Investigators:
Dwayne A. Henclewood, Ph.D. Student, Georgia Tech
Dr. Angshuman Guin, Research Engineer II, Georgia Tech
Dr. Richard Fujimoto, Professor, Georgia Tech
Project Overview:
Traffic congestion is an $87 billion “cost” to the US economy. In 2007, Americans spent approximately four billion additional hours and purchased an estimated three billion gallons of additional gas due to congestion [1]. Historically, road congestion was “fixed” by increasing the roadway’s capacity. However, expanding roads, especially in urban areas, has become increasingly difficult and expensive. Thus, there has been a significant push by the government, private industry, and the research community to develop and implement alternate means of alleviating congestion. This research project is developing a methodology to provide both public and transportation facility managers with current and near future arterial performance measures. It will utilize point sensors to develop an online, data-driven, microscopic traffic simulation approach to determine and provide the necessary information to aid in the decision making process as to how to use and operate transportation facilities with greater efficiency. More specifically, the problem that this research effort will be addressing is the lack of available real time and near-future traffic performance measures along arterials that can be used by facility managers to better manage facility operations. Such performance measures include speed, travel time, delay estimates and queue lengths. It is envisioned that such information will be disseminated through a variety of media including dynamic message signs (DMS), highway advisory radio (HAR), the internet, and in-vehicle and other portable GPS navigation systems.
The research methodology involves the use of point sensor traffic data to drive a microscopic traffic simulation, VISSIM, in real time. The data from the detectors will be transmitted and then be used as input in an empty VISSIM model of the area being studied. Once the necessary data is implemented in VISSIM, current and near-future traffic states, as well as performance measures, will be estimated.
Figure 1 illustrates the conceptual framework for developing a real time, online, data-driven simulation tool. The first step is to obtain traffic-related data from the network’s roadway detectors. These processed detector data will then be used to populate a VISSIM model of the study area. Once the traffic’s current state is captured in the simulated environment, the model will be used to provide an estimate of future traffic conditions. From these future states, a probable future state is then estimated. In this framework, both the probable future traffic state as well as the traffic’s current state can be transmitted to the various end users.
A number of preliminary tests have been conducted to determine the feasibility of the proposed methodology. One such test was aimed at determining whether a VISSIM simulation can be driven by detector data streaming in real time, and having its performance measures reflect those of the area being simulated. To conduct this experiment, a road corridor on Georgia Tech’s campus was selected as the arterial to be studied. The data stream came from temporary mid-block and boundary detectors that were placed throughout the corridor.
The results from the collected data and the output from the VISSIM model were compared to determine how well the data-driven simulation was able to reflect the performance measures of the real world. Travel time was the primary performance measure that was analyzed. It was concluded that the research results supported the likely feasibility of the proposed method. Sources of error have been identified and future work will implement mechanisms to address these errors.
1. D. Schrank and T. Lomax, “Urban Mobility Report 2009,” Report for the Texas Transportation Institute, 2009
Sponsored by the Georgia Department of Transportation (RP 09-01).
April 2009-October 2011