Project

Full title: Safety-Enhanced and Physics-Informed Foundation Models of Vehicle Behavior

This project develops physics-informed AI foundation models to enhance safety validation in automated vehicles. By embedding physical constraints into AI learning, the models improve predictive reliability, interpretability, and scalability across diverse driving conditions. The approach integrates automated test case generation, enabling comprehensive validation of safety-critical scenarios and reducing reliance on costly physical trials. Outcomes include improved simulation accuracy, proactive safety interventions, and deployment within real-world SIL/HIL frameworks. The project directly supports FFI’s goals of safe, sustainable, and intelligent road transport while advancing Europe’s leadership in AI-driven vehicle automation.

Traffic safety benefit: Enhances automated vehicle safety through physics-informed AI for reliable prediction, validation, and proactive risk reduction.

Key words: Foundation models, physics-informed neural networks, vehicle behavior, automation, time-series analysis, generative models, test case generation