Coordinated real-time control algorithm for multi-crossing traffic lights

To coordinate the traffic lights for adjacent crossings makes profound challenges to the intelligent control of traffics in modern cities. In particular, the complexity of the model for the multi-crossing traffic is difficult to solve in a time scale acceptable for efficient traffic control. This study proposed to solve the optimization problem of the multi-crossing traffic model by co-evolving particle swarm optimization, and to learn the optimal strategy for multi-crossing traffic control by an artificial neural network (ANN). With this well trained ANN, the realtime control over the multi-crossing traffic lights can be achieved. The systematic simulation results showed us that the proposed approach is a promising control algorithm for the multi-crossing traffic lights.