An Unscented Particle Filter Approach to Estimating Real-Time Traffic State

Estimating the realtime traffic state is important to fulfill the intelligent traffic management, it is therefore of great interest to obtain accurate estimation of the realtime traffic so that adaptive control mechanisms can be carried out accordingly. The macroscopic traffic flow is adopted as the model of freeway, it is considered as connected by same distance segments; the traffic sensors are placed at the conjunction of these segments, their numbers are much less than the traffic state to be estimated. The compression state space is adopted, the model parameters is taken as traffic state to be estimated not constant value. An unscented particle filter (UPF) method is proposed to improve the estimation accuracy of real-time traffic state. The simulation results indicate that the unscented particle filter can increase the accuracy of the estimation in terms of the root mean square error(RMSE), compared with the extended Kalman filter (EKF).