Nested Sampling for Non-Gaussian Inference in SLAM Factor Graphs

We present nested sampling for factor graphs (NSFG), a novel approach to approximate inference posterior distributions expressed over factor-graphs. Performing such is key step in simultaneous localization and mapping (SLAM). Although the Gaussian approximation often works well, other more challenging SLAM situations, distribution non-Gaussian cannot be explicitly represented with standard distributions. Our technique applies settings where substantially (e.g., multi-modal) thus needs expressive representation. NSFG exploits methods directly sample represent without parametric density models. While are known their powerful capability multi-modal distributions, application of not straightforward. leverages structure construct informative prior which efficiently sampled provide notable computational benefits methods. simulated experiments demonstrate that robust computes solutions an order magnitude faster than state-of-the-art techniques. Similarly, we compare approaches notably describing posteriors.