projects

My research develops system-level frameworks for online and adaptive learning-enabled control in large-scale cyber-physical systems. A recurring theme is that learning and control should be designed around the physical realities of dynamic environments: safety constraints, bounded or adversarial disturbances, unreliable predictions, communication delays, and limited sensing.

The projects below are organized around two connected directions: foundations for safe, adaptive, and scalable online control, and applications in sustainable energy and industrial systems.

Foundations

Adversarial Stabilization

Online, adaptive stabilization of unknown dynamics under worst-case, non-stochastic perturbations.

System Identification Under Bounded Noise

Online system identification with finite-time, minimax-optimal rates under bounded disturbances.

Scalable, Communication-Constrained Control

Online and adaptive distributed control with locality, delay, and communication constraints.

Prediction-Augmented Control

Online, adaptive networked control with unreliable ML-generated forecasts.

Applications

Learning-Enabled Control for Sustainable Energy Systems

Online, adaptive, uncertainty-aware control for power grids and buildings.

ML for Manufacturing

Online, adaptive monitoring and process analytics for industrial manufacturing.