Enhancing Safety Standards in Automated Systems Using Dynamic Bayesian Networks
Kranthi Kumar Talluri, Anders L. Madsen, Galia Weidl
TL;DR
The paper addresses unsafe cut-in maneuvers in high-speed traffic by introducing a Dynamic Bayesian Network (DBN) that fuses lateral evidence with proactive safety checks. By decomposing safety into three probabilistic hypotheses—$LE$, $Safe\_LAT$, and $Safe$—and operating at 10 Hz, the approach enables earlier lane-change recognition and safer braking in critical scenarios. Results show that the DBN reduces crashes in high-speed conditions (≈9.22% vs ≈25% for baselines) while maintaining competitive low-speed performance, demonstrating robust safety validation potential and explainability. Overall, the work provides a scalable probabilistic framework that enhances automated driving safety validation and offers actionable insights for regulation and deployment.
Abstract
Cut-in maneuvers in high-speed traffic pose critical challenges that can lead to abrupt braking and collisions, necessitating safe and efficient lane change strategies. We propose a Dynamic Bayesian Network (DBN) framework to integrate lateral evidence with safety assessment models, thereby predicting lane changes and ensuring safe cut-in maneuvers effectively. Our proposed framework comprises three key probabilistic hypotheses (lateral evidence, lateral safety, and longitudinal safety) that facilitate the decision-making process through dynamic data processing and assessments of vehicle positions, lateral velocities, relative distance, and Time-to-Collision (TTC) computations. The DBN model's performance compared with other conventional approaches demonstrates superior performance in crash reduction, especially in critical high-speed scenarios, while maintaining a competitive performance in low-speed scenarios. This paves the way for robust, scalable, and efficient safety validation in automated driving systems.