Event-Driven Decision-Making for Autonomous Vehicles in Mixed-Traffic Roundabouts

Autonomous vehicle operation in roundabouts remains a challenging task due to continuous traffic flow, limited visibility, and the coexistence of human and automated drivers. Traditional optimizationor learning-based decision frameworks struggle to handle such conditions because they rely on perfect perception, continuous control, or full vehicle connectivity. This paper presents a discrete event system based decision-making framework for autonomous vehicles navigating single-lane roundabouts in mixed traffic. The proposed method models vehicle interactions as event-triggered state transitions, allowing transparent and modular decision logic between offensive and defensive driving behaviors. Simulations demonstrate that the proposed approach achieves collision-free operation and stable flow performance under partial observability, while maintaining interpretability and low computational cost. The results highlight that discrete event reasoning provides a scalable and explainable alternative to existing continuous and data-driven models for autonomous driving in complex urban environments.