Road Similarity-Based BEV-Satellite Image Matching for UGV Localization
🤖 AI Summary
For autonomous ground vehicles (UGVs) operating in GNSS-denied off-road environments, this paper proposes a vision-aided global localization method based on cross-domain matching between LiDAR-derived bird’s-eye view (BEV) maps and satellite imagery. The core method addresses the significant viewpoint disparity between ground and aerial perspectives by constructing a road-structure similarity embedding space: robust BEV representations are generated via tightly coupled LiDAR-inertial odometry fused with multimodal perception. Localization is then achieved through normalized cross-correlation (NCC)-based similarity scoring integrated within a Bayesian particle filter framework, enabling long-range, GNSS-free global positioning. Experimental validation over a 10-km off-road trajectory demonstrates no localization drift, with mean lateral error of 0.89 m and planar Euclidean error of 3.41 m. The approach maintains high accuracy and robustness under challenging nighttime conditions.
Application Category
📝 Abstract
To address the challenge of autonomous UGV localization in GNSS-denied off-road environments,this study proposes a matching-based localization method that leverages BEV perception image and satellite map within a road similarity space to achieve high-precision positioning.We first implement a robust LiDAR-inertial odometry system, followed by the fusion of LiDAR and image data to generate a local BEV perception image of the UGV. This approach mitigates the significant viewpoint discrepancy between ground-view images and satellite map. The BEV image and satellite map are then projected into the road similarity space, where normalized cross correlation (NCC) is computed to assess the matching score.Finally, a particle filter is employed to estimate the probability distribution of the vehicle's pose.By comparing with GNSS ground truth, our localization system demonstrated stability without divergence over a long-distance test of 10 km, achieving an average lateral error of only 0.89 meters and an average planar Euclidean error of 3.41 meters. Furthermore, it maintained accurate and stable global localization even under nighttime conditions, further validating its robustness and adaptability.
Problem
Research questions and friction points this paper is trying to address.
UGV localization in GNSS-denied off-road environments
Matching BEV perception images with satellite maps
Reducing viewpoint discrepancy for accurate positioning
Innovation
Methods, ideas, or system contributions that make the work stand out.
BEV-satellite image matching in road similarity space
LiDAR-inertial odometry with LiDAR-image fusion
Particle filter for pose probability distribution estimation
🔎 Similar Papers
No similar papers found.
