Dynamic State Estimation in Vehicle Platoon System by Applying Particle Filter and Unscented Kalman Filter

This study applies a particle filter (PF) and an unscented Kalman filter (UKF) to estimate the headway and velocity of a six- vehicle platoon system. These two feedback estimators were used to estimate headway and velocity indirectly from several measurement variables, such as acceleration rate and velocity, of selected vehicles in the platoon. To evaluate the performance of the proposed two estimators, artificial car-following data were created to cover various speed ranges that include some acceleration and deceleration scenarios. Also, a comparison of estimation accuracy is conducted when varying the number of probe cars installed in the platoon system. Numerical analysis showed that the PF succeeded in estimating headway and velocity more accurately than the UKF, even when the number of probe cars installed is fewer and their location is varied within the platoon. The estimations by the UKF were inaccurate and the filter was unstable during all probe car penetrations except during the 100% installation scenario. The UKF is considered to yield stable and accurate estimates only when all vehicles are equipped with the sensing system, whereas the PF does not require numerous probe cars to generate accurate estimates regardless of their location in the platoon.