Project Summary
As California accelerates the deployment of Connected Automated Vehicles (CAVs) and Intelligent Traffic Management Systems (ITMS), ensuring pedestrian safety at complex intersections becomes increasingly urgent. This project addresses a growing concern: The vulnerability of ITMS perception systems, particularly camera and LiDAR sensors, to physical adversarial attacks such as laser interference, acoustic spoofing, and electromagnetic disruptions. These stealthy perturbations can lead to dangerous detection failures, disproportionately affecting vulnerable pedestrians including children, elderly individuals, and transit-dependent populations. We propose an applied research initiative to systematically evaluate these threats in both lab and real-world settings, with a focus on California’s evolving transportation infrastructure. The project includes (1) characterizing ITMS vulnerabilities under coordinated physical attacks, (2) assessing the robustness of current sensor fusion strategies, and (3) developing novel, multi-layered defenses that combine robust ML models with sensor-level protections. The resulting end-to-end evaluation pipeline will serve as a critical tool for public agencies such as DMV, CPUC, and local DOTs, enabling data-driven risk assessments and safety certification processes. Our findings will guide the safe integration of ITMS and CAVs, reduce pedestrian fatalities, and promote equitable, future-ready urban mobility in line with UC ITS priorities.
