Intelligent Traffic Signal Phase Distribution System Using Deep Q-Network

Traffic congestion is a worsening problem owing to an increase in traffic volume. Traffic congestion increases the driving time and wastes fuel, generating large amounts of fumes and accelerating environmental pollution. Therefore, traffic congestion is an important problem that needs to be addressed. Smart transportation systems manage various traffic problems by utilizing the infrastructure and networks available in smart cities. The traffic signal control system used in smart transportation analyzes and controls traffic flow in real time. Thus, traffic congestion can be effectively alleviated. We conducted preliminary experiments to analyze the effects of throughput, queue length, and waiting time on the system performance according to the signal allocation techniques. Based on the results of the preliminary experiment, the standard deviation of the queue length is interpreted as an important factor in an order allocation technique. A smart traffic signal control system using a deep Q-network , which is a type of reinforcement learning, is proposed. The proposed algorithm determines the optimal order of a green signal. The goal of the proposed algorithm is to maximize the throughput and efficiently distribute the signals by considering the throughput and standard deviation of the queue length as reward parameters.

A Comparative Analysis of Machine Learning Algorithms in Design Process of Adaptive Traffic Signal Control System

In recent decades machine learning technology has proved its efficiency in most sectors by making human life easier. With this popularity and efficiency, it is applied to design traffic signal control systems to mitigate traffic congestion and distribute waiting delays. Hence, many researchers around the world are working to address this issue. As a part of the solution, this article presents a comparative analysis of various machine learning models to come up with a suitable model for an isolated intersection. In this context, eight machine learning models including Linear Regression, Ridge, Lasso, Support Vector Regression, k-Nearest Neighbour, Decision Tree, Random Forest, and Gradient Boosting Regression Tree are selected. Shivakumara Swamiji Circle (SSC), one of the intersections in Tumakuru, Karnataka, India is selected as a case study area. Essential data is collected from SSC through videography. The selected models are developed to predict green time based on traffic classification and volume in Passenger Car Units (PCU) for each phase on the PyCharm platform. The models are evaluated based on various performance metrics. Results revealed that all the selected models predict green splits with 91% accuracy using traffic classification as input, whereas, models showed 85% accuracy with PCU as input. And also, Gradient Boosting Regression Tree is the best suitable model for the selected intersection, whereas, Decision Tree is not referred model for this application.

Signal Control Strategy for Supersaturated Traffic at Urban Road Intersections with a Large Technical Grade Gap

To reduce the risk of queuing overflow on the urban minor road at the intersection under supersaturation where the capacity of the arterial and minor roads shows extreme disparity, reduce the adverse effects caused by long queues of vehicles on the minor road, and comprehensively balance the multiobjective requirements such as priority of the main road, queuing restrictions, and delay on the minor road, the minor road queue model at the end of red, a road remaining capacity model, and multiparameter coordinated signal control model were established, and a multiobjective genetic algorithm was used to optimize this solution. As an example, the multiparameter coordinated control strategy decreased the delay per vehicle by approximately 17% and the queue length by approximately 30%–50% on the minor road and slightly increased the delay per vehicle at the intersection and the length on the main road queue. This control strategy can make full use of the capacity of the main road to control the queue length on the minor road, effectively reduce the risk of minor road queue overflow blocking local road network traffic operation involved, and comprehensively balance the traffic demand between arterial and minor roads. It provides a reference control method for coping with the transfer of the main traffic contradiction under the oversaturated state of the road intersection with large disparity.