Improving freeway speed estimates from single-loop detectors

Many existing freeway traffic management systems (FTMS) use speed data from loop detectors as input to various traffic management functions, such as automatic incident detection, and to traveler information system components. In many cases, these FTMS include single-loop detectors that are not able to measure vehicle speed. Typically, speed estimates are made on the basis of single-loop traffic volume, occupancy measurements, and estimates of average vehicle length. Unfortunately, the accuracy of these speed estimates is generally very poor. This paper presents a method for improving these speed estimates. The proposed method is applicable to FTMS that contain both single- and dual-loop detector stations. It does not require modification to field hardware or additional field equipment. The proposed method reduces the root mean squared speed estimation error by 23% on average over the traditional speed estimation method of using a constant, average, effective vehicle length for the entire day.