Prof. Saber Fallah is Professor of Safe AI and Autonomy and Director of the Cognitive Autonomous Vehicles Laboratory (CAV-Lab) at the University of Surrey, UK. He leads multidisciplinary research on safe, trustworthy, and human-centric artificial intelligence for connected and autonomous vehicles, robotics, and intelligent transportation systems. His research spans Safe AI, neuro-symbolic reinforcement learning, autonomous vehicle safety assurance, model-based optimal control, and explainable AI, with a strong emphasis on the safe deployment of AI-enabled autonomous systems in real-world environments. Prof. Fallah collaborates extensively with industry, government, and international academic partners to advance next-generation autonomous mobility technologies. He previously served as an Independent Scientific Advisor at The Alan Turing Institute and is the recipient of the 2019 Guildford Innovation Award in the Emerging Technologies category for his contributions to autonomous systems and AI research.

Artificial intelligence is rapidly reshaping the future of mobility, offering the potential to significantly improve road safety, transport efficiency, and accessibility. However, despite remarkable advances in machine learning and autonomous driving technologies, deploying AI-driven vehicles at scale remains constrained by fundamental challenges surrounding safety assurance, trustworthiness, explainability, and regulatory acceptance. Existing AI systems often excel under expected operating conditions but struggle to provide robust, transparent, and verifiable behaviour in the complex, uncertain, and safety-critical environments encountered on public roads.

This presentation explores the broader scientific and engineering challenges that must be addressed before AI-driven vehicles can become trusted members of mixed human–robot traffic. It examines why conventional data-driven approaches alone are insufficient for safety-critical autonomy and argues for the development of AI systems capable of richer reasoning, continual adaptation, and human-centred decision-making. Particular attention is given to the need for AI that can understand context, reason under uncertainty, learn safely from experience, and provide explanations that support certification, public confidence, and regulatory approval.

The presentation also discusses how recent advances in foundation models, neuro-symbolic AI, continual learning, and trustworthy AI are creating new opportunities to overcome many of the limitations of current autonomous driving systems. These technologies have the potential to bridge the gap between high-performing AI models and the stringent requirements imposed by real-world deployment.

Finally, the presentation outlines a broader vision for the next generation of safe and trustworthy autonomous vehicles, emphasising the importance of integrating technical innovation with human values, governance frameworks, and rigorous safety assurance. Achieving this vision will require close collaboration between researchers, industry, policymakers, and regulators to ensure that future AI-driven vehicles are not only intelligent, but also demonstrably safe, transparent, and worthy of public trust.

时间：2026年7月9日（周四）14:00—15:00

地点：清华大学汽研所301
 

请通过下列二维码报名，主办方将于7月8日17:00前通过邮箱发送确认信息。