Measuring delay at signalized intersections
A typical Dutch signalized intersection has over 50 detectors: induction loops and buttons. The detectors are used to determine where to give green light and for how long. The information from detectors and signal phases is logged in a standardized format (VLOG). This makes it possible to estimate things like intensity over a period of time, vehicles that crossed during red, maximum waiting time, and so on.
CodingConnected has developed a server-client application (YAVC) that gathers and analyzes this kind of logging data. The software is used by multiple municipalities (like Rotterdam and Nijmegen) and engineering companies. Alongside the sever-client application, CodingConnected also develops a stand-alone desktop app that can read VLOG data from disk (YAVV).
One much sought after indicator for the quality of the controlling process is time-loss. Given the nature of the data, time-loss can only be estimated based on VLOG. Floating car data would allow for a much better calculation, however, this kind of data is still relatively rare and considered privacy sensitive. Therefore, I am working on an algorithm to calculate and estimation of time-loss based on VLOG data. To get an idea of the quality of the algorithm, SUMO was used to gather actual time-loss values from a simulation. These are then compared with time-loss values based on logging data from the same simulation.