This Cyber-Physical Systems (CPS) grant will advance structural health monitoring of concrete structures by relying on data acquired by a novel sensing technology with unprecedented scalability and spatial resolution. Modern society depends critically on sound and steadfast functioning of a variety of engineering structures and infrastructures, such as bridges, buildings, pipelines, geotechnical structures, aircrafts, wind turbines, and industrial facilities. Due to aging, massive urbanization, and climate change, there is a growing need for accurate and reliable assessment of the health condition, performance, and operation of these structures in order to ensure their continuous functioning and safe use. The researched technology enables pervasive and scalable sensing of concrete structures with high resolution by transforming concrete into a smart self-sensing material, thereby enabling reliable long-term structural health monitoring. This in turn contributes to the nation?s sustainability and resilience and to advancing the nation?s prosperity, welfare, and security. The project advances multiple core research areas in structural health monitoring including CPS system architectures using embedded devices, multi-parameter sensing and networking based on radio frequency sensors, and machine learning for accurate and reliable data analytics. The research outcomes are highly translational to various other CPS domains. The project also contributes to secondary education and outreach activities in multiple ways as well as to undergraduate and graduate education.

The aim of this project is to create a novel sensing system comprised of radio frequency sensors that are pervasively embedded in large volumes of concrete structures and that sense their localities using radio frequency properties. The objective is the assessment of key parameters that reflect the behavior of the monitored structure under operational conditions, such as deformation, temperature, and humidity, as well as detection and characterization of damages. The project has the following intellectual contributions: 1) Passive radio frequency-based sensing that operates over a wide range of frequencies; architectures of smart exciters and networked radio frequency sensors that communicate among themselves via backscatter modulation; solar-powered radio frequency exciter platform that powers the sensors. 2) Energy-based sensing and network optimization of the radio frequency sensor network in terms of its monitoring ability and network connectivity given the constraints on the available harvested power at the exciters. 3) Machine learning methods for function estimation based on the principle of ensemble modeling with Gaussian processes and applied to self-localization and to inference of three-dimensional distributions of material parameters within large volumes of concrete structures.

The vision for Advanced Air Mobility (AAM) or formerly Urban Air Mobility (UAM) is to enable an air transportation system that moves people and cargo between places previously underserved by the current aviation market (local, regional, intraregional, urban) using revolutionary new electric vertical take-off and landing (eVTOL) aircraft. AAM has received significant attention from federal agencies. Companies around the globe are competing to build and test eVTOL aircraft to ensure the AAM will become an integral part of people?s daily life. The AAM has enormous economic potential and societal impact, but its success will depend on its ability to scale the operations to the expected high demand with safety guarantee.

This project lays the foundation of safe and scalable learning-based planning and control for autonomous air mobility. Concretely, the project will (i) focus on algorithmic advances of scalable multi-agent aircraft autonomy for real-time separation assurance to increase the airspace capacity; (ii) develop and integrate the online safety guard and offline adaptive stress testing model to provide safety enhancement for the multi-agent aircraft autonomy; (iii) design the collaborative traffic flow planning framework for flight operators and the airspace service provider to improve safety and efficiency when facing demand and capacity uncertainties on the AAM network; and (iv) integrate the developed models and algorithms to build an autonomous AAM ecosystem testbed to perform simulation/flight tests and system level validation. The multidisciplinary approach is based on multi-agent reinforcement learning, safe reinforcement learning, multi-agent stochastic game, and bi-level robust optimization. The proposed effort has transformative impacts to enable safe and scalable advanced air mobility. It could have impact in the way that other CPS tools are designed and implemented to support increasing autonomy and unmanned operations in civil aviation, autonomous cars/trucks, and robotics. The project has an integrated education plan in (i) student innovation competitions, teams and clubs; (ii) interdisciplinary curriculum development and improvement for AI and autonomy in aerospace; (iii) bringing industry experts to students in classroom; and (iv) international student research exchange. The project will engage elected officials and policy makers in AI and machine learning via podcast series, which will provide basic knowledge and insights on legal, ethical, and societal implications of AI. The project will establish a workforce pipeline from high school to postdoc for women in in aerospace via AI and computing.

This Cyber-Physical Systems (CPS) project aims at designing theories and algorithms for scalable multi-agent planning and control to support safety-critical autonomous eVTOL aircraft in high-throughput, uncertain and dynamic environments. Urban Air Mobility (UAM) is an emerging air transportation mode in which electrical vertical take-off and landing (eVTOL) aircraft will safely and efficiently transport passengers and cargo within urban areas. Guidance from the White House, the National Academy of Engineering, and the US Congress has encouraged fundamental research in UAM to maintain the US global leadership in this field. The success of UAM will depend on the safe and robust multi-agent autonomy to scale up the operations to high-throughput urban air traffic. Learning-based techniques such as deep reinforcement learning and multi-agent reinforcement learning are developed to support planning and control for these eVTOL vehicles. However, there is a major challenge to provide theoretical safety and robustness guarantees for these learning-based neural network in-the-loop models in multi-agent autonomous UAM applications. In this project, the researchers will collaborate with committed government and industry partners on the use-case-inspired fundamental research, with a focus on promoting safety and reliability of AI, machine learning and autonomy in students with diverse backgrounds.

The technical objectives of this project include (1) Safety and Robustness of Single-Agent Reinforcement Learning: in order to address the ?safety critical? UAM challenge, the PIs plan the min-max optimization for single agent reinforcement learning to formally build sufficient safety margin, constrained reinforcement learning to formulate safety as physical constraints in state and action spaces, and the novel cautious reinforcement learning that uses variational policy gradient to plan the safest aircraft trajectory with minimum distributional risk; (2) Safety and Robustness of Multi-Agent Reinforcement Learning: in order to address the ?heterogeneous agents and scalability? challenge, a novel federated reinforcement learning framework where a central agent coordinates with decentralized safe agents to improve traffic throughput while guaranteeing safety, and a scaling mechanism to accommodate a varying number of decentralized aircraft; (3) Safety and Robustness from Simulations to the Real World: in order to address the ?high-dimensionality and environment uncertainty? challenge, the researchers will focus on the agents? policy robustness under distribution shift and fast adaptation from simulation to the real world. Specifically, value-targeted model learning to incorporate domain knowledge such as the aircraft and environment physics, and a safe adaptation mechanism after the RL model is deployed online for flight testing or execution is planned.

Artificial Neural Networks (ANN) are increasingly being employed to monitor and control Cyber-Physical Systems (CPS), as in autonomous ground and aerial vehicles. With increasing complexity and safety criticality of these systems, formal-verification techniques that provide rigorous guarantees are urgently needed. The broad goal of the research is to develop novel algorithms and software tools for formal verification of ANN-controlled CPS (ANN-CPS). One of the main challenges of existing techniques is their scalability to large number of neurons and complex physical dynamics.

Using the novel concept of Interval Neural Networks, coupled with ideas from formal methods such as counter-example guided abstraction refinement and approximate bisimulation, the project investigates scalable formal verification techniques for ANN-CPS. The results of the project will enable rigorous analysis of complex ANN-CPS possible, thereby enhancing their reliability in applications such as autonomous driving. Further, the PI is engaged in course development, mentorship of undergraduate and graduate students, and outreach activities for K-12 students, with the broader aim of motivating and building the workforce for formal analysis of cyber-physical systems.

Light Detection and Ranging (LiDARs) and cameras are an indispensable part of the sensor suite used in autonomous cyber-physical systems such as self-driving cars and unmanned aerial vehicles. The data generated by these sensors is often processed by a deep neural network that transforms it into state estimates used in control loops. Although one can analyze the impact that erroneous state estimates have on control loops, less is known about how to characterize the errors produced by deep neural networks. The objective of this project is to develop analysis and design techniques that provide formally guaranteed bounds on how large these errors can be. Formally establishing error bounds will enable the verification of existing systems as well as the design of new autonomous systems for which formal guarantees of safety and performance can be given.

This project addresses the challenge of using deep neural networks in the perception pipeline of autonomous cyber-physical systems by following two different approaches, termed correctness-by-training and correctness-by-supervision. The first approach, correctness-by-training, is based on the use of monotone neural networks for which deterministic generalization bounds can be established. The challenge of using monotone neural networks is that their training is more challenging and several novel training techniques will be investigated. The second approach, correctness-by-supervision, consists of attaching a supervisor to the neural network that overrides the network output so as to enforce guaranteed error bounds. A supervisor will be developed in the context of localization using LiDAR measurements using novel point-set registration techniques based on moments. Both approaches aim to provide guaranteed error bounds on the state estimates computed by deep neural networks. The ultimate contribution is to use these error bounds in the formal analysis of safety and performance of control loops using deep neural networks in the perception pipeline.

The goal of this project is to improve how perception uncertainty is modeled in Cyber-Physical Systems (CPS). Complex autonomous CPS, from airplanes and unmanned aerial vehicles to future self-driving cars, permeate our daily lives. These systems consist of many interdependent components operating in highly uncertain environments and exhibiting complex dynamics. This interdependency makes not only their modeling harder but also quantifying their robustness more difficult. A single undetected faulty reading in sensors, delay in processing or error in communication protocols can lead to catastrophic events such as airplane or car accidents. Such events can lead to loss of life as well as fear or loss of confidence in the public. The approach of this project is to consider uncertainty as a function of time, rather than static estimates, which will enable researchers to quantify the robustness of the overall system. The broader impacts of the project include organization of a drone competition to be held at the University of Southern California (USC).

The modeling of uncertainty and reasoning about the robustness of highly complex CPS designs is crucial. While post-hoc analysis calls for improving sensing technology or fault tolerance through redundant sensors (or sensor fusion), in this project, we construct mathematical and algorithmic foundations to address research challenges in (1) mathematical models of time-varying uncertainty; (2) modeling of interdependent CPSs for analysis of interdependence as well as environment uncertainty; (3) quantification of robustness against such uncertainty; and (4) design of control strategies for these systems. Our approach is to develop a temporal logic-based framework for these complex interconnected CPS models. Augmenting formal specification techniques based on temporal logic with notions from statistics and information theory enables our framework to engineer high-confidence CPS applications that are adaptable and resilient.

Despite claims in popular media, current ?self-driving? and advanced driver assist systems (ADAS), based on purely data-driven, machine learning algorithms may still suffer from catastrophic failures. This tendency of ?theoretical statistical accuracy? but ?demonstrated fragility in practice? makes current deep learning algorithms unsuitable for use within feedback loops for safety-critical, cyber-physical applications such as assisted or unsupervised self-driving cars in traffic. Regardless of these shortcomings, it is certain that automation and autonomy will play a crucial role in future mobility solutions, either for personally owned or shared-mobility vehicles; and regardless of the level of automation, at least in the foreseeable future, the driver should be in the loop. There is currently a need to quantify the impact of the human driver within the autonomy loop, both from an individual experiential perspective, as well as in terms of safety. In addition, the next generation of ?self-driving? or ?driver-assist? systems should be able to sense, learn and anticipate driver?s habits, skills and adapt accordingly, thus making driving more intuitive and safer at the same time. How to best integrate the driver?s learning goals and preferences in a transparent manner to enhance the ?driving experience? without sacrificing safety requires further work, however.

The main objective of this research is to utilize techniques and models from reinforcement learning and formal methods to develop the next generation of ADAS that can accommodate the driver preferences and habits with safety constraints. The aim is to increase the performance and safety guarantees of deep neural network architectures operating within a feedback loop that includes the driver by: a) using redundant architectures that blend model-free and model-based processing pipelines; and b) adding safety guarantees both during training and during execution by leveraging recent advances of formal methods for safety-critical applications. Specifically, the technique consists of learning a state prediction model to estimate the internal reward function of the driver using a novel neural network architecture, accompanied by a federated, lifelong learning approach to identify heterogeneous driver preferences and goals. The proposed approach will further add a layer of safety and robustness by incorporating the neural network architecture with a differentiable Signal Temporal Logic (STL) framework to meet temporal safety constraints, and will meet with an additional safety layer using a run-time assurance (RTA) mechanism that combines reachability analysis with a monitoring approach to ensure that system cannot be steered to unsafe conditions. The proposed framework will be validated and tested in two stages. The first stage will involve simulations and experiments on several non-trivial problems using high-fidelity driving simulation platforms such as CARLA. The second stage will conduct human-in-the-loop experiments using a driving simulator developed at Georgia Tech. The research will involve both graduate and undergraduate students. The results of this research will be disseminated to the community by journal and conference publications, organization of invited workshops and seminar presentations, and by targeted exposure (press releases, interviews) to popular media.

Buildings are vitally important because they contribute to the well-being and productivity of their occupants - however, these benefits come at a high environmental cost. Collectively, buildings account for 40% of the US primary energy usage and CO2 emissions and 70% of the electricity consumption. Furthermore, buildings put a tremendous strain on the power grid as they are largely responsible for the peaks in energy demand. Making buildings smarter through the deployment of sensors, actuators, and controllers, which collectively serve as the backbone of building cyber-physical systems (CPS), can achieve more than 30% annual energy savings and can also significantly smooth peak demand. Thus, smart buildings are vital to a sustainable energy future. However, the road to large-scale realization of smart buildings is inhibited by their heterogeneity, which requires engineering customized, site-specific, and, thereby, costly solutions.

The goal of this project is to develop a CPS solution for autonomous buildings that will enable non-expert building managers to deploy asset-specific, smart control policies. The advantage of the proposed solution relies on the fact that the approach can be applied on a large-scale even without any human intervention. The resulting software solution is the Artificial-Intelligence-Enabled Building Energy Expert (AI-BEE) and it will be demonstrated using simulations and experiments at the Center for High Performance Buildings at Purdue University. The proposed research will result in foundational contributions in core CPS areas, including machine learning and control, that will be translational to other application areas, such as large-scale energy systems (power grid), transportation, civil infrastructure, and unmanned vehicles.

The technical details of our approach are as follows. First, a taxonomy of building types is being developed. The idea is that the energy behavior of every building should be completely specified by a finite set of variables in a machine-readable format. Second, each complete building description is associated with a set of dynamical systems that describes the energy consumption. In this way, non-experts will be able to specify building characteristics and get a set of plausible dynamical systems that include a description of the building. This set of dynamical systems is what is called the relevant model universe to the building at hand. Third, meta reinforcement learning is being used to discover a self-improving control algorithm that works well for all dynamical models in the relevant model universe. The final step is to deploy the discovered algorithm to the building and let it self-improve further.

Cyber-physical systems (CPS), such as automobiles, planes, and heavy equipment rely on complex distributed supply chains that source parts from manufacturers across the world. A fundamental problem that these systems face is ensuring the safety, security, and integrity of both the cyber components and physical parts that they receive through their supply chain. Because of the separation between the manufacturer and the consumer of the part, there are immense challenges in ensuring that physical parts arrive from the desired source and are not modified or swapped for inferior copies in transit. For example, the Aerospace Industries Association states that "though we know counterfeit parts enter the aerospace supply chain, the time and place of their entry is unpredictable." If either the cyber-components or the physical parts being incorporated into these systems have been tampered with, significant cyber-physical security risk is introduced. As an example, an attacker who has a part's cyber-information can simply produce a counterfeit part, clone any physical identifiers (serial numbers, etc.), and claim that the cyber-information is for the cloned part.

While cyber-security techniques, such as roots of trust and signing chains, exist to help ensure software integrity, there are no commensurate roots of trust and signing chains that can guarantee the source and integrity of both the cyber components and physical parts. As such, there is a risk that the algorithms and control approaches used in a supply chain will not identify the inferior performance characteristics of a counterfeit part and control its operation in an unsafe manner. The primary goal of this research is to create an integrity mechanism based on physically unclonable functions to ensure that an entire CPS is built from both trusted software and physical parts. To achieve this goal, the research investigates (i) a physical measurement technique (electro-mechanical impedance) to provide parts an unclonable physical identity and (ii) the cyber signing approaches to build chains of trust from these identities.

There is no question that indoor environments are often uncomfortable or unhealthy for occupants. This is an even more critical issue in healthcare facilities, where patients may experience the stressful effects of poor thermal, luminous, and acoustic environments more acutely. With complementary expertise from engineering and psychology, the proposed research is focused on creating a human-on-the-loop, responsive indoor environmental system with the potential to offer better quality of care in hospitals. The outputs of this project will have profound societal impacts on the wellbeing of both healthy individuals and on recovering sick individuals. Research outcomes will enable real time human-built environment interaction to minimize stress and optimize performance in any built environment, and ultimately lead towards economic benefits achieved through wellness and higher productivity. Improved indoor environmental quality in hospital settings will improve patient healing, which is an important societal benefit. Similar strategies can be used for educational facilities, and office buildings. This research encourages Broadening Participation through inclusion of individuals from underrepresented groups (female and Latinx Co-PIs), female and minority students, and a minority serving lead institution from an EPSCoR state. Results will be disseminated broadly through scientific publications and seminars, and K-12 outreach, including STEM competitions, and summer programs.

Indoor environmental quality (IEQ) not only impacts the physical health of patients, but also their psychological health. Yet environmental controls for heating, cooling and ventilation, noise attenuation, and lighting in hospitals are based on outdated models of how hospitals function, who occupies these settings, and what emerging technologies are available. As a result, many hospitals are just functionally adequate, often likely to be too cold or hot, too loud, or too bright. In order to capitalize on the healing potential of the hospital?s built environment, we propose a three-year collaborative effort between the University of Hawaii at Manoa, Arizona State University, and Drexel University to develop innovative biosensor technologies, deep-learning health data analytics, and user-centric control algorithms to connect these three domains in which the interdependencies of the physiological, physical, and psychological will be investigated, quantified, and addressed. The team is partnering with the Children?s Hospital of Philadelphia (CHOP) to validate the approach. Specific anticipated engineering/science contributions include: 1) innovative cyber-physical system architecture using heterogeneous biosensing and data analytics for real-time control; 2) new sensor fusion based technology for non-invasive, precise physiological measures that are surrogate stress indicators; 3) progressive development of innovative human centric deep model linking physiological biometrics to psychological measures, and connecting environmental factors to psychological measures facilitated with physiological biometrics; 4) new stress responsive real-time supervisory control strategies including optimal environmental adjustment, and 5) multi-level system evaluation via virtual, laboratory, and field testing at a hospital environment at CHOP.