BEV-Patch-PF: Particle Filtering with BEV-Aerial Feature Matching for Off-Road Geo-Localization
Not on the map. Why?
Abstract (arXiv)
We propose BEV-Patch-PF, a GPS-free sequential geo-localization system that integrates a particle filter with learned bird's-eye-view (BEV) and aerial feature maps. From onboard RGB and depth images, we construct a BEV feature map. For each 3-DoF particle pose hypothesis, we crop the corresponding patch from an aerial feature map computed from a local aerial image queried around the approximate location. BEV-Patch-PF computes a per-particle log-likelihood by matching the BEV feature to the aerial patch feature. On two real-world off-road datasets, our method achieves 9.7x lower absolute trajectory error (ATE) on seen routes and 6.6x lower ATE on unseen routes than a retrieval-based baseline, while maintaining accuracy under dense canopy and shadow. The system runs in real time at 10 Hz on an NVIDIA Tesla T4, enabling practical robot deployment.
Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 33 of its 51 references that are in the map. First posted on arXiv 17 Dec 2025.
Builds on
33 of 51 references.
References · 33
- 2015Deep Residual Learning for Image Recognitioncited by 52,096
- 2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,974
- 2016Feature Pyramid Networks for Object Detectioncited by 5,036
- 2025DINOv3cited by 1,414
- 2018Unified Perceptual Parsing for Scene Understandingcited by 1,271
- 2021Swin Transformer V2: Scaling Up Capacity and Resolutioncited by 1,228
- 2020Lift, Splat, Shoot: Encoding Images From Arbitrary Camera Rigs by Implicitly Unprojecting to 3Dcited by 1,044
- 2021KITTI-360: A Novel Dataset and Benchmarks for Urban Scene Understanding in 2D and 3Dcited by 575
- 2020VIGOR: Cross-View Image Geo-localization beyond One-to-one Retrievalcited by 189
- 2025FoundationStereo: Zero-Shot Stereo Matchingcited by 173
- 2022TransGeo: Transformer Is All You Need for Cross-view Image Geo-localizationcited by 168
- 2023AnyLoc: Towards Universal Visual Place Recognitioncited by 163
21 more
- 2021Coming Down to Earth: Satellite-to-Street View Synthesis for Geo-Localizationcited by 113
- 2020TartanVO: A Generalizable Learning-based VOcited by 108
- 2022Beyond Cross-view Image Retrieval: Highly Accurate Vehicle Localization Using Satellite Imagecited by 88
- 2023OrienterNet: Visual Localization in 2D Public Maps with Neural Matchingcited by 84
- 2017Cross-View Image Matching for Geo-localization in Urban Environmentscited by 69
- 2024CricaVPR: Cross-image Correlation-aware Representation Learning for Visual Place Recognitioncited by 61
- 2022Uncertainty-aware Vision-based Metric Cross-view Geolocalizationcited by 54
- 2022Visual Cross-View Metric Localization with Dense Uncertainty Estimatescited by 48
- 2022SliceMatch: Geometry-guided Aggregation for Cross-View Pose Estimationcited by 46
- 2023Boosting 3-DoF Ground-to-Satellite Camera Localization Accuracy via Geometry-Guided Cross-View Transformercited by 45
- 2020Ford Multi-AV Seasonal Datasetcited by 39
- 2023Convolutional Cross-View Pose Estimationcited by 36
- 2015Semantic Cross-View Matchingcited by 31
- 2019Image-Based Geo-Localization Using Satellite Imagerycited by 26
- 2021Super Odometry: IMU-centric LiDAR-Visual-Inertial Estimator for Challenging Environmentscited by 26
- 2023Learning Dense Flow Field for Highly-accurate Cross-view Camera Localizationcited by 24
- 2024TartanDrive 2.0: More Modalities and Better Infrastructure to Further Self-Supervised Learning Research in Off-Road Driving Taskscited by 23
- 2019You Are Here: Geolocation by Embedding Maps and Imagescited by 19
- 2024BEVRender: Vision-based Cross-view Vehicle Registration in Off-road GNSS-denied Environmentcited by 5
- 2024BEVLoc: Cross-View Localization and Matching via Birds-Eye-View Synthesiscited by 4
- 2025Learning to be Smooth: An End-to-End Differentiable Particle Smoothercited by 4
Built on by
Not cited yet.