BEV-Patch-PF: Particle Filtering with BEV-Aerial Feature Matching for Off-Road Geo-Localization
Dongmyeong Lee, Jesse Quattrociocchi, Christian Ellis, Rwik Rana, Amanda Adkins, Adam Uccello, Garrett Warnell, Joydeep Biswas
TL;DR
BEV-Patch-PF introduces a GPS-free sequential geo-localization framework that couples a particle filter with a learned BEV–aerial feature similarity for continuous pose estimation in off-road environments. By sampling aerial feature patches corresponding to each particle pose and weighting them with a BEV-ground feature match, the method yields a smooth, discriminative likelihood over SE(2) without discretizing heading. Training uses InfoNCE for discriminative alignment and a confidence loss to calibrate sampling reliability, enabling robust operation under canopy and shadows. Empirical results on multiple off-road datasets show significant ATE improvements over retrieval-based baselines and real-time performance on consumer-level GPUs, with a public CDS dataset and ROS 2 integration enhancing practical deployment.
Abstract
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 7.5x lower absolute trajectory error (ATE) on seen routes and 7.0x 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.