Edge Server-Based AI Application for Dilemma Zone and Traffic Conflict Events Detection

While AI-based techniques have been broadly embraced by transportation researchers and professionals due to their huge potential to significantly improve the efficiency and accuracy of analyzing transportation data, they typically require a large dataset to train the models effectively and experience difficulties in supporting real-time transportation operations and management. This project will use AI-based video analytics techniques to continuously track road users for real-time dilemma zone quantification and traffic conflict analysis. The objectives of this research are to (1) develop an AI-based video analytics application for real-time dilemma zone events detection; (2) develop an AI-based video analytics application for real-time detection of traffic conflicts among road users based on trajectory analysis; (3) design the system of implementing the AI-based applications on the edge server with ultra-low latency 5G communication; and (4) demonstration of the proof-of-concept with real-time data.