Density-Based Classification Network Control Approach for a Single Stoplight Traffic System on a Simplified Traffic Simulation using MATLAB

A more dynamic method of improving traffic flow on a road network is made possible by intelligent traffic management systems (ITMS). This is done by determining the optimal routing configuration of a stoplight system to maximize vehicle throughput. One such algorithm is the classification network which uses a deep neural network to approximate a function that determines the optimal routing state by projecting it as a classification task. This paper uses a classification network-based control algorithm for a single smart traffic light module implemented through the MATLAB software. The proposed system was able to route vehicles 15.1% faster than a static system on a single intersection road network. The classification network control scheme was able to determine the best routing configuration for the stoplight system in real time to optimize traffic flow. Thus, the system is a feasible improvement in stoplight control as it can reduce traffic congestion by increasing vehicle throughput in the system.