Developing an Agent-based Simulator Model to Improve Cities' Traffic Flow

Nowadays, cities’ population has been encountered with rapid growth. However, the urban transport infrastructures are not developed with this rate. This issue yield to the common problem which is the traffic jam in streets. The practical approach to solve this problem is the efficient use of the infrastructure and proper traffic control which plays an important role in this performance. Based on the aforementioned issues, the aim of this study is an urban traffic control system using the agent-based simulation approach. In this study, the microscopic model for efficient network traffic junctions in the context of the flow of traffic at intersections has been used. The proposed approach, by defining the main parameters of traffic as a collection of micro-scale smart agents, is expected to address the issue mentioned more effectively. These agents consist of three categories (vehicles, traffic control centers and traffic lights) with different functions. In our approach in addition to the defined scenarios for intersection with a focus on optimizing time, total optimization of traffic flow will also be followed. Also, path finding is used to test the performance of the model. Having the traffic lights with three phase and two methods of path finding, six scenarios are defined and are implemented in a simulated environment. The results of the comparison parameter stop time and the average speed of vehicles reveals the fact that using the intelligent path finding and smart traffic light (sixth scenario) would lead to the downturn about 3.12 seconds per kilometer for stop time and upturn near 1.83 km/h for average speed of cars using the proposed method. We used the controlled data for the evaluation process and the relative accuracy was 83% for stop time and 94% for average speed of vehicles. The results of this study revealed the efficiency as well as reliability of the developed agent-based model in traffic smoothing.