Bayesian implementation of a Lagrangian macroscopic traffic flow model

In this paper we apply state-estimation techniques to a model which describes the time-evolution of observed traffic patterns. We develop a switched linear state-space formulation of a macroscopic traffic flow model and then use Sequential Monte Carlo filtering and regime-based Kaiman Filter (RKF) to reconstruct the underlying traffic patterns, where observations are provided by a microscopic traffic flow simulation which runs in parallel with our model.