United States Department of Transportation
i
Traffic Signal Vehicle Detection, One Size Does Not Fit All
-
2019-06-28
[PDF-8.52 MB]
Details:
-
Creators:
-
Corporate Creators:
-
Corporate Contributors:
-
Subject/TRT Terms:
-
Publication/ Report Number:
-
Resource Type:
-
Geographical Coverage:
-
Edition:Final, January 20, 2017- June 30, 2019
-
Corporate Publisher:
-
Abstract:Traffic signals are traffic control devices that detect vehicles at intersections and assign right of way to road users of all types (including motorized and non-motorized vehicles and pedestrians). Vehicle detection sensors used today include inductive wire loops, video, radar, and magnetometers. Although used extensively, the performance of non-intrusive video-detectors is adversely affected by weather impacts of shadows, sun glare, fog, rain, and snow. Efficiency of remote sensing radar sensors is affected by spatially surrounded buildings and trees. All-weather operations of in-pavement sensors are disrupted by pavement degradation. The primary objectives of this research study are to review prior vehicle detection sensor evaluation studies, conduct field evaluations of vehicle detection call errors, evaluate the error in vehicle detection for selected sensor models, and interview the traffic signal engineers for field performance and cost. Field data sets were collected for four types of sensor call errors: dropped, missed, false, and locked. Data of 20 signal cycles were collected for nine vehicle detection sensor models at 30 signalized intersections (18 cities in 13 counties of the State of Mississippi). The statistical significance of the main effects of key factors (signal regions and sensor types, sensor models) using the collected call error data were analyzed. The result of statistical inference analysis including hypothesis testing and multiple comparisons at 90% certainty were used to evaluate the sensor models. There is no statistically significant difference among three signal regions and among eight sensor models and the difference in the means is relatively small. One radar, one radar/video, and two video sensor models outperformed other sensor models evaluated in this study. Additionally, hourly vehicle volume, calculated from the total vehicle counts in the 21st signal cycle, was used to estimate harmful vehicular emissions for each signal site. Emissions are higher for the signal site with the higher hourly traffic volume.
-
Format:
-
Collection(s):
-
Main Document Checksum:
-
Download URL:
-
File Type:
