Occlusion-Robust Relative Pose Estimation for Multi-Robot Systems Via Geometric-Aware Diffusion Matching
Suyoung Kang, Rishav Dutta, Peng Gao, Maggie Wigness, John G. Rogers III, Donghyun Kim, Hao Zhang
AI summary
Problem
Classical and learning-based relative pose estimation methods degrade under occlusions, texture scarcity, and partial view overlap, lacking explicit geometric consistency needed for reliable multi-robot collaboration in dynamic, GPS-denied environments.
Approach
GADM uses a diffusion model to progressively refine feature correspondences between robots, guided by a graph neural network and explicit geometric consistency losses to suppress spurious matches and enforce epipolar constraints.
Key results
- Novel GADM framework integrating diffusion processes with geometric constraints for correspondence refinement
- Robust 6-DoF relative pose estimation under partial view overlap and severe occlusions
- Real-time, decentralized pose estimation on physical robot teams without GPS or global maps
- Superior performance over classical and learning-based baselines in simulation and real-world experiments
Why it matters
Provides a reliable, map-free visual localization solution critical for coordinated multi-robot operations in GPS-denied, cluttered, or dynamic environments like search-and-rescue or urban exploration.
Abstract
Relative pose estimation is crucial for coordinated multi-robot navigation. However, robots in close proximity often face intra-team occlusions, where teammates partially block each other’s field of view, while dynamic environments further introduce environmental occlusions. Classical relative pose esti- mation methods degrade under occlusion and texture scarcity, whereas learning-based methods often lack explicit geometric consistency, which limits their accuracy during real deployments. To address multi-robot relative pose estimation in complex 3D environments, we introduce Geometric-Aware Diffusion Matching (GADM), which enables a team of robots to estimate relative 6-DoF poses using only RGB-D sensors, even under occlusions. GADM uses a diffusion model to progressively exploit global and higher-order structural constraints encoded by a graph network, guiding smoother optimization and faster convergence to robust correspondence distributions under noise and occlusions. By integrating geometric consistency, GADM explicitly addresses occlusions by producing geometrically consistent matches suit- able for real-time deployment on physical robots. The resulting correspondences are then used with geometry-based solvers to estimate 6-DoF relative poses, providing robustness even under partial view overlap and limited keypoint visibility. We conducted experiments using both robotics simulations and physical robot teams, and our results show that GADM achieves robust 6-DoF pose estimation performance in occluded scenarios. More details are provided on the project website: https:// gadm2026.github.io.