Probability hypothesis density filtering for real-time traffic state estimation and prediction
Matthieu Canaud, ,Université de Lyon, F-69000, Lyon, Lyudmila S. Mihaylova, Jacques Sau, Nour‐Eddin El Faouzi · Networks and Heterogeneous Media · 2013
The probability hypothesis density (PHD) methodology is widely usedby the research community for the purposes of multiple objecttracking. This problem consists in the recursive state estimation ofseveral targets by using the information coming from an observationprocess. The purpose of this paper is to investigate the potentialof the PHD filters for real-time traffic state estimation. Thisinvestigation is based on a Cell Transmission Model (CTM) coupledwith the PHD filter. It brings a novel tool to the state estimationproblem and allows to estimate the densities in traffic networks inthe presence of measurement origin uncertainty, detectionuncertainty and noises. In this work, we compare the PHD filterperformance with a particle filter (PF), both taking into accountthe measurement origin uncertainty and show that they can providehigh accuracy in a traffic setting and real-time computationalcosts. The PHD filtering framework opens new research avenues andhas the abilities to solve challenging problems of vehicularnetworks.