Shufang Zhu

Shufang Zhu

(she/her/hers)

Sapienza University of Rome

Automated Reasoning, Planning, Agent Program Synthesis

Shufang Zhu is a Postdoctoral researcher at the Dipartimento di Ingegneria Informatica, Automatica e Gestionale. Sapienza University of Rome, working with Prof. Giuseppe De Giacomo on his Advanced ERC project WhiteMech. Her research concerns interdisciplinary knowledge across artificial intelligence (AI) and formal methods (FM), focusing on automated reasoning, planning and synthesis. She received her Ph.D. degree in 2020, at East China Normal University (ECNU), Shanghai, China, under the supervision of Prof. Geguang Pu. During her Ph.D., she received the scholarship from Chinese Scholarship Council (CSC) and studied as a visiting Ph.D. student (August 2016 to Feb 2018) at Rice University, under the supervision of Prof. Moshe Y. Vardi. 

 Towards Resilient and Trustworthy Autonomous Agents

Autonomy is one of the grand objectives in artificial intelligence (AI), which aims at building AI agents that autonomously deliberate how to act in a changing, incompletely known, unpredictable environment, especially in a challenging environment when human instructions are lacking or delayed. Crucially, an intelligent agent must be trustworthy, as empowering autonomous agents with the ability to self-deliberate carries significant risks. Further, intelligent agents should also be resilient, which allows them to recover gracefully from disruptive changes.

My research vision is to develop trustworthy and resilient autonomous agents through an interdisciplinary research area of AI and formal methods (FM) – Agent Program Synthesis (APS), a fundamental mechanism for autonomous deliberation under formal guarantees. Towards this goal, my first research thrust aims at synthesizing trustworthy autonomous agents of tasks specified in Linear Temporal Logic on finite traces (LTLf). My second research thrust aims to empower trustworthy autonomous agents with resilience wrt disruptions from the environment. My third research thrust tackles the challenge of boosting trustworthy autonomous agents with resilience wrt disruptions from the agent itself.

My long-term research vision is to make autonomous agents trustworthy, resilient, safe, and reliable for applications in safety- and security-critical domains.