Automated Travel Time Measurement Using Vehicle Lengths From Loop Detector Speed Traps

Problem

Travel time data is a useful tool for traffic engineers, offering the possibility of better incident detection, ramp meter control, and traveler information. It will become even more useful as input for emerging technologies such as dynamic traffic assignment; more importantly, it can be used now to quantify the benefit of such new technologies before making significant infrastructure investments.

This report presents a method of measuring travel time using existing loop detector infrastructure, inexpensive computers, and real traffic data.

Method and Findings

This method uses an algorithm to match vehicles detected at one loop detector station with the same vehicle at the next station; the difference in time between stations is the actual travel time of that vehicle.

Because the loop detectors measure effective vehicle length, it is possible to distinguish individual vehicles as they pass over it. However, such a length measurement may be accurate to only two feet, depending on the speed of traffic, making it difficult to match pairs from one detector to another. It is simple to eliminate unlikely matches from this scenario; however, the data produces a series of "possible matches" that must then be distinguished in order to make sense of the information.

To eliminate false matches, the algorithm matches platoons where vehicles pass both detectors in the same relative order. The platoon should produce a contiguous sequence of possible matches; if these are long enough, the problem posed by false matches can be eliminated.

Three closely related algorithms use different strategies to eliminate spurious sequences due to false matches. The algorithms were used to measure travel times on a large data set and the average measurement error for the different algorithms ranged between 0.7 percent and 4.5 percent, corresponding to an average segment velocity error between 0.4 mph and 1.5 mph.