Probabilistic Topological Map Inference with Belief Propagation
Houzhe Wang, Jingqi Jiang, Shida Xu, Eric Yeatman, Sen Wang
AI summary
Problem
Metric SLAM back-ends are ill-suited for confined, branched environments where topological connectivity matters more than geometric accuracy, and existing probabilistic topological mapping methods lack scalability and robustness to perceptual aliasing.
Approach
The method models topological inference as a factor graph and uses belief propagation to compute a posterior distribution over plausible node correspondences and connectivity, leveraging pairwise similarity and sequential transition statistics.
Key results
- Factor-graph formulation for topology-space inference
- Specialized pairwise and sequential connectivity factors to handle aliasing
- Superior topology recovery over PTM baseline across noise levels and scales
- Perfect matching accuracy and edge F1 under low-to-medium noise
Why it matters
Enables reliable navigation and monitoring in complex, confined robotic environments where metric accuracy is unreliable but topological structure is critical.
Abstract
Metric Simultaneous Localization and Mapping (SLAM) prioritizes geometric accuracy of estimated robot poses and maps. However, in many real-world robot appli- cations, such as inspection robots operating inside pipelines or other confined network environments, metric accuracy is less critical than correctly capturing the underlying topological connectivity. In this paper, we investigate back-end optimization for topological mapping/SLAM, and propose a probabilistic topological map inference algorithm. Given noisy front-end measurements, our approach explicitly models the topological map inference problem within a factor graph framework. It performs inference using belief propagation, which yields a posterior distribution over multiple plausible topological maps rather than a single estimate. We evaluate our method on topologies derived from an open-source pipeline network dataset, spanning various topology sizes and degrees of per- ceptual aliasing. Extensive experiments demonstrate that our algorithm infers high-quality topological maps across varying conditions.