Abstract
The collection of vehicle classification data, and of heavy goods vehicle or truck data in particular, is fundamental to the design and maintenance of transportation infrastructure. Classification monitoring efforts also improve the availability and reliability of volume data, which are the backbone of traffic analyses and transportation management systems. Nonintrusive sensors minimize the adverse impact on traffic and surrounding communities (e.g., closing traffic, cutting the pavement) when installed and maintained. The application of nonintrusive sensors for vehicle classification is possible for a simplified scheme of classes, and accuracy expectations improve, along with the rapid development of the nonintrusive technology. Three sensors were tested to evaluate the accuracy and reliability of vehicle classification: Autoscope RackVision Terra, a length-based classification sensor utilizing video imaging technology; the Infra-Red Traffic Logger (TIRTL), an axle-based classification sensor utilizing active infrared technology; and SmartSensor HD, a length-based classification sensor utilizing microwave radar technology. The accuracy of traffic sensors for vehicle classification was established with simultaneous field observations (direct or videotaped) or comparison of classification data collected simultaneously by different sensors. The conclusion is that only TIRTL can provide reliable classification under ideal conditions, while Autoscope can be accurate for some simplified classes. The SmartSensor HD did not provide good class counts on the basis of various deployments in Honolulu, Hawaii.