Abstract: 

In recent years, we have witnessed rapid advances in Assured Autonomy through reactive synthesis for developing trustworthy autonomous Artificial Intelligence (AI) systems, with a strong focus on efficient and scalable techniques that are soon ready for applications. However, autonomous systems are increasingly deployed in dynamic, uncertain, and unpredictable environments, where unexpected situations are not the exception but the norm. In such settings, resilience, the ability to gracefully recover from unforeseen circumstances, is not just desirable, but essential. While resilience has long been a foundational principle in distributed systems, current approaches to autonomous AI largely lack mechanisms to detect, handle, and recover from exceptional or unforeseen conditions. As a result, the challenge of embedding resilience into autonomous systems, alongside trustworthiness, remains largely unaddressed. In this talk, I will present our recent work on tackling this critical gap. We explore new approaches in automated reactive synthesis that go beyond trustworthiness and efficiency, integrating formal guarantees with mechanisms for exception handling and recovery. These advances represent a step toward more resilient and dependable autonomous systems.

Short bio:
Shufang Zhu is a Lecturer (Assistant Professor) at the Department of Computer Science, University of Liverpool. She is also an Associate Member of the Department of Computer Science, University of Oxford. Prior to joining Liverpool, she was a Senior Research Associate at the Department of Computer Science, University of Oxford. Her expertise lies in the interdisciplinary research area of artificial intelligence (AI) and formal methods (FM), with a focus on automated planning and synthesis. She was selected as a Future Digileader by Digital Futures, Sweden in 2023 and an UT Austin EECS Rising Star in 2022.