{"id":{"repo_id":"umn","oai_identifier":"oai:conservancy.umn.edu:11299/276790"},"canonical_url":"https://search.dev.ndltd.org/etd/umn/oai:conservancy.umn.edu:11299/276790","repository":{"repo_id":"umn","name":"University of Minnesota","base_url":"https://conservancy.umn.edu/server/oai/request"},"display":{"title":"Enabling safe, smart, and sustainable Transportation Cyber-Physical Systems with Artificial Intelligence","abstract":"The transportation engineering landscape is undergoing a pivotal transformation. Although transportation engineering research roots lie in the traditional engineering principles underpinning road, bridge, and infrastructure design, the past decade has seen a merger of computational and data-driven paradigms. Consequently, the forerunners of this sector are not just traditional civil engineering entities anymore. Technology-centric companies like Uber, Lyft, Waymo, and Tesla are now leading figures, reshaping the traditional role of civil engineering entities in transportation. Ongoing urbanization and densification of cities highlight the critical need for evolved traffic operation strategies to better manage, control, and optimize transportation systems. This evolution calls for high-level and efficient traffic flow prediction models, essential in creating resilient urban environments. Concurrently, the rise in vehicle automation—from the basic SAE Level 1 features like adaptive cruise control (ACC) to the advanced SAE Level 5 fully automated vehicles devoid of human intervention—is reshaping the transportation sphere. However, challenges persist. While traditional traffic flow models frequently falter in mimicking the intricate realities of our roads, purely data-driven methods might overlook vital domain-specific nuances of traffic engineering. This dissertation bridges this gap by harnessing the prowess of Artificial Intelligence (AI) to elucidate the intricacies of Transportation-Cyber-Physical Systems (T-CPS) across multiple scales. This research combines domain expertise with AI-driven insights to facilitate the modeling and control of evolving transportation systems, where humans continue to play an essential role. The central theme is to explore the potential of machine learning, informed by transportation engineering priors, to enhance our understanding and management of T-CPS. The first two chapters focus on microscopic traffic phenomena, specifically driving behavior. The first chapter, \"Modeling and Learning Driving Behavior Part I: Merging Physical Constraints with AI in Car-following Models,\" introduces RACER, the Rational Artificial Intelligence Car-following model Enhanced by Reality. This deep learning car-following model integrates Rational Driving Constraints to predict ACC driving behavior with high accuracy and realism. RACER's ability to satisfy partial derivative constraints sets it apart from conventional models, demonstrating superior performance in metrics like acceleration, velocity, and spacing. This highlights the value of incorporating physical constraints within AI models, particularly for enhancing safety in transportation. The second chapter, ``Modeling and Learning Driving Behavior Part II: Car-following-response Based Vehicle Classification,\" builds on this foundation by exploring the classification and identification of driving behavior in traffic flow. Here, a time-series-based deep learning classification method is proposed to rapidly assess car-following dynamics and identify human-driven and ACC vehicles in real-time using only car-following trajectory data. This method's high accuracy and speed outperform existing state-of-the-art techniques, marking an advancement in real-time traffic control strategies tailored to individual driving characteristics. Shifting the focus to macroscopic phenomena, the third chapter, ``Data-Driven Approaches in Macroscopic Modern Transportation Analysis Part I: Mobility,\" addresses the optimization of taxi utilization rates amid the disruptions caused by transportation network companies. Leveraging LSTM architecture for dynamic demand prediction, a real-time taxi trip optimization strategy is developed. Testing this strategy in normal and COVID-19 scenarios reveals improvements in mileage utilization rates, demonstrating the effectiveness of the proposed approach and highlighting the broader impacts of pandemics on urban mobility. The fourth chapter, \"Data-Driven Approaches in Macroscopic Modern Transportation Analysis Part II: Safety,\" investigates driver-pedestrian interactions at unsignalized intersections. Transportation is not only about theory but is closely related to our daily lives. Large-scale data collection and experiments play a crucial role in transportation research. This study introduces an extensive open-source dataset compiled from video data at 18 intersections, documenting over 3000 interactions and more than 50 contextual variables. Using logistic regression, a classification model is developed to predict driver yielding behavior based on identified variables. This comprehensive dataset and analysis provide valuable insights for traffic safety improvements and intersection design, contributing to safer and more sustainable transportation systems. Finally, the fifth chapter, \"Cybersecurity in the Era of Intelligent Transportation Systems,\" explores the cybersecurity risks associated with advanced driver-assistance systems and automated vehicles. It presents a traffic modeling framework for potential cyberattacks and their impacts on vehicle dynamics and traffic flow. An anomaly detection model based on generative adversarial networks is introduced for real-time detection of such attacks, outperforming contemporary neural network models in identifying irregular driving patterns in ACC vehicles. Given the extensive work on driving behavior, it is essential to consider future scenarios and potential cybersecurity challenges. In conclusion, this dissertation represents a contribution to the field of transportation engineering by offering innovative solutions to contemporary challenges through the integration of AI and domain-specific knowledge. The common theme between each of these chapters is the utilization of rich datasets and advanced technologies to enhance and optimize transportation systems. The findings and methodologies presented have the potential to guide future research and practical applications, fostering the development of safer, more efficient, and sustainable transportation systems.","abstract_html":"The transportation engineering landscape is undergoing a pivotal transformation. Although transportation engineering research roots lie in the traditional engineering principles underpinning road, bridge, and infrastructure design, the past decade has seen a merger of computational and data-driven paradigms. Consequently, the forerunners of this sector are not just traditional civil engineering entities anymore. Technology-centric companies like Uber, Lyft, Waymo, and Tesla are now leading figures, reshaping the traditional role of civil engineering entities in transportation. Ongoing urbanization and densification of cities highlight the critical need for evolved traffic operation strategies to better manage, control, and optimize transportation systems. This evolution calls for high-level and efficient traffic flow prediction models, essential in creating resilient urban environments. Concurrently, the rise in vehicle automation—from the basic SAE Level 1 features like adaptive cruise control (ACC) to the advanced SAE Level 5 fully automated vehicles devoid of human intervention—is reshaping the transportation sphere. However, challenges persist. While traditional traffic flow models frequently falter in mimicking the intricate realities of our roads, purely data-driven methods might overlook vital domain-specific nuances of traffic engineering. This dissertation bridges this gap by harnessing the prowess of Artificial Intelligence (AI) to elucidate the intricacies of Transportation-Cyber-Physical Systems (T-CPS) across multiple scales. This research combines domain expertise with AI-driven insights to facilitate the modeling and control of evolving transportation systems, where humans continue to play an essential role. The central theme is to explore the potential of machine learning, informed by transportation engineering priors, to enhance our understanding and management of T-CPS. The first two chapters focus on microscopic traffic phenomena, specifically driving behavior. The first chapter, "Modeling and Learning Driving Behavior Part I: Merging Physical Constraints with AI in Car-following Models," introduces RACER, the Rational Artificial Intelligence Car-following model Enhanced by Reality. This deep learning car-following model integrates Rational Driving Constraints to predict ACC driving behavior with high accuracy and realism. RACER's ability to satisfy partial derivative constraints sets it apart from conventional models, demonstrating superior performance in metrics like acceleration, velocity, and spacing. This highlights the value of incorporating physical constraints within AI models, particularly for enhancing safety in transportation. The second chapter, ``Modeling and Learning Driving Behavior Part II: Car-following-response Based Vehicle Classification," builds on this foundation by exploring the classification and identification of driving behavior in traffic flow. Here, a time-series-based deep learning classification method is proposed to rapidly assess car-following dynamics and identify human-driven and ACC vehicles in real-time using only car-following trajectory data. This method's high accuracy and speed outperform existing state-of-the-art techniques, marking an advancement in real-time traffic control strategies tailored to individual driving characteristics. Shifting the focus to macroscopic phenomena, the third chapter, ``Data-Driven Approaches in Macroscopic Modern Transportation Analysis Part I: Mobility," addresses the optimization of taxi utilization rates amid the disruptions caused by transportation network companies. Leveraging LSTM architecture for dynamic demand prediction, a real-time taxi trip optimization strategy is developed. Testing this strategy in normal and COVID-19 scenarios reveals improvements in mileage utilization rates, demonstrating the effectiveness of the proposed approach and highlighting the broader impacts of pandemics on urban mobility. The fourth chapter, "Data-Driven Approaches in Macroscopic Modern Transportation Analysis Part II: Safety," investigates driver-pedestrian interactions at unsignalized intersections. Transportation is not only about theory but is closely related to our daily lives. Large-scale data collection and experiments play a crucial role in transportation research. This study introduces an extensive open-source dataset compiled from video data at 18 intersections, documenting over 3000 interactions and more than 50 contextual variables. Using logistic regression, a classification model is developed to predict driver yielding behavior based on identified variables. This comprehensive dataset and analysis provide valuable insights for traffic safety improvements and intersection design, contributing to safer and more sustainable transportation systems. Finally, the fifth chapter, "Cybersecurity in the Era of Intelligent Transportation Systems," explores the cybersecurity risks associated with advanced driver-assistance systems and automated vehicles. It presents a traffic modeling framework for potential cyberattacks and their impacts on vehicle dynamics and traffic flow. An anomaly detection model based on generative adversarial networks is introduced for real-time detection of such attacks, outperforming contemporary neural network models in identifying irregular driving patterns in ACC vehicles. Given the extensive work on driving behavior, it is essential to consider future scenarios and potential cybersecurity challenges. In conclusion, this dissertation represents a contribution to the field of transportation engineering by offering innovative solutions to contemporary challenges through the integration of AI and domain-specific knowledge. The common theme between each of these chapters is the utilization of rich datasets and advanced technologies to enhance and optimize transportation systems. The findings and methodologies presented have the potential to guide future research and practical applications, fostering the development of safer, more efficient, and sustainable transportation systems.","abstract_has_math":false,"creators":["Li, Tianyi"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2024,"date_issued":"2024-07","date_published":"2024-07","updated_at":"2026-07-24T05:19:52Z","subjects":["AI","Automated vehicle","Intelligent transportation systems","Traffic","Transportation engineering","Urban mobility"],"languages":["en"],"rights":[],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"https://hdl.handle.net/11299/276790","outbound_label":"Handle","outbound_source":"dc:identifier.uri"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:creator","label":"Author","values":["Li, Tianyi"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2025-10-24T11:34:07Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2025-10-24T11:34:07Z"]},{"key":"dc:date.issued","label":"Date","values":["2024-07"]},{"key":"dc:type","label":"Dc Type","values":["Thesis or Dissertation"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["AI","Automated vehicle","Intelligent transportation systems","Traffic","Transportation engineering","Urban mobility"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/11299/276790"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["University of Minnesota Ph.D. dissertation. July 2024. Major: Civil Engineering. Advisor: Raphae; Stern. 1 computer file (PDF); xx, 202 pages."]},{"key":"dc:description.abstract","label":"Abstract","values":["The transportation engineering landscape is undergoing a pivotal transformation. Although transportation engineering research roots lie in the traditional engineering principles underpinning road, bridge, and infrastructure design, the past decade has seen a merger of computational and data-driven paradigms. Consequently, the forerunners of this sector are not just traditional civil engineering entities anymore. Technology-centric companies like Uber, Lyft, Waymo, and Tesla are now leading figures, reshaping the traditional role of civil engineering entities in transportation. Ongoing urbanization and densification of cities highlight the critical need for evolved traffic operation strategies to better manage, control, and optimize transportation systems. This evolution calls for high-level and efficient traffic flow prediction models, essential in creating resilient urban environments. Concurrently, the rise in vehicle automation—from the basic SAE Level 1 features like adaptive cruise control (ACC) to the advanced SAE Level 5 fully automated vehicles devoid of human intervention—is reshaping the transportation sphere. However, challenges persist. While traditional traffic flow models frequently falter in mimicking the intricate realities of our roads, purely data-driven methods might overlook vital domain-specific nuances of traffic engineering. This dissertation bridges this gap by harnessing the prowess of Artificial Intelligence (AI) to elucidate the intricacies of Transportation-Cyber-Physical Systems (T-CPS) across multiple scales. This research combines domain expertise with AI-driven insights to facilitate the modeling and control of evolving transportation systems, where humans continue to play an essential role. The central theme is to explore the potential of machine learning, informed by transportation engineering priors, to enhance our understanding and management of T-CPS. The first two chapters focus on microscopic traffic phenomena, specifically driving behavior. The first chapter, \"Modeling and Learning Driving Behavior Part I: Merging Physical Constraints with AI in Car-following Models,\" introduces RACER, the Rational Artificial Intelligence Car-following model Enhanced by Reality. This deep learning car-following model integrates Rational Driving Constraints to predict ACC driving behavior with high accuracy and realism. RACER's ability to satisfy partial derivative constraints sets it apart from conventional models, demonstrating superior performance in metrics like acceleration, velocity, and spacing. This highlights the value of incorporating physical constraints within AI models, particularly for enhancing safety in transportation. The second chapter, ``Modeling and Learning Driving Behavior Part II: Car-following-response Based Vehicle Classification,\" builds on this foundation by exploring the classification and identification of driving behavior in traffic flow. Here, a time-series-based deep learning classification method is proposed to rapidly assess car-following dynamics and identify human-driven and ACC vehicles in real-time using only car-following trajectory data. This method's high accuracy and speed outperform existing state-of-the-art techniques, marking an advancement in real-time traffic control strategies tailored to individual driving characteristics. Shifting the focus to macroscopic phenomena, the third chapter, ``Data-Driven Approaches in Macroscopic Modern Transportation Analysis Part I: Mobility,\" addresses the optimization of taxi utilization rates amid the disruptions caused by transportation network companies. Leveraging LSTM architecture for dynamic demand prediction, a real-time taxi trip optimization strategy is developed. Testing this strategy in normal and COVID-19 scenarios reveals improvements in mileage utilization rates, demonstrating the effectiveness of the proposed approach and highlighting the broader impacts of pandemics on urban mobility. The fourth chapter, \"Data-Driven Approaches in Macroscopic Modern Transportation Analysis Part II: Safety,\" investigates driver-pedestrian interactions at unsignalized intersections. Transportation is not only about theory but is closely related to our daily lives. Large-scale data collection and experiments play a crucial role in transportation research. This study introduces an extensive open-source dataset compiled from video data at 18 intersections, documenting over 3000 interactions and more than 50 contextual variables. Using logistic regression, a classification model is developed to predict driver yielding behavior based on identified variables. This comprehensive dataset and analysis provide valuable insights for traffic safety improvements and intersection design, contributing to safer and more sustainable transportation systems. Finally, the fifth chapter, \"Cybersecurity in the Era of Intelligent Transportation Systems,\" explores the cybersecurity risks associated with advanced driver-assistance systems and automated vehicles. It presents a traffic modeling framework for potential cyberattacks and their impacts on vehicle dynamics and traffic flow. An anomaly detection model based on generative adversarial networks is introduced for real-time detection of such attacks, outperforming contemporary neural network models in identifying irregular driving patterns in ACC vehicles. Given the extensive work on driving behavior, it is essential to consider future scenarios and potential cybersecurity challenges. In conclusion, this dissertation represents a contribution to the field of transportation engineering by offering innovative solutions to contemporary challenges through the integration of AI and domain-specific knowledge. The common theme between each of these chapters is the utilization of rich datasets and advanced technologies to enhance and optimize transportation systems. The findings and methodologies presented have the potential to guide future research and practical applications, fostering the development of safer, more efficient, and sustainable transportation systems."]},{"key":"dc:title","label":"Title","values":["Enabling safe, smart, and sustainable Transportation Cyber-Physical Systems with Artificial Intelligence"]}]}],"canonical_facts":{"dc:creator":["Li, Tianyi"],"dc:date.accessioned":["2025-10-24T11:34:07Z"],"dc:date.available":["2025-10-24T11:34:07Z"],"dc:date.issued":["2024-07"],"dc:description":["University of Minnesota Ph.D. dissertation. July 2024. Major: Civil Engineering. Advisor: Raphae; Stern. 1 computer file (PDF); xx, 202 pages."],"dc:description.abstract":["The transportation engineering landscape is undergoing a pivotal transformation. Although transportation engineering research roots lie in the traditional engineering principles underpinning road, bridge, and infrastructure design, the past decade has seen a merger of computational and data-driven paradigms. Consequently, the forerunners of this sector are not just traditional civil engineering entities anymore. Technology-centric companies like Uber, Lyft, Waymo, and Tesla are now leading figures, reshaping the traditional role of civil engineering entities in transportation. Ongoing urbanization and densification of cities highlight the critical need for evolved traffic operation strategies to better manage, control, and optimize transportation systems. This evolution calls for high-level and efficient traffic flow prediction models, essential in creating resilient urban environments. Concurrently, the rise in vehicle automation—from the basic SAE Level 1 features like adaptive cruise control (ACC) to the advanced SAE Level 5 fully automated vehicles devoid of human intervention—is reshaping the transportation sphere. However, challenges persist. While traditional traffic flow models frequently falter in mimicking the intricate realities of our roads, purely data-driven methods might overlook vital domain-specific nuances of traffic engineering. This dissertation bridges this gap by harnessing the prowess of Artificial Intelligence (AI) to elucidate the intricacies of Transportation-Cyber-Physical Systems (T-CPS) across multiple scales. This research combines domain expertise with AI-driven insights to facilitate the modeling and control of evolving transportation systems, where humans continue to play an essential role. The central theme is to explore the potential of machine learning, informed by transportation engineering priors, to enhance our understanding and management of T-CPS. The first two chapters focus on microscopic traffic phenomena, specifically driving behavior. The first chapter, \"Modeling and Learning Driving Behavior Part I: Merging Physical Constraints with AI in Car-following Models,\" introduces RACER, the Rational Artificial Intelligence Car-following model Enhanced by Reality. This deep learning car-following model integrates Rational Driving Constraints to predict ACC driving behavior with high accuracy and realism. RACER's ability to satisfy partial derivative constraints sets it apart from conventional models, demonstrating superior performance in metrics like acceleration, velocity, and spacing. This highlights the value of incorporating physical constraints within AI models, particularly for enhancing safety in transportation. The second chapter, ``Modeling and Learning Driving Behavior Part II: Car-following-response Based Vehicle Classification,\" builds on this foundation by exploring the classification and identification of driving behavior in traffic flow. Here, a time-series-based deep learning classification method is proposed to rapidly assess car-following dynamics and identify human-driven and ACC vehicles in real-time using only car-following trajectory data. This method's high accuracy and speed outperform existing state-of-the-art techniques, marking an advancement in real-time traffic control strategies tailored to individual driving characteristics. Shifting the focus to macroscopic phenomena, the third chapter, ``Data-Driven Approaches in Macroscopic Modern Transportation Analysis Part I: Mobility,\" addresses the optimization of taxi utilization rates amid the disruptions caused by transportation network companies. Leveraging LSTM architecture for dynamic demand prediction, a real-time taxi trip optimization strategy is developed. Testing this strategy in normal and COVID-19 scenarios reveals improvements in mileage utilization rates, demonstrating the effectiveness of the proposed approach and highlighting the broader impacts of pandemics on urban mobility. The fourth chapter, \"Data-Driven Approaches in Macroscopic Modern Transportation Analysis Part II: Safety,\" investigates driver-pedestrian interactions at unsignalized intersections. Transportation is not only about theory but is closely related to our daily lives. Large-scale data collection and experiments play a crucial role in transportation research. This study introduces an extensive open-source dataset compiled from video data at 18 intersections, documenting over 3000 interactions and more than 50 contextual variables. Using logistic regression, a classification model is developed to predict driver yielding behavior based on identified variables. This comprehensive dataset and analysis provide valuable insights for traffic safety improvements and intersection design, contributing to safer and more sustainable transportation systems. Finally, the fifth chapter, \"Cybersecurity in the Era of Intelligent Transportation Systems,\" explores the cybersecurity risks associated with advanced driver-assistance systems and automated vehicles. It presents a traffic modeling framework for potential cyberattacks and their impacts on vehicle dynamics and traffic flow. An anomaly detection model based on generative adversarial networks is introduced for real-time detection of such attacks, outperforming contemporary neural network models in identifying irregular driving patterns in ACC vehicles. Given the extensive work on driving behavior, it is essential to consider future scenarios and potential cybersecurity challenges. In conclusion, this dissertation represents a contribution to the field of transportation engineering by offering innovative solutions to contemporary challenges through the integration of AI and domain-specific knowledge. The common theme between each of these chapters is the utilization of rich datasets and advanced technologies to enhance and optimize transportation systems. The findings and methodologies presented have the potential to guide future research and practical applications, fostering the development of safer, more efficient, and sustainable transportation systems."],"dc:identifier.uri":["https://hdl.handle.net/11299/276790"],"dc:language.iso":["en"],"dc:subject":["AI","Automated vehicle","Intelligent transportation systems","Traffic","Transportation engineering","Urban mobility"],"dc:title":["Enabling safe, smart, and sustainable Transportation Cyber-Physical Systems with Artificial Intelligence"],"dc:type":["Thesis or Dissertation"]},"updated_at":"2026-07-24T05:19:52Z"}