A Conceptual Deep Learning Model for Real-Time Routing

Urban congestion has been a known problem since the first urban revolution throughout the world. Today's major metropolises are synonymous with traffic congestion and complicated urban circulation. This paper introduces a traffic management system based on agent technology and deep learning model. It acts on the phase layouts constituted of sequences and length to reduce average delay time, maximize intersection throughput, and suggest a routing planning scheme to avoid congested intersection. In the optimization model, lights of each intersection are controlled by a group of agents. The system uptake the ability of the agent to communicate, coordinate and collaborate, to decompose the routing process into decentralized architecture. The proposed model was instantiated in the ANYLOGIC simulator.