Reliability of Automatic Traffic Monitoring with Non-Intrusive Sensors

摘要

Automatic traffic monitoring improves the availability of traffic data and provides fundamental and essential information for developing strategies of traffic control and operation and meeting increasing needs for a variety of traffic data to support the implementation of ITS. Traditional intrusive technologies continue to be widely employed in the United States. However, in addition to pavement cutting and boring, serious traffic delay may occur due to temporary closure of traffic lanes for installation and periodic maintenance of intrusive traffic detectors. An alternative to obtain traffic information without traffic disruption is to use non-intrusive technologies that are easier to be installed, operated, and maintained. Non-intrusive detection sensors are usually deployed over roadways or at roadsides, which minimize the interruption to traffic operations and eliminate the risk of exposing service personnel to oncoming traffic. There is a growing interest and activities in testing and applications of non-intrusive sensing technologies for portable or permanent traffic data collection. This paper is to present an analysis on the performances of non-intrusive sensors under various traffic and environmental conditions. We focus on evaluation of automatic traffic monitoring, including collection of vehicle classification, volume and speed, with video image processing (Autoscope), microwave radar (SmartSensor HD) and active infrared (TIRTL) technology and discuss the reliability of the three sensing technologies with emphasis on the deployments on high speed and high volume arterials. We interpret the relationship between the performance of volume/speed and classification detection, and present the deficiencies of these sensing technologies. The main conclusion is that the sensitivity to external influences varies for volume, speed and classification detection. We note that good performance in vehicle classification is a sufficient but not necessary condition for good performance in volume and speed detection.