The objective of this research is to meet the urgent global need for improved safety and reduced maintenance costs of important infrastructures by developing a unified signal processing framework coupling spatiotemporal sensing data with physics-based and data-driven models. The approach is structured along the following thrusts: investigating the feasibility of statistical modeling of dynamic structures to address the spatiotemporal correlation of sensing data; developing efficient distributed damage detection and localization algorithms; investigating network enhancement through strategic sensor placement; addressing optimal sensor collaboration for recursive localized structural state estimation and prediction. Intellectual merit: This innovative unified framework approach has the potential of being more reliable and efficient with better scalability compared to the current state-of-the-art in structural health monitoring. The proposed research is also practical as it allows analysis of real-world data that accounts for structural properties, environmental noise, and loss of integrity over sensors. Probabilistic representation of significant damages allows more informative risk assessment. Broader impacts: The outcome of this project will provide an important step toward safety and reliability albeit increasing complexity in dynamic systems. New models and algorithms developed in this project are generic and can contribute in many other areas and applications that involve distributed recursive state estimation, distributed change detection and data fusion. This project will serve as an excellent educational platform to educate and train the next generation CPS researchers and engineers. Under-represented groups such as female students and Native American students will be supported in this project, at both the graduate and undergraduate levels.

The objective of this research is to develop methods to monitor and ensure the robustness of a class of cyber-physical systems termed "physical networks," such as electric, water, sewage, and gas networks. The approach is to analyze such networks using the mathematical formalism of graphical models. The project models a physical network as a graph, whose variables have a concrete physical interpretation, such as voltage, satisfying known physical laws, such as Kirchoff laws. The machinery of graphical models is used to develop methods to monitor and ensure the robustness of such networks, using the electric power network as a representative. By studying puzzling network-wide interactions, the project has the potential to clarify the role of complexity in large scale networks. Potential contributions will be made to the fields of distributed inference algorithms and fast numerical methods. Physical networks play a crucial role in modern society, and yet, often exhibit fragile behavior, such as black-outs in electric power networks, resulting in economic loss, as well as causing a security risk. This project seeks to understand the robustness behavior of such networks and to train a broad class of students in their theory and practice. Results from this research are to be incorporated into courses and disseminated via research publications. The Carnegie Mellon Conference on the Electricity Industry allows students to interact with faculty and electricity industry veterans. Interaction with the electricity industry aims to provide it with an understanding of cyber-intelligence, to ensure effective robustness monitoring capabilities in the power grid.

The objective of this research is to scale up the capabilities of fully autonomous vehicles so that they are capable of operating in mixed-traffic urban environments (e.g., in a city such as Columbus or even New York or Istanbul). Such environments are realistic large-city driving situations involving many other vehicles, mostly human-driven. Moreover, such a car will be in a world where it interacts with other cars, humans, other external effects, and internal and external software modules. This is a prototypical CPS with which we have considerable experience over many years, including participation in the recent DARPA Urban Challenge. Even in the latter case, though, operation to date has been restricted to relatively “clean” environments (such as multi-lane highways and simpler intersections with a few other vehicles). The approach is to integrate multidisciplinary advances in software, sensing and control, and modeling to address current weaknesses in autonomous vehicle design for this complex mixed-traffic urban environment. All work will be done within a defined design-and-verification cycle. Theoretical advances and new models will be evaluated both by large-scale simulations, and by implementation on laboratory robots and road-worthy vehicles driven in real-world situations. The research address significant improvements to current methods and tools to enable a number of formal methods to move from use in limited, controlled environments to use in more complex and realistic environments. The theory, tools, and design methods that are investigated have potential application for a broad class of cyber-physical systems consisting of mobile entities operating in a semi-structured environment. This research has the potential to lead to safer autonomous vehicles and to improve economic competitiveness, the nation's transportation infrastructure, and energy efficiency. The richness of the domain means that expected research contributions can apply not only to autonomous vehicles but, also, to a variety of related cyber-physical systems such as service robots in hospitals and rescue robots used after natural disasters. The experimental research laboratory for the project is used for undergraduate and graduate courses and supports new summer outreach projects for high-school students. Research outcomes are integrated with undergraduate and graduate courses on component-based software.