Resilient {CPS}

Safety is critical in Cyber-Physical Systems (CPS) applications, including autonomous driving systems, power grids, and medical robots, and faces multiple challenges.

CPS are deployed in environments with sensor noise, faults, actuator failures, and adversarial attacks. Adversarial environments invalidate assumed dynamical and measurement models of the system, rendering safety guarantees that are based on nominal models inapplicable.

The goal of my research is to design safe CPS and ensure continued safe operation under faults and attacks (resilience). My background allows me to seamlessly integrate control theory, machine learning, and security, enhancing interdisciplinary collaboration.

In the area of resilient CPS, my research aims to construct robust and safe controllers in the presence of faults, disturbances, and adversarial attacks. We constructed the first Fault-Tolerant (FT) CBFs in \cite{clark2020control} to ensure joint safety and stability under sensor spoofing attacks. We generalized our FT scheme in \cite{zhang2022safe} to incorporate high relative degree CPS under sensor and actuator faults as well as neural CBFs \cite{zhang2024fault}.

I have also investigated attack-resilience from a system security perspective in my recent work that focuses on spoofing attacks on LiDAR sensors.