarks are matched using a triangle invariant descriptor and geometric hashing, allowing a RANSAC-based alignment process to estimate the rover's global pose. Integrated within an Extended Kalman Filter (EKF), this system demonstrates reliable global pose recovery and bounded drift correction in simulated Mars environments.

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Submission Number: 11