YOLOv8 vs. Sensor-Based Traffic Control: Kalman Filter Integration for Hybrid Management

Abstract

Modern traffic management systems of an urban environment presuppose adaptive systems of response to changing conditions of pedestrian and vehicle presence. Research is a comparative analysis of sensor-based and YOLOv8 traffic control systems, where the detection response time, the spatial gap analysis, environmental sensitivity, and accuracy are evaluated. Experimental validation indicates unique operational benefits since sensor-based systems are 92% accurate in high pedestrian environments, whilst the YOLOv8 is 84% effective in vehicle recognition. A novel hybrid architecture is proposed, implementing strategic technology assignment where sensors manage pedestrian monitoring and YOLOv8 handles vehicle detection, integrated through Kalman filter decision engines. Sophisticated 15-day simulation with Poisson distribution modeling shows 35% reduction in mean delay time, 84% accuracy of vehicle identification, and 78% accuracy of pedestrian identification in crowded scenarios. Kalman filter integration reaches 78% diminution of noise and enables predictive traffic control with 91% accuracy for a 10 minute horizon. The hybrid model overcomes the specific system constraints with a maximization of complementary benefits to intelligent urban traffic management.