Heterogeneous traffic estimation with particle filtering

This article considers the state estimation problem for heterogeneous traffic characterized by various vehicle sizes. The interactions between vehicle classes display unique features that are distinct from homogeneous traffic flows. Modeling and estimation methods for heterogeneous traffic, however, still remain relatively unexplored. This article adopts a particle-filtering approach to sequentially estimate the traffic state with a heterogeneous traffic flow model that allows overtaking and creeping behaviors. Numerical experiments are introduced to evaluate the estimator performance, indicating that the filter can reduce estimation errors by up to 48% when compared to pure forward simulation of the model.