ABSTRACT
The behavior prediction of surrounding vehicles in a mixed scenario plays a vital role in developing advanced driver-assistance systems (ADASs) for autonomous driving, because it is an essential criterion for decision making and trajectory planning. It predicts the future actions of the surrounding vehicles based on the present and past observations of the environment, thus reducing future threats. This chapter proposes a game-theoretic approach of behavior prediction based on the Stackelberg game and Gaussian mixture model (GMM)-hidden Markov model (HMM) to improve behavior prediction performance in a mixed vehicle scenario. The Stackelberg game is used to model the reasoning of the driver. The GMM-HMM model identifies driver behavior through the past trajectory of the vehicle. The benchmark data set, Next Generation SIMulation (NGSIM), is used to train the parameters in the GMM-HMM and driver intention models. The simulated results indicate that the proposed approach predicts the future behavior of surrounding vehicles better than the HMM-based model.
