Upgrade the Video Vehicle Detector Verification System(V2DVS). Final Report June 30, 2008-June 29, 2011.

摘要

The accurate detection of the presence, speed and/or length of vehicles on roadways is recognized as critical for effective roadway congestion management and safety. Vehicle presence sensors are commonly used for traffic volume measurement and control of signalized intersections and ramp meters. In addition, vehicle speed, either measured directly from a speed trap or estimated based on vehicle length, and classification from length are important for automated incident detection and the characterization and prediction of traffic demand. The market has recently seen an increasing number of out-of-pavement and wireless in-pavement detection systems. These systems are easier to install and maintain than conventional inductive loop detectors, but there is evidence from pilot installations, however anecdotal, that they have different detection characteristics and can yield different results. Due to situations including lane-changing, high ground clearance, trailers and occlusion, detectors will occasionally fail to detect, falsely detect or multiply detect individual vehicles. These errors tend to cancel each other when detection data is aggregated into bins, as in common practice. Non-aggregated (i.e. individual vehicle) data is needed on how these detectors perform in different situations in order to optimally specify equipment for certain detection requirements and environments. There was previously no way to compile individual detection data or quantify detection errors without manually comparing each individual detection event record to video ground-truth, which is a labor-intensive process that is practically impossible for large datasets. Selected out-of-pavement and wireless in-pavement detection systems, along with duplex inductive loop detectors, are installed on Route 405 in the city of Irvine to detect the passage of each northbound vehicle. Installed at the same site is the Video Vehicle Detector Verification System (V2DVS) that automates the collection and accuracy assessment of output data from all detectors under test.