Visual Relocalization from Sparse Views in Aliased and Low-Texture Environments via Novel View Synthesis

Maria Peribañez, Javier Civera, Rudolph Triebel, Riccardo Giubilato

Visual localization becomes extremely challenging in planetary-like terrains characterized by low texture, perceptual aliasing, harsh illumination, and sparse, weakly overlapping viewpoints induced by forward rover motion and unconstrained driving directions. Under these conditions, state-of-the-art image-to-image and image-to-map matching pipelines suffer significant performance degradation. In this work, we propose a visual relocalization method that departs from classical correspondence-based pipelines by directly estimating camera poses against a differentiable map representation built with 3D Gaussian Splatting (3DGS). Our key contribution is a geometry-aware training strategy that combines photometric and geometric losses, where the geometric supervision is provided for the first time by combining multi-view stereo (MVS) and LiDAR depths. We show that this joint optimization produces a 3DGS model that better fits the underlying scene geometry, leading to improved photometric and geometric consistency and more robust, accurate single-image 6-DoF pose estimation. Extensive experiments on data acquired in planetary-analog environments validate the effectiveness of our approach, showing substantial gains in relocalization accuracy under challenging conditions. Code is available at https://github.com/DLR-RM/multimodal-gsplat-relocalization.