2012 NATMEC Dallas, Texas
Maryam Moshiri, EIT Jonathan Regehr, Ph.D., P.Eng.
Jeannette Montufar, Ph.D., P.Eng. University of Manitoba, Civil Engineering
Jonathan Foord, EIT
Michael Cantor, P.Eng., PTOE City of Winnipeg
Background
Study Methodology
Field Test Design
Evaluation Methodology
Concluding Remarks
Inductive loop sensors traditionally used for vehicle detection at intersections
The City of Winnipeg is searching for viable alternatives to loops at intersections to: Overcome limitations of inductive loops
Obtain reliable and comprehensive traffic data
Improve data collection through traffic monitoring program
Design and implement an intersection field test to evaluate the performance of advanced vehicle detection sensors: At stop-bar and advance location For vehicle detection, count, and classification
Under various weather, illumination, and traffic conditions
For 3 months of data collection
Literature Review
Jurisdictional survey
Vendor interviews
Discussions with the City of Winnipeg
Two or more jurisdictions identified the product as preferred or currently in testing
Literature is available on the field testing
Wireless capabilities
Remote viewing capabilities
Two or more functionalities
Manufacturer and vendor support
TECHNOLOGY PRODUCTS (identified through survey)
Video Autoscope Solo Terra (new version: Encore) Iteris Vantage Naztec Traficon Traficam Aldis Gridsmart
Microwave Wavetronix SmartSensor Matrix and Advance MS Sedco TC26B
Magnetometer Sensys LED Tomar Strobecom II
LeddarTech Leddar d-tec
N Test Cabinet
N Test Cabinet
Axis Surveillance video (ground-truth) Height ~ 35 feet
Iteris video Height ~ 40 feet Autoscope video
Height ~ 37 feet
Wavetronix SmartSensor microwave Height ~ 20 feet
LOOPS
ITERIS
WAVETRONIX
AUTOSCOPE
Detection Count Classification On a per zone basis and aggregate level across all lanes
Stop-bar and advance detection capabilities to provide “call” and “extension” functions to traffic signal
Missed, false, stuck on, and dropped calls
Safety vs. efficiency
Inductive loops as baseline
Algorithms developed to perform analysis
Count data from all three test sensors
Classification data from Autoscope video sensor only
Miovision as baseline
15 minute count intervals
5 level classification scheme
Adverse weather conditions
Illumination conditions (for video sensors)
Wind conditions
Traffic conditions
Continuous advancements in vehicle detection and traffic data collection technologies necessitates testing their performance This research provides a comprehensive test-bed Adequate time for data collection Uniform test bed for all sensors Evaluation of effect of specific conditions (e.g., weather, traffic) Winnipeg’s extreme weather conditions Evaluation of detection, count, and classification performance
Inputs from all detection zones are scanned every 10th of a second Customized detection data collection
procedure to reduce number of data-loggers Digital to analog conversion boards
designed for each sensor to give a unique resistance value for inputs from all zones