Cyber-Physical Systems

Cyber-physical systems connect sensors, software, networks, and controllers to physical processes, so computational decisions can affect safety and real-world operation. Researchers use model-based control and digital twins alongside formal methods such as Signal Temporal Logic verification, and data-driven approaches including reinforcement learning and graph-based anomaly detection. These methods are applied to settings such as industrial control systems, power grids, autonomous robots, and connected vehicles, where they must handle faults, cyberattacks, changing conditions, and tight timing constraints. A central challenge is assuring safety and security when learned or language-model-based components are difficult to verify, while testing them under realistic physical and network conditions.