This research develops innovative methods for improving intersection safety through three integrated approaches: optimized intersection planning, connected and automated vehicle (CAV) integration, and AI-driven pedestrian safety enhancement. The project addresses the safe accommodation of diverse users including private vehicles, trucks, transit, and pedestrians while incorporating emerging technologies such as sensors, control systems, and CAVs. Using population- based metaheuristic algorithms and VISSIM microsimulation modeling, the research will optimize intersection development through cost minimization that includes construction, maintenance, user costs, delays, accidents, and emissions. The CAV integration component focuses on infrastructure readiness for varying levels of vehicle autonomy through simulation and analysis models, cooperative perception systems, and Vehicle-to-Everything communication. The pedestrian safety advancement leverages multi- modal RGBT sensor data, SAM2 AI tracking models, LiDAR integration, and surrogate safety measures to create predictive safety systems. The methodology builds on extensive University of Maryland experience in transportation network optimization and incorporates real-world sensor deployments at Maryland sites with over 25 hours of interaction data collection and analysis.
Universities Involved
University of Maryland
Principal Investigators
Cinzia Cirillo,
Paul Schonfeld,
Terry Yang
Expected Research Outcomes & Impacts
The application of this research will transform intersection safety management by enabling transportation agencies to design and retrofit infrastructure that accommodates diverse vehicle types, user needs, and emerging technologies. State agencies and local authorities will gain enhanced capabilities to address aging infrastructure challenges while managing implementation costs and reducing pedestrian-involved traffic accidents. The optimized planning frameworks will enable agencies to make data-driven investment decisions that minimize total lifecycle costs while maximizing safety and efficiency improvements. CAV integration methodologies will prepare transportation systems for future vehicle technologies, reducing delays, conflicts, and crash risks through improved vehicle-to- infrastructure communication and adaptive signal control. The AI-driven pedestrian safety systems will significantly enhance detection capabilities in challenging conditions, particularly low-light environments, raising the standard of data quality and traffic safety evaluations nationwide.
Transportation professionals will benefit from scalable, cost-effective retrofitting strategies that allow phased implementation aligned with budget constraints and technology adoption timelines. Long-term impacts include reduced intersection conflicts, improved multimodal safety, enhanced traffic flow efficiency, and strengthened infrastructure readiness for autonomous vehicle deployment. The modular upgrade approaches will enable agencies to efficiently allocate resources while maintaining service quality during infrastructure transitions, ultimately creating safer, more efficient intersections that serve all users effectively.
Subject Areas
Traffic Management, Intelligent Transportation Systems, Artificial Intelligence