Model Estimation for Car-following Dynamics based on Adaptive Filtering Approach

Identification of driver behavior models using data has been an essential problem for the development of high-fidelity micro-simulation and design of vehicle-based intelligent systems. In this research, our focus is on model estimation of car-following, a crucial element of tactical driver behavior, using data collected from real traffic. By theoretical exploration of the relation between the Kalman filter and the recursive least square (RLS) method, a mathematical model estimation framework is proposed based on iterative usage of the extended Kalman filter (EKF). Numerical experiments have been conducted in the estimation and evaluation of a generalized GM model using closed-loop simulations. Accordingly, the applicability of the approach has been identified with further research potential.