What if Trip Lengths are not Homogeneous?
In this work, together with the colleagues at EPFL, we study how trip lengths heterogeneity can affect the aggregated dynamics on a network-level approach.
We propose a new dynamical model for capturing the inflow/outflow relationship in a large urban area, considering the trip length distribution. We prove that by just taking into account the mean and variance of the distribution one can substantially improve the performance of state-of-the-art models that consider only the behaviour of average dynamics. The proposed model is calibrated with microsimulation data and utilized for state estimation and perimeter control. This work includes the entire pipeline (modelling, system identification, estimation, control) for studying and managing traffic flow in urban networks.
