SNGR: Selective Non-Gaussian Refinement for Ambiguous SLAM Factor Graphs
For SLAM practitioners, SNGR addresses the problem of ambiguous factor graphs where Gaussian approximations are inaccurate, offering a selective refinement method that balances accuracy and computational efficiency.
SNGR augments iSAM2 with selective nested sampling on windows where Gaussian approximations fail, achieving high-precision failure detection and consistent local likelihood improvements in range-only SLAM with wrong data association, while reducing computational cost relative to exhaustive non-Gaussian inference.
We present Selective Non-Gaussian Refinement (SNGR), a SLAM framework that augments iSAM2 with targeted nested sampling on windows where Gaussian approximations are likely to fail. We detect such regions using the condition number of joint marginal covariances and selectively refine them using the full nonlinear factor graph likelihood, with a gating mechanism to avoid degradation in multimodal cases. Experiments on range-only SLAM with wrong data association show that SNGR achieves high-precision failure detection and consistent local likelihood improvements while reducing computational cost relative to exhaustive non-Gaussian inference. These results highlight both the promise and the limitations of selective refinement for approximate SLAM posteriors.