PROSAFE-TWIN

PROSAFE‑TWIN delivers a safety‑first digital twin fusing physics-based traffic dynamics with AI to prevent safety‑critical events before crashes occur. While current management is reactive—acting only after incidents emerge—
PROSAFE-TWIN identifies high-risk segments in advance. It transforms the Safe System Approach into an operational
reality, providing road authorities with the predictive intelligence to intervene before escalation, directly supporting the
EU’s "Vision Zero" goals.
The technical core is a Hybrid Digital Twin bridging the gap between raw data and human decision-making. By
combining the "laws of physics" with AI, the system fuses streams from sensors, connected vehicles, and weather into a
real-time risk map. This detects hazards like hidden queues or dangerous interactions with vulnerable road users (VRUs).
Crucially, the project integrates Social Sciences and Humanities (SSH) to translate insights into usable workflows, using
human-factors engineering to reduce cognitive overload and ensure effective human-in-the-loop oversight.
The project delivers high-impact outputs focused on trust and scalability. PROSAFE‑TWIN addresses the AI "black box"
by providing uncertainty-aware predictions with explicit confidence bounds, eliminating alert fatigue. A fairness-bydesign framework ensures
safety benefits are equitably distributed across socioeconomic contexts. To support resource constrained networks, the SAVI framework optimizes data efficiency, making proactive safety viable without dense, expensive hardware. Validation occurs across four diverse European testbeds: rural hubs, urban streets, high-speed corridors, and datasparse regional roads.
By encoding risk intelligence into ITS standards and providing an EU AI Act conformity pack, PROSAFE-TWIN establishes a replicable, privacy-preserving blueprint.