A Narrative Review: V2X-based Artificial Intelligence for Vulnerable Road User Protection of Urban Autonomous Vehicles

Abstract

Since vulnerable road user protection is a critical challenge for vehicles operating in urban areas, Artificial intelligence (AI) and vehicle-to-everything (V2X) communication is becoming one of core enabling technologies in the development of intelligent transportation systems. In automated vehicle control, traditional onboard sensing is generally limited by occlusion, sensing range, and the complexity of dynamic traffic environments. V2X communication can provide additional information from surrounding ve hicles, roadside infrastructure, and traffic systems, which creates new opportunities for improving vehicle perception, decision- making, and motion control. However, existing reviews rarely focus on the systematic integration of V2X communication and AI algorithms for VRU protection in urban autonomous driving scenarios, especially the coordination mechanism between multi-source data fusion and trajectory prediction. This review focuses on the application of V2X-based artificial intelligence in automated vehicle control in the awareness of road user movement. It summarises the key technologies of V2X communication, cooperative perception, multi-source data fusion, vulnerable road user trajectory prediction, collision risk assessment, and decision-making control. The main objective is to investigate how V2X information can enhance environmental awareness and support safer and more efficient control strategies for automated vehicles.