Research thrusts

Cyber-physical AI

How does a system that senses the physical world turn what it senses into decisions it can be trusted to act on?

This thrust in 25 seconds

Transcript

Research thrust Cyber-physical AI. How does a system that senses the physical world turn what it senses into decisions it can be trusted to act on? Lines of work: Digital twins, sensing, and state recovery; Learning without moving the data; High-performance computing and data integrity; Agentic control within limits. Cyber-physical AI. Co-leads: Yan Luo and Seung Woo Son smartcyberphysical.org/research-cpai.html

This thrust owns the AI methods the other thrusts apply: sensing and fusion, state estimation, learning across sites without moving data, and computing that keeps its data honest under load. It builds machine awareness of what a physical system is doing, made to feed the people who decide, and it puts limits on what an AI agent may do to infrastructure on its own.

What the group builds

Digital twins, sensing, and state recovery

Live models that stay in step with a grid, a network, or a manufactured part, built from incomplete and sometimes corrupted measurements with physics in the model, not only data: SUMMIT's real-time simulators with hardware in the loop, NATIG's co-simulation of a grid and its wireless network, twins delivered from hybrid clouds, and physics-informed models that predict a part's reliability without destroying it.

Learning without moving the data

Federated learning across utilities, hospitals, and trial sites, so the model travels and the data does not, with the privacy guarantees written in from the start.

High-performance computing and data integrity

Compression and anomaly detection for scientific data, and computing systems that detect silent data corruption before it reaches a result.

Agentic control within limits

AI agents that plan and act on physical systems, from autonomous robotic planning for the Army to grid-aware data centers, with the guardrails that decide what they may do without a person.

Recent papers

The newest work from the faculty on this thrust. 99 papers since 2019 carry one of their names.

  1. H. Rabbani, A. Rezaee, F. Arpanaei, H. Rabbani, V. M. Vokkarane, M. Brandt-Pearce
    Optical Fiber Technology, vol. 103, art. 104806, Jan. 2027
  2. C. A. Ng, S. Pagsuyoin, Y. Luo
    Sensors and Actuators A: Physical, vol. 410, art. no. 118210, Nov. 2026
  3. A. Rezaee, F. Arpanaei, R. McCann, H. Rabbani, J. A. Hernández, M. Brandt-Pearce, V. M. Vokkarane
    IEEE/Optica Journal of Optical Communications and Networking, vol. 18, no. 10, Oct. 2026IF 5.1 (2025)
  4. C. Pozzi, C. Ng, S. Lyon, Y. Luo, C. Niezrecki, M. Inalpolat
    Wind Energy, vol. 29, no. 10, Sept. 2026IF 4.1 (2023)
  5. A. Rezaee, R. McCann, V. M. Vokkarane
    IEEE/Optica Journal of Optical Communications and Networking, vol. 18, no. 9, pp. D90-D105, Sept. 2026 (Special Issue on Benchmarking in Optical Networks)IF 5.1 (2025)
  6. A. Rezaee, F. Arpanaei, R. McCann, L. Nadal, J. A. Hernández, V. M. Vokkarane
    IEEE/Optica Journal of Optical Communications and Networking, vol. 18, no. 8, pp. C160-C172, Aug. 2026IF 5.1 (2025)
  7. Y. Jeong, A. Moon, H. Lee, S. W. Son
    IGARSS 2026 - 2026 IEEE International Geoscience and Remote Sensing Symposium, pp. 3035-3035, Aug. 2026
  8. B. R. Jyoti Arka, M. Z. Islam, Y. Lin, V. M. Vokkarane, J. Zhao
    IEEE Power & Energy Society General Meeting (PESGM), pp. 1-5, July 2026

All publications