Timing Optimization and Control for Smart Traffic

Currently, traffic signals in most countries have preset time periods for the red/yellow/green signals. Different roads in a city are configured with different time periods based on type and location of roads and the peak or off-peak hours. Such preset timings are inadequate in coping with unexpected situations such as a traffic accident or other natural disasters or specific unplanned events. Because the traffic control system is generally not adaptive enough, traffic congestion occurs often in most metropolitan cities. As a solution, we propose a smart traffic optimization system that can adjust the setting of traffic signals in real time based on sensor and camera inputs. In this work, we developed both a microscopic simulator with heuristic optimization, as well as, an embedded system prototype for traffic optimization. A genetic algorithm based optimization method is employed. The traffic simulator allows the user to collect statistical information on roads and also to check if an optimization decision (change in traffic signal timings) works or not. The embedded system prototype helps to realize and verify the actual deployment of the optimization decision. The integration of the simulator and the prototype helped to verify the feasibility of the proposed smart traffic optimization system. Experiment results validate that both the intersection average waiting rate and the intersection average waiting time decreased rapidly after optimization.