Uses of Probe Vehicle Data in Traffic Light Control

详细信息    查看全文

• 作者：Robbin Blokpoel ; ^{robbin.blokpoel@dynniq.com" class="auth_mail" title="E-mail the corresponding author} ; Jaap Vreeswijk
• 关键词：traffic light control ; probe vehicle data
• 刊名：Transportation Research Procedia
• 出版年：2016
• 出版时间：2016
• 年：2016
• 卷：14
• 期：Complete
• 页码：4572-4581
• 全文大小：270 K

文摘

Dynamic traffic light control requires traffic detection to function. Without detectors, only fixed time control is possible. Since traditional sensors like inductive loop detectors have significant installation and maintenance costs, alternative detection techniques may be an interesting alternative. Moreover, loop detectors provide point data based on vehicles passing or occupying the loops, but the prediction of the vehicle dynamics is limited due to the fixed locations of the detectors. Cooperative vehicle--infrastructure technology can monitor an intersection approach continuously and thus provide extensive information of approaching vehicles as they frequently transmit a Cooperative Awareness Message (CAM) containing all required relevant information. The most important information for a traffic light controller is the queue length for each signal group. Therefore, this paper proposes three algorithms that improve the queue measurement. The first uses GPS data only, but has the advantage of having no accumulating errors over time as integration of point measurements for traditional detection has. The second also uses information of the traffic light status and determines the queue length at the start of green by using a model of the wave speed of accelerating vehicles. The third focusses on lower penetration and combines traditional stop line detection with cooperative detection to estimate queue length while not requiring entry detection. The first two algorithms were compared with traditional queue estimation algorithms and resulted in an improvement in the average queue estimation error from 5.6 vehicles to 2.6 for the GPS-only algorithm and 1.7 for the wave speed algorithm. These algorithms were subsequently applied to traffic control by using the improved queue estimations for better planning of when to cut off green phases. This resulted in a 31% reduction in delay time and a 60% reduction in stops. The third algorithm was applied directly to the traffic control strategy due to its direct dependence on it and resulted in a delay reduction of up to 33.6%. Using the accurate positioning information from vehicles to determine the inter green times more accurately resulted in an additional reduction of only the average delay time. Overall the proposed algorithms show a great potential benefit of using cooperative data for traffic control, especially considering that no expensive equipment is required to acquire data from CAM messages.