6. PERFORMING ORGANIZATION CODE

Traffic Management applications such as ramp metering, incident detection, travel time prediction, and vehicle classification greatly depend on the accuracy of data collected from inductive loop detectors, but these data are prone to various errors caused by hardware and software problems. The impact of these errors will propagate to the subsequent measurements of flow, occupancy, and speed and can end up affecting traffic control decisions and traveler information. In an effort to improve the quality of loop detector data, this research developed algorithms to automatically identify two types of detector errors that are difficult to diagnose in the field: splashover, the false detection in one lane of a vehicle from an adjacent lane; and pulse breakup, when a single vehicle produces two or more pulses in its lane of travel. While the focus of this study is loop detectors, the present work is likely applicable to emerging detectors seeking to replace loop detectors using wayside mounted sensors. Most of these detectors emulate the operation of loop detectors, so, while the mechanism of splashover and pulse breakup differs across the different detection technologies, its manifestation should b