Uncertainty Estimation and Out-of-Distribution Detection for LiDAR Scene Semantic Segmentation

TL;DR

This paper introduces a novel approach for LiDAR scene segmentation that effectively detects out-of-distribution samples and quantifies both epistemic and aleatoric uncertainties using a single deterministic model and Gaussian Mixture Models.

Contribution

It presents a method that distinguishes in-distribution from OOD samples and estimates uncertainties without requiring additional OOD training data, outperforming existing techniques.

Findings

Superior OOD detection accuracy in real-world scenarios

Effective quantification of epistemic and aleatoric uncertainties

Detection of highly uncertain samples missed by deep ensembles

Abstract

Safe navigation in new environments requires autonomous vehicles and robots to accurately interpret their surroundings, relying on LiDAR scene segmentation, out-of-distribution (OOD) obstacle detection, and uncertainty computation. We propose a method to distinguish in-distribution (ID) from OOD samples and quantify both epistemic and aleatoric uncertainties using the feature space of a single deterministic model. After training a semantic segmentation network, a Gaussian Mixture Model (GMM) is fitted to its feature space. OOD samples are detected by checking if their squared Mahalanobis distances to each Gaussian component conform to a chi-squared distribution, eliminating the need for an additional OOD training set. Given that the estimated mean and covariance matrix of a multivariate Gaussian distribution follow Gaussian and Inverse-Wishart distributions, multiple GMMs are generated by sampling from these distributions to assess epistemic uncertainty through classification variability. Aleatoric uncertainty is derived from the entropy of responsibility values within Gaussian components. Comparing our method with deep ensembles and logit-sampling for uncertainty computation demonstrates its superior performance in real-world applications for quantifying epistemic and aleatoric uncertainty, as well as detecting OOD samples. While deep ensembles miss some highly uncertain samples, our method successfully detects them and assigns high epistemic uncertainty.