Industrial Expert System for Intelligent Traffic Lane Allocation Using Machine Learning and Pattern Recognition

Abstract

This paper highlights how existing infrastructure can be used to feed data to an expert system for smart vehicle traffic management. The use of existing video cameras in road intersections for traffic pattern analysis is often overlooked due to their perceived limitations and varying technical specifications. We compare the advantages and disadvantages of both a system architecture that incorporates a Convolutional Neural Network and one that makes use of a Logistic Regression algorithm for traffic lane occupancy detection and optimizes the allocation of additional lanes based on occupancy data, thereby improving vehicle traffic flow, additionally, traffic patterns are viewed analyzed from point of origin to point of dissipation, the system allocating lanes accordingly on defined segments not just in individual intersections. Using existing video camera infrastructure for data collection offers a cost-effective approach that enhances traffic safety, enables emergency corridors, and allows flexibility in lane allocation without requiring a complete replacement of costly infrastructure already in use in most major cities around the world.