Method for real-time recognition of vehicle platoons

An approach for the real-time recognition of vehicle platoons based on the time epochs of vehicles passing over a point (detector) on a link is introduced. Since platoons on a link depend on its traffic volume and the distribution of the vehicle headways, the algorithm first computes second-by-second vehicle density (vehicles per second) for each network link. To identify platoon beginnings and endings, this link density is multiplied by two fixed factors, one greater than 1 and the other less than 1, to obtain upper link density and lower link density, respectively; these are subsequently utilized in the platoon-recognition process. Using detector data from microsimulations of traffic scenarios, the platoon-recognition method is evaluated with respect to the following measures of effectiveness: percentage of vehicles captured in platoons, platoon size distribution, platoon densities, number of platoons versus percentage of free time (defined as the time between platoons), model robustness, and computational time.