Traffic Research Earns Wellington Prize

BY 
June 25, 2019

ASCE has honored S. P. Anusha, Ph.D.; Lelitha Vanajakshi, Ph.D.; Shankar Subramanian, Ph.D.; and Laurence Rilett, Ph.D., P.E., M.ASCE, with the 2019 Wellington Prize for their paper “Cycle-by-Cycle Analysis of Signalized Intersections for Varying Traffic Conditions with Erroneous Detector Data,” Journal of Transportation Engineering, Part A: Systems, August 2017.  

Traffic congestion at signalized intersections on urban arterials is a major issue faced by system operators and road users. Two important variables that characterize traffic congestion at intersections are queue length and delay. Traffic management strategies aim to reduce both queue length and delay with the ultimate goal of achieving smoother and more efficient traffic flow and lower congestion. To achieve these goals, congestion metrics such as queue length and delay are required in real-time.

Queue length and delay are spatial variables that are difficult to measure empirically. Typical engineering practice is to estimate these variables using data from location-based sensors such as inductive loop detectors. The data obtained from these sensors are prone to errors that, in turn, lead to inaccurate estimates of queue length and delay. This paper developed real-time estimation methodologies that take into account the uncertainties associated with field measurements, including variations in traffic flow and detector errors. Model-based estimation approaches were adopted because they are better adapted to handle traffic flow variations, and a Kalman Filter (KF) estimation methodology was used because it can effectively handle the detector errors inherent in field measurements.

The major contribution of this paper is the development of model-based estimation schemes in the presence of detector errors. Two traffic scenarios were studies based on the location of the end of queue and the location of the point detector. These are known as queue within advance detector (QWAD) and queue beyond advance detector (QBAD). The appropriate variables were identified for each scenario and each metric (e.g., queue length and delay) to be estimated. A comparison was made between the KF estimation models with and without the incorporation of the detector error properties. The results indicate that the KF estimation scheme was able to handle the detector errors inherent in the input data and that the proposed approach yielded better and more accurate estimates than the approach that ignored detector error. It was concluded that the proposed approach, which explicitly accounts for detector errors, can improve the estimation accuracy for queue length and delay in real-time applications.

The authors identified two engineering applications for the proposed models. The first is for estimating queue length and delay in real-time on urban arterials for real-time traveler information systems. This information would be useful for both system operators and individual drivers. The second application is for the design of intersections. A key design question is where to place the advance detectors at signalized intersections. The proposed models can be used to identify detector locations, based on traffic volume and traffic signal settings, with the goal of avoiding oversaturation conditions.

The Arthur M. Wellington Prize is awarded to the author or authors of a paper about transportation on land, on the water, in the air, or on foundations, and closely related subjects.

Leave a Reply

— required *

— required *