As automobiles become more connected and intelligent, ensuring their cybersecurity is essential to maintaining safety and trust in future transportation systems. Self-driving cars are expected to play a significant role in future transportation. However, their advanced features—such as sensors, cameras, wireless networks, computer vision and machine learning—create new ways for attackers to interfere with vehicle operations. This project aims to protect autonomous vehicles by developing methods that secure both high-level software that makes driving decisions and low-level hardware that controls physical actions like steering and braking. The intellectual merit of the project lies in addressing critical gaps in understanding security vulnerabilities resulting from the complex interactions between self-driving software and vehicle control systems. The project will generate new insights into how information flows between these levels, particularly in adversarial environments, and investigate defense mechanisms against cyber-physical threats. The broader impacts of the project include improving the security of self-driving cars and, with them, the safety of shared civil transportation infrastructure. The outcomes will extend to other cyber-physical systems that integrate computer vision, automation, and real-time networks, such as drones, smart agriculture, and manufacturing systems. The project will also create educational programs to train future cybersecurity professionals at both undergraduate and graduate levels, building a skilled workforce to address cybersecurity challenges in intelligent systems. This project advances the understanding of security-relevant interactions between self-driving autonomy functions—such as perception systems and machine learning algorithms—and the vehicular control systems’ automaticity functions, including electronic control units and in-vehicle networks. These functions operate within sensor-actuator control loops, exchanging cyber-physical state information that introduces opportunities for emerging attacks with potentially catastrophic consequences. This research investigates the nuanced interactions across these interfaces, particularly in adversarial environments, and investigates hybrid attacks targeting sensor-to-controller and controller-to-actuator channels to compromise vehicle operation. The project explores real-time security mechanisms, graceful degradation of system performance under attack, intrusion tolerance, and safe recovery of compromised systems. By addressing these integration points, the project aims to enhance the understanding of cyber-physical threats in automotive systems and contribute to broader advancements in detecting, mitigating, and recovering from attacks. 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.