ABSTRACT
This research proposed a particle filter and unscented Kalman filter to estimate the velocity and headway of three platooned vehicles and to predict the deceleration intent of the last vehicle expected to occur a few seconds later. In the state equations of the PF and the UKF, the trend and the GHR model were both tested as the updated algorithms of the vehicle velocity. Numerical analyses using the field test data showed that both PF an UKF successfully estimated the velocity and headway of platooned vehicles and also predicted the trend and intensity of the driver’s deceleration intent that occurred 1.5 s later. However, the prediction contains some uncertainty, because the acceleration and deceleration behaviors include the randomness of the human driver. Relaying the forecasted uncertain intent to the following car driver is the key issue, although the inferred intention was not completely accurate and appropriate interface is needed.
