This research will focus on reconstructing fundamental diagrams of heterogeneous traffic from trajectory data. The NGSIM data is the basis of the study, and some filtering methods are applied to look at the traffic states that sustain a period of time. Specifically, the researchers focus on the auto and truck data, considering four vehicle-following types: auto following auto, auto following truck, truck following auto, and truck following truck. The speed-density relations under the four categories are plotted using kernel smoothing regressions. There are some interesting findings and potential issues with the data, and the researchers are working on further checking the data and deriving some useful criteria to filter stable states.
