Abstract
Traffic state estimation represents one of the important ingredients for traffic prediction and forecasting. The work presented in this paper deals with the estimation of traffic state variables (density and speed), using the so called Super- Twisting Sliding Mode Observer (STSM). Several numerical simulations, using simulated and real data, show the relevance of the proposed approach. In addition, a comparative study with the Extended Kalman Filter (EKF) is carried-out. The comparison indices concern convergence and stability, dynamic performance and robustness. The design of the two observers is achieved using a nonlinear second order traffic flow model in the same highway traffic and geometric conditions.
Kalman filter and sliding mode observer in artificial pancreas: an in-silico comparison