Professor Dezong Zhao

The safe and reliable deployment of autonomous vehicles will play a key role in shaping future mobility. However, ensuring that autonomous vehicles are truly safe in all situations remains a major challenge. Real-world driving is full of rare, unpredictable, and complex events – from sudden pedestrian crossings to unusual road layouts and extreme weather. Testing autonomous vehicles on public roads alone cannot cover all these possibilities, while traditional simulation tools often rely on limited or unrealistic scenarios. Without systematic and rigorous testing, autonomous vehicles may behave unpredictably in critical situations, raising serious safety concerns and public trust issues. 

This project aims to transform how autonomous vehicles are tested by developing a new framework for scenario generation and autonomous testing. Using cutting-edge techniques in generative AI, the project will create realistic, diverse, and safety-critical scenarios – especially those rarely seen in real-world driving. These scenarios will then be used in automated testing platforms that can rapidly and cost-effectively evaluate autonomous vehicle behaviour under a wide range of conditions. This work will bridge the gap between simulation and reality, accelerate safe deployment, and help ensure autonomous vehicles are ready for the complexities of real-world roads.