An AI-Driven Smart Street Lighting with Object Recognition for Energy Optimization

Energy-efficient street lighting is a critical component of sustainable smart city infrastructure, particularly in regions with extensive nighttime road networks. This study presents a data-driven evaluation of occupancy-aware smart street lighting using a large-scale, real-world nighttime traffic dataset collected from multiple road environments in Abu Dhabi, including urban, suburban, intercity, and rural areas. Instead of relying on physical system deployment or prototype-based field testing, the analysis leverages aggregated traffic occupancy statistics derived from real traffic monitoring systems to assess the potential impact of adaptive lighting control strategies. Nighttime road occupancy rates were analyzed across different road categories and time intervals, revealing pronounced temporal and spatial variations in traffic demand. The results indicate that average nighttime occupancy levels during off-peak hours typically remain below 20–30% for several road types, with late-night periods exhibiting even lower utilization. Based on these occupancy patterns, estimated illumination dimming levels correspond to potential energy savings in the range of 30–60%, depending on road category and time of operation, while maintaining higher lighting levels during periods of increased traffic to satisfy road safety requirements. Overall, the findings confirm that leveraging real-world traffic statistics provides a robust and scalable foundation for designing intelligent, occupancy-aware street lighting policies in smart city environments.