CAREER: Human-Physical Interaction-Driven Coadaptation for Resilient Human-Cyber-Infrastructure Systems

This CAREER project establishes the scientific foundations of human-physical interaction and interaction-driven coadaptation within human-cyber-physical systems (HCPS). Humans and physical systems interact bidirectionally in HCPS, actively influencing and being influenced by each other. Under external disruptions, coordinated adaptations driven by bidirectional interactions – where adaptations in each domain both inform and respond to those in the other – enhance collective resilience beyond one-sided or independent adaptations. Biological systems underscore interaction-driven coadaptation as a fundamental organizing principle: interacting species coordinate adaptations based on interaction traits to thrive under changing conditions. To achieve this goal, the project formalizes interaction-driven coadaptation between humans and civil infrastructure systems (e.g., transportation, power, and water systems). This formalization strengthens collective resilience to increasingly frequent and severe disasters by enabling coordinated adaptations between humans and infrastructure systems, reducing disruption losses and accelerating recovery. The project will advance the U.S. national interests by improving life stability, infrastructure operability, economic continuity, and response to disaster-induced disruptions. It will also promote HCPS education through an attract-train-reward pipeline, research-education integration, and public engagement. The project advances the science of HCPS by moving beyond prevailing paradigms – such as human-in/on-the-loop and human-aware systems – that often model humans as passive or exogenous agents. Instead, it establishes bidirectional human-physical interaction as the foundation for synergistic coadaptation between human and physical agents under uncertainty in coupled human-physical dynamics. Drawing inspiration from biological models of coadaptation, including replicator dynamics and evolutionarily stable strategies, the project extends these principles to HCPS to enable robust modeling and control of interactive and coadaptive human-physical dynamics for greater system adaptability and resilience. To model and control these dynamics, the project introduces four key innovations: 1) a spatiotemporal, generative, and context-aware learning framework that enables reliable and scalable modeling of human and infrastructure states under sparse and uncertain observations; 2) an asynchronous, cross-domain, and adaptive learning framework that enables reliable and generalizable modeling of lagged, distributed, and context-dependent human-infrastructure interaction dynamics based on inferred human and infrastructure states; 3) a hybrid coadaptation modeling framework that enables integration of deterministic infrastructure adaptation with behaviorally grounded stochastic human adaptation to simulate coadaptation dynamics under uncertainty; and 4) a replicator-inspired, equilibrium-aware multi-agent reinforcement learning framework that enables stable and anticipatory interaction-driven coadaptation policy optimization under multi-agent competition and non-stationary dynamics. Collectively, these contributions transform how HCPS are modeled, simulated, and optimized – opening new scientific pathways for interactive, coadaptive, and resilient system design across CPS, infrastructure, and disaster resilience domains to advance HCPS theory and the resilience of real-world systems. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria. NSF Award ID: 2542834 | Program: 01002627DB NSF RESEARCH & RELATED ACTIVIT | Principal Investigator: Kaijian Liu | Institution: Stevens Institute of Technology, HOBOKEN, NJ | Award Amount: $657,779 View on NSF Award Search: https://www.nsf.gov/awardsearch/show-award/?AWD_ID=2542834 View on Research.gov: https://www.research.gov/awardapi-service/v1/awards/2542834.html