Prediction based real time traffic management using connected autonomous vehicles

Abstract

The increasing population of big cities and hence the increasing rate of vehicle use with the population bring important environmental and economic problems. Traf- c congestion is one of the main causes of these problems. The presence of factors that may cause tra c to slow down or even stop locally increases the density of tra c, especially in highly populated cities, and the e ect of these factors can cease to be local and a ect the entire road network. Therefore, the e ective management of tra c plays an essential role in reducing these negative e ects. In this thesis, the real-time management using the connected autonomous vehicles, namely SWSCAV, [1] was tested in the 11 km long road network using the SUMO (Simulation of Urban Mobility) environment. Then, SWSCAV [1] with and without the prediction was compared with two real-time tra c management methods, namely the Variable Speed Limits and Lane Control Systems. 2400 di erent scenarios were created changing the parameters: the control distance and the percentage of the connected autonomous vehicles in the tra c ow. SWSCAV [1] with prediction where there are 50% connected autonomous vehicles decreased the density by an average of 58.18%. This scenario provided a 61.61% decrease in the density locally with a control distance of 1250 meters.