Water and nitrogen represent two of the most expensive inputs to agricultural systems, and two of the critical constraints on overall agricultural productivity. Today, farmers generally over apply nitrogen fertilizer, because the potential cost of over application is less than the potential cost of achieving suboptimal yields. Similarly, in farm settings water is often over applied, particularly when studies are conducted at high resolution within individual center-pivot fields. We willdesignand validatean integrated cyber-physical system to collect and integrate data from remote sensing and low-cost field deployed wearable sensors and use machine learningand mathematical modeling to guide precision water and nutrient interventions in farmer's fields. This would mean that agricultural productivity can be sustained or increased while reducing overall nitrogen fertilizer and irrigation applications. Among the many beneficial effects to society as a whole would be 1) a decrease the environmental impact of agriculture; 2) decreased competition for scarce water supplies between agriculture and growing urban centers; and 3) increased farmer profitability, improving the economic viability of rural economies.The CPS will enable fusion of a large volume of spatio-temporally distributed multi-modal information to create a data-driven decision support platform that provides actionable information on optimal agricultural managementstrategies.The team will continue to leverage and develop extensive outreach and educational activities to train the next generation of scientists, through many existing STEM programs in Iowa State University and University of Nebraska-Lincoln.

Water and nitrogen represent two of the most expensive inputs to agricultural systems, and two of the critical constraints on overall agricultural productivity. Today, farmers generally over apply nitrogen fertilizer, because the potential cost of over application is less than the potential cost of achieving suboptimal yields. Similarly, in farm settings water is often over applied, particularly when studies are conducted at high resolution within individual center-pivot fields. We willdesignand validatean integrated cyber-physical system to collect and integrate data from remote sensing and low-cost field deployed wearable sensors and use machine learningand mathematical modeling to guide precision water and nutrient interventions in farmer's fields. This would mean that agricultural productivity can be sustained or increased while reducing overall nitrogen fertilizer and irrigation applications. Among the many beneficial effects to society as a whole would be 1) a decrease the environmental impact of agriculture; 2) decreased competition for scarce water supplies between agriculture and growing urban centers; and 3) increased farmer profitability, improving the economic viability of rural economies.The CPS will enable fusion of a large volume of spatio-temporally distributed multi-modal information to create a data-driven decision support platform that provides actionable information on optimal agricultural managementstrategies.The team will continue to leverage and develop extensive outreach and educational activities to train the next generation of scientists, through many existing STEM programs in Iowa State University and University of Nebraska-Lincoln.

Harnessing recent progresses on wastewater treatment, food security, data analytics, and machine intelligence, we propose to study novel optimized technology-driven controlled-environment agriculture (CEA) systems that can achieve high areal vegetable productivity to increase the food and nutritional security in urban areas with low operating cost and reduced energy consumption.Our project focuses on two core CPS research areas, i.e., control and data analytics, inspired by the design and operations of a pilot testbed at Georgia Tech for coupling the water and nutrients in domestic wastewater (DWW) to high-productivity CEAs. Food production in the CEAs must warrant the high cost of land in urban areas, which makes it necessary to reduce the total DWW-CEA operating cost and increase productivity. However, it is highly challenging to control and optimize this complex system of subsystems. In our case, we need to coordinate the Pilot-Plant and Pilot-Farm, examine their inter-correlation, and support dynamic and robust optimal decisions to achieve the highest production yield, while simultaneously satisfying various performance specifications on nutrient compositions, operating cost and energy consumption, and meeting other safety requirements of DWW-CEAs. Moreover, the profound impact of numerous operating conditions and parameters on the vegetable phenotype, yield and nutrient compositions during different growth periods need to be thoroughly understood. We have to move from model-driven CPS fundamentals to an integrated data-driven model-based approach. The objective of this 3-year interdisciplinary project is to maximize food productivity and nutrition while minimizing cost, energy and waste. The research is organized as four thrusts, each objective-driven and delineated as follows.InThrust 1, first-principles water models are adapted to the physical system and calibrated before on-site experimental validation. As first-principles models are not available for hydroponic agriculture, inThrust 2we adopt a hybrid approach where parameter-dependent biochemical reactions are supported by machine-learning algorithms with the purpose of parameter estimation and capture of possible un-modeled dynamics. A distinguishing feature ofThrust 2is the implementation of non-invasive spectroscopy and imaging to gain additional plant morphology and nutrition information. The subsystems (e.g., water and hydroponics) are then integrated as a system-wide model and validated experimentally. InThrust 3, we devise control algorithms with the purpose of achieving desired closed-loop performance despite the presence of disturbances and/or uncertain dynamics. As a part ofThrust 4, the resulting CPS is implemented in simulation software and experimentally validated to determine a feasible deployment scale. We expect that our novel, foundational research contributions will be immediately transferable to the water, waste, agriculture and remote sensing industries, but benefit other complex system control applications entirely.

The goal of this project is to use proprioception, localized sensing, and observed forces to develop robust, autonomous fruit picking methods. Fresh market tree fruit growers still rely on a large seasonal labor force for harvesting operations. Despite extensive research over the past thirty years, robotic harvesters are not yet commercially available. Prior work has considered manipulation a robot position control problem, disregarding the need for sensor input after physical contact with the fruit. However, when picking fruit such as apples and pears, professional pickers use active perception, incorporating both visual and tactile input about fruit orientation, stem location, and the fruit's immediate surroundings. We propose to embrace this physical contact by incorporating a rich set of in-hand sensors in an extended manipulation feedback loop with the goal of providing fine control over how the fruit is separated from the tree. To overcome the constraints of data collection in the field, we will develop a learning framework for compartmentalizing the tasks and design an instrumented proxy to serve as a training environment.While our primary focus in this project is fresh market apple and pear harvesting, we believe that this framework will be useful for numerous other agricultural applications that involve physical manipulation. For example, harvesting methods used for greenhouse sweet peppers and tomatoes are highly dependent on knowledge of peduncle orientation. However, automating production has been difficult due to similar challenges with occlusions and determining crop orientation. Another potential area of application for this learning framework is plant phenotyping, using soft tactile sensors, in addition to other sensor types, to measure a plant's physical properties.

The potential economic and societal impacts of realizing fully autonomous cyber-physical systems (CPS) are astounding. If the Federal Aviation Administration (FAA) allows integration of unmanned aerial vehicles (UAVs) into the national civilian airspace, the private-sector drone industry is estimated to generate more than 100K high-paying technical jobs over a ten-year span and contribute $82B to the U.S. economy. Self-driving cars are predicted to annually prevent 5M accidents and 2M injuries, conserve 7B liters of fuel, and save 30K lives and $190B in healthcare costs associated with accidents in the U.S. Successful mission pursuit of such fully autonomous CPS hinges on possessing full situational awareness including precise knowledge of its own location. Current CPS are far from possessing this capability, particularly in dynamic, uncertain, poorly modeled environments where GPS coverage may be spotty, obscured, or otherwise impaired. This necessitates developing a coherent analytical foundation to deal with this emerging class of CPS, in which situational awareness and mission planning and execution are intertwined and must be considered simultaneously to address uncertainty, model mismatch, and compensate for potential GPS coverage gaps.

This project is has four main objectives: (1) Analyze the observability of unknown dynamic, stochastic environments comprising multiple agents. This analysis will establish the minimum a priori knowledge needed about the environment and/or agents for stochastic observability. (2) Develop adaptation strategies to refine the agents models of the environment, on-the-fly, as the agents build spatiotemporal maps. Adaptation is crucial, since it is impractical to assume that agents have high-fidelity models describing the environment. (3) Design optimal, computationally efficient information fusion algorithms with performance guarantees. These algorithms will consider physically realistic nonlinear dynamics and observations with colored, non-Gaussian noise, commonly encountered in CPS. (4) Synthesize optimal, real-time decision making strategies to balance the potentially conflicting objectives of information gathering and mission fulfillment. This investigation will enable autonomous CPS to navigate complex tradeoffs, leading to autonomous identification and adoption of the optimal strategy.

This research has far-reaching impact- it will evolve autonomous CPS from merely sensing the environment to making sense of the environment, bringing new capabilities in environments where direct human control is not physically or economically possible. The project has a vertically-integrated education plan spanning K-12, undergraduate, and graduate students. The project will engage economically disadvantaged middle and high school students in the same UAV testbed used for research verification. Also, research outcomes will be infused into new and existing undergraduate and graduate courses.

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.

Autonomous driving is on the verge of revolutionizing the transportation system and significantly improving the well-being of people. An autonomous vehicle relies on multiple sensors and AI algorithms to facilitate sensing and perception for navigating the world. As the automotive industry primarily focuses on increasing autonomy levels and enhancing perception performance in mainly benign environments, the security and safety of perception technologies against physical attacks have yet to be thoroughly investigated. Specifically, adversaries creating physical-world perceptual illusions may pose a significant threat to the sensing and learning systems of autonomous vehicles, potentially undermining trust in these systems. This research project aims to deepen our understanding of the security and safety risks under physical attacks. The project endeavors to bolster sensing and learning resilience in autonomous driving against malicious perceptual illusion attacks. The success of the project will significantly advance the security and safety of autonomous driving in the face of emerging physical-world threats, paving the way for the safe deployment of autonomous vehicles in next-generation transportation systems.

The goal of this project is to investigate advanced sensing and learning technologies to enhance the precision and robustness of autonomous driving in intricate and hostile environments. The team?s approach includes: (i) a comprehensive framework to evaluate key vulnerabilities in software/hardware components of autonomous driving systems and devise effective attack vectors for generating false and deceptive perceptions; (ii) a real-time super-resolution radar sensing technology and a data fusion approach that integrates features from various sensor types at both the middle and late stages to effectively bolster the robustness of each sensing modality against illusions; and (iii) a systematic framework to enhance the algorithmic generality and achieve robust perception against multi-modal attacks using multi-view representation learning. The presented solutions will undergo rigorous testing using simulations and experiments to validate their effectiveness and robustness. These solutions contribute to the development of more secure and robust autonomous driving systems, capable of withstanding perceptual illusion attacks in real-world scenarios. The project will also offer research training opportunities for underrepresented students across diverse levels and age groups. The resulting novel technology will be shared as open-source for broader dissemination and advancement of the knowledge developed through this project.
 

This project aims to radically transform traffic management, emergency response, and urban planning practices via predictive analytics on rich data streams from increasingly prevalent instrumented and connected vehicles, infrastructure, and people. Road safety and congestion are a formidable challenge for communities. Current incident management practices are largely reactive in response to road user reports. With the outcome of this project, cities could proactively deploy assets and manage traffic. This would reduce emergency response times, saving lives, and minimizing disruptions to traffic. Efforts are planned in Kindergarten-12 outreach, undergraduate education, outreach to women and minority students, and incorporation of the research into courses, with the goal to inspire and train a diverse cohort for the next-generation of scientists and prepare them for taking on challenges arising from smart and connected communities.

To realize the envisioned system, an integrated research approach is taken to tackle the following closely related research tasks: (1) integration of heterogeneous data streams using a new sparse multi-task multi-view feature fusing method; (2) prediction of traffic incidents by designing a novel high-order low-rank model; (3) teaming of connected vehicles and roadside sensor systems; (4) verification of traffic condition prediction by crowdsourcing the ground truth from user reports in real-time; (5) selection of crowdsourcing participants that recruits and selects voluntary operators of instrumented connected vehicles to provide onboard sensing readings; (6) selection of high quality and diverse images and videos from crowdsourcing vehicles to provide better data for traffic prediction; and 7) design of optimal rerouting strategies to improve commuters' routes in the time of potential traffic disruption.

Cyber-physical systems such as self-driving cars, drones, and intelligent transportation rely heavily on machine learning techniques for ever-increasing levels of autonomy. In the example of autonomous vehicles, deep learning or deep neural networks can be employed for perception, sensor fusion, prediction, planning, and control tasks. However powerful such machine learning techniques have become, they also expose a new attack surface, which may lead to vulnerability to adversarial attacks and potentially harmful consequences in security- and safety-critical scenarios. This project investigates adversarial machine learning challenges faced by autonomous cyber-physical systems with the aim of formulating defense strategies. The project will collaborate with the Center for STEM (Science, Technology, Engineering and Math) Education at Northeastern University and the Office of Access and Inclusion Center at University of California at Irvine to engage undergraduates, women, and minority students in independent research projects.

This project is composed of two interdependent research thrusts, one for investigating adversarial attacks and one for devising countermeasures, aiming to secure the key deep learning-equipped software components of autonomous cyber-physical systems, such as perception, obstacle prediction, and vehicle planning and control. The main deep learning techniques of interest to autonomous cyber-physical systems include convolutional neural networks for detection, recurrent neural networks for prediction, and deep reinforcement learning for control. The technical innovations of the project include ADMM (Alternating Direction Method of Multipliers) based attack generation, concurrent adversarial training and model compression, and multi-sourced defense schemes incorporating adversarial training and ensemble learning. This project will implement and evaluate the proposed attack and defense approaches on real-world prototypes of autonomous cyber-physical systems for autonomous vehicles and unmanned aerial vehicles in the investigators' labs. The investigators will release all the developed models, algorithms, and software to GitHub to facilitate community usage.

Cancer remains the second leading cause of death in the US. Immunotherapy is a cancer treatment that aims to help the body?s immune system fight cancer. While excellent responses have been observed for a large number of patients with varying disease types, a considerably larger number of patients have received little to no benefit from immunotherapy. This varied outcome has been attributed to the highly heterogeneous physical and physiological profile within and around tumors that suppress the immune system?s response. Various physical, chemical, and biological treatment modalities are under investigation for altering the tumor environment from a state where immune effects are suppressed, to one supportive of an anti-tumor immune response. However, these approaches are hampered by the lack of techniques for monitoring the tumor state in response to candidate treatments. Technologies that enable continuous monitoring of the tumor?s immune state, and thereby guide precise delivery of interventions to drive tumors to an immunostimulatory state, offer the promise of unlocking the full potential of immunotherapies. A cyber-physical systems (CPS) perspective is uniquely suited to addressing this challenge, treating the tumor as an ?in body CPS? with the development of sensors and analytical techniques for longitudinal assessment of the tumor, coupled with co-located methods for delivering physical/chemical treatments for modulating the environment within the tumor towards an immunostimulatory state. If successfully developed and translated, the CPS framework for immunomodulation investigated in this project may ultimately guide selection and optimal delivery of priming interventions prior to immunotherapy delivery, determine when priming interventions have successfully modulated the tumor to an immunogenically favorable state, and for assessing treatment response. The investigator team will develop a graduate-level course on biomedical cyber-physical systems along with modules on implantable biomedical sensors for undergraduate courses. Further, this project will provide summer research opportunities for students from under-represented groups via the Pathways to STEM program.

This project will investigate a CPS framework for immunomodulation of the tumor microenvironment (TME), integrating: (1) a unique 3D micro-array sensor and treatment (MIST) device consisting of a sensing/actuation platform for longitudinal sensing and control of physical and physiological parameters within the TME; (2) novel model-informed machine learning techniques for determining tumor immune state from TME physical/physiologic characteristics; and (3) model-guided therapy via the MIST device for driving the TME to an immunostimulatory state. Advanced 3D fabrication technology will provide implantable micromachined multimodal sensing devices to enable longitudinal in vivo sensing of TME parameters such as tissue oxygenation, pH, pressure, and metabolism, and co-located treatment on a single device. Data gathered from implantable sensors will be fused with computational models of biophysical parameters informed by tumor-specific vasculature maps using a graph neural tensor completion approach. The novel hybrid machine learning approach for data imputation and fusion will systematically incorporate uncertainties and provide the basis to infer the immune state of a tumor, validated against gold-standard molecular biomarkers of immune state in experimental small animals. A graph-based clustering approach integrated with a recurrent neural network will be used for the prediction of tumor state changes. Finally, we will evaluate the efficacy of model-guided delivery of energy-based interventions to transform the TME to a pro-immunogenic state and the impact of these interventions on immunotherapy outcomes in small animals.

This project is jointly funded by the Cyber-Physical Program and the Established Program to Stimulate Competitive Research (EPSCoR).