Belief-Space Residual Risk for Automated Driving under Localization Uncertainty

2026-05-12

AI Summary

This work addresses a critical limitation in existing residual risk assessment methods, which neglect the uncertainty inherent in ego-vehicle localization and thus fail to accurately capture safety risks in real-world driving scenarios. To bridge this gap, the study introduces a novel belief-space–based formulation of expected degraded risk that explicitly incorporates localization uncertainty into the residual risk evaluation framework. The ego-vehicle’s pose uncertainty is modeled using a Gaussian distribution, and its state covariance is jointly integrated with that of surrounding obstacles within a particle-based risk estimation scheme to compute collision probabilities in a unified manner. Experimental results demonstrate that the proposed approach significantly enhances the accuracy of safety risk quantification for autonomous driving systems, particularly in complex urban environments and under adverse weather conditions.

Abstract

Residual risk metrics have recently been introduced to assess the safety implications of automated driving systems. Existing approaches typically assume a deterministic ego pose and concentrate mainly on perception errors related to surrounding objects and latency effects. In practice, however, automated vehicles operate under considerable localization uncertainty, especially in complex urban settings and in adverse weather conditions. This work extends the spatial residual risk formulation to the belief space by explicitly modeling ego pose uncertainty as a Gaussian distribution. Residual risk is reformulated as the expected degradation-induced risk over the ego pose belief distribution. Within a particle-based risk estimation framework, localization uncertainty is incorporated into the computation of collision probabilities through covariance fusion of ego and object uncertainties.

Problem

• residual risk
• localization uncertainty
• automated driving
• belief space
• ego pose

Innovation

• belief-space
• residual risk
• localization uncertainty
• Gaussian distribution
• covariance fusion