Skip to content
Open access

Ground-Vehicle-Assisted Monocular vSLAM Correction for UAV Pose Estimation

2026 · IEEE Access · Vol 14, pp. 165709-165732 · 0 citations · 58 references

TL;DR

The results demonstrate that an ordinary globally localized ground vehicle can provide useful metric drift correction for monocular UAV vSLAM without artificial markers, UWB ranging, or additional onboard UAV localization sensors.

Abstract

Monocular visual SLAM is attractive for lightweight UAV localization, but its unknown metric scale and accumulated drift limit long outdoor missions. This paper presents a cooperative monocular vSLAM method in which a globally localized ground vehicle of known geometry acts as a moving metric anchor. A fine-tuned YOLO-OBB detector locates vehicle candidates in downward-looking UAV images, and a ReID gate selects the instrumented ground vehicle. The selected box corners, known vehicle geometry, and synchronized global vehicle pose are used with PnP to estimate UAV camera-position priors. Each prior is assigned an uncertainty for subsequent optimization. A stability-tested Sim(3) transformation initializes the monocular reconstruction in the global metric frame. A translation-only keyframe graph then combines visual relative displacements with intermittent absolute and relative position-prior constraints while retaining the Sim(3)-initialized keyframe orientations. The accepted correction is propagated coherently to the map, after which joint soft-pose bundle adjustment refines keyframe poses and map points under visual reprojection residuals and position priors. The resulting corrected ORB-SLAM3 map can be stored, shared, and reused for subsequent localization. The method is evaluated on six outdoor UAV–UGV sequences from the AeroScale-Drive dataset. Stable metric initialization is obtained for five sequences, and all five benefit from optimization. Across three complete runs, the mean translation-registered three-dimensional trajectory RMS error decreases from 28.532 m to 6.906 m, corresponding to a 75.8% reduction, while the corrected maps retain low internal image-reprojection residuals and remain reloadable. The vehicle-derived pose-prior front end achieves a mean RMS error of 4.941 m. These results demonstrate that an ordinary globally localized ground vehicle can provide useful metric drift correction for monocular UAV vSLAM without artificial markers, UWB ranging, or additional onboard UAV localization sensors. AeroScale-Drive is available through IEEE DataPort at https://dx.doi.org/10.21227/ng66-jy89

Read PDF

Similar papers

Sep 2026

HAT: Hypothesis-Anchored Tracking for Video Monocular Spacecraft Pose Estimation

Monocular 6-DoF pose estimation of non-cooperative targets is important for on-orbit servicing and debris removal. A single-image estimator can confuse near-symmetric spacecraft orientations, and tracking can preserve an incorrect pose. We present Hypothesis-Anchored Tracking (HAT), a causal framework that uses inter-f...

André Lopo, Atabak Dehban, Rodrigo M. M. Ventura · 1 citation
Open access Oct 2026

A Pose-Only Point–Line Geometric Representation Framework for Visual–Inertial Localization of UAVs and UGVs in Complex Environments

A unified pose-only measurement model is developed for MSCKF point and line features, which improves localization accuracy while maintaining processing efficiency comparable to that of the corresponding point-line system without the pose-only formulation.

Ai-Ping Wang, Zhao-Long Yang, Shu-Wen Chen et al. · 0 citations
Open access 2026

Dynamic-Aware Point-Line Visual SLAM with Loop-Consistent Semantic Mapping

PLCS-SLAM is presented, a visual localization and mapping system for dynamic and low-texture environments that combines depth-aware semantic gating with probabilistic geometric verification to suppress dynamic outliers while preserving valid static features.

Bo Jiang, Meng-Chen Xu, Cheng-Long Li et al. · 0 citations
Open access Oct 2026

MSR-SLAM: A Multi-Stage Map-Refinement Framework for Reliable LiDAR–Inertial SLAM in GPS-Denied Environments

MSR-SLAM is presented, a multi-stage map-refinement framework built on an existing LiDAR–inertial odometry front end that rejects constraints whose candidate-induced translational or rotational correction exceeds preset limits before they can affect the active pose graph.

Munef El Muhammed, Anas Mahyoub Naji Saeed Alqadhi, M. A. M. S. Bajhaw et al. · 0 citations
Book Open access Nov 2026

LaMP: Last-Meter Manifold Pose Estimation for Image-Goal Terminal UAV Navigation

Image-goal terminal UAV navigation guides a UAV from its current viewpoint to a goal viewpoint specified by a reference image. Local feature matching is brittle under the large viewpoint change between the two views, and monocular geometry alone struggles to supply the target-centric metric range the task demands. We p...

Hua-Feng Li, Si-Qi Li, Xuan-Bin Wang et al. · 1 citation
Sep 2026

Robust trajectory triangulation for UAV visual localization via iteratively reweighted least squares.

Accurate localization of ground targets using monocular vision from a fixed-wing unmanned aerial vehicle (UAV) in tracking mode is a challenging problem in computer vision and photogrammetry. Classical triangulation methods often suffer from rank deficiency due to degenerate observation geometry. The trajectory triangu...

Yin-Bo Xu, Zhi-Jian Xiao, Jia-Run Yan et al. · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.