A human-like longitudinal decision model of intelligent vehicle at signalized intersections

Recently there are significant researches on behavior decision of automatic driving vehicle at intersections. However, previous studies pay most attention to the issues of fuel consumption, efficiency and safety, less consider the feelings of occupants and drivers. This paper proposes a Longitudinal Decision Model (LDM) which can take comfort indicator and timeliness indicator into account to get an optimal speed profile based on Markov Decision Process (MDP) at signalized intersections; and a method of trapezoidal speed planning which is designed based on human driving process is embedded in this model. Not only is the model able to consider the immediate effect of the current vehicle's action but also take the long-term influencing factors into account to obtain the optimal traveling scheme of the vehicle. The experimental results show that the model is reliable, and these results of the actual implementation are highly consistent with the planning.