A Multi-Object Adaptive Control Method Of Large Multi-Branch Roundabouts

Roundabout appears in the early 20th century. It has played a very important role in making the traffic flow smooth for a long time. But when the traffic flow exceed the roundabout's capacity, its operation performance will deteriorate, especially the Large Multi-branch Roundabouts (LMR). Every driver will try their best to enter the LMR without signal control. The filled LMR can't release the vehicles at a high speed, for the vehicles in the inside lane will be obstructed by the outside dense traffic streams when they want to leave the LMR. And even if they can pull out the LMR, they are bound to cause some delays to the outside traffic streams. In addition, vehicles travel slowly when they in a filled LMR.In this paper, we propose an innovative control model of LMR based on accurate traffic perception. We present a concept of Optimal Number of Vehicles within the Roundabout (ONVR) and prove that LMR can get a higher real-time capacity if the Current Number of Vehicles within the Roundabout (CNVR) is kept at the ONVR level. We calculate the ONVR of the LMR base on its topology at first. We accurately perceive the LMR traffic flows and control the entering vehicles according to the leaving vehicles, keeping the CNVR on the ONVR level and getting a higher real-time capacity.In this paper, we used the VISSIM traffic simulator, which supports microscopic traffic simulation exactly and apply this control model to Zhongshan Square in Dalian, China, a typical LMR. We deployed the Wireless Sensor Networks and the Video Capture Device to detect the vehicles accurately and collected traffic information from 8am to 9am ten consecutive days. We reproduce the traffic flows and control the traffic light to make the real-time capacity of the LMR on a higher level. The result shows us that the Real-Time Capacity of Zhongshan Square raises about 40% and the vehicle travel time reduces 30% on average during the rush hour. This control method has a high practical value and can be easily applied to other similar LMR.