Mdp Agent-based Traffic Lights Control Method

Traffic is a bottleneck of urban development. It plays an important role in the development of urban economy and the improvement of the situation of people's life. As the direction of 21th century transportation system, Intelligent Traffic System (ITS) is widely researched by the specialists and scholars of many countries in the world, and it is the essential way of eliminating congestion and improving the utilization of roads.With the fast development of Intelligent Traffic System, there are a lot of breakthroughs in the study of traffic light control. The current problem is that the traffic light control system is usually time-fixed and static, the lights change by presetted period, but can't execute real-time control according to real situation, so non effective waiting time is increased and road resources are wasted. Traditionally, a centralized method was used to address this problem. However, it may be inefficient, time-consuming, costly, and not easily extended. This paper proposes a multi-agent cooperation method to solve the traffic light control problem in distributed, and put the problem into the architecture of optimal constraint satisfaction problem (OCSP) to make a better performance. In this framework, each agent is responsible for a part of the solution, through the communication and corporation of agents in the system, the total solution will finally be reached. Using this method, we may avoid the shortcomings mentioned above in the traditional ways. Thus, we can achieve the intelligent real time control, extension dynamically and well stability of the whole system. At the end of this paper, we build a virtual traffic grid to prove the performance of this MDP-Based Multi-agent Traffic Light Control.The multi-agent cooperation method can improve the traffic capacity of intersection and reduce the delay time of vehicles in comparison with traditional methods. The algorithms above go through system emulation and testing, and get good results. They're feasible for engineering application.