Two-Stage Lane-Changing Driving Strategy Based on Driving Habits and Vehicle Dynamics for Autonomous Electric Vehicles

Lane-changing (LC) critically affects traffic efficiency and safety, making it a key focus in autonomous driving strategy development. In the human-machine co-driving phase, assisted driving systems must integrate driver habits to enable effective driver-vehicle collaboration. To this end, this paper proposes an LC strategy for autonomous electric vehicles (EVs) that integrates driver habits and vehicle dynamic characteristics. It solves two crucial issues: 1) how to guarantee drivers' LC habits in the proposed strategy, and 2) how to maximize the utilization of electric vehicle (EV) dynamics on the LC performance. In the lane-changing decision (LCD) stage, we estimate the LC probability to obtain a range of LC starting positions that align with driver habits, and we select one to enhance the EV performance. In addition, in the lane-changing implementation (LCI) stage, we propose an anthropomorphic EV control to ensure the LC trajectory is consistent with driver habits, while the EV dynamics are optimized with different trajectory objectives. The simulation results show the driver's LCD is dependent on the longitudinal position difference between the preceding vehicles in the original and target lanes, and the LCD predicted accuracy reaches 95.2%. In addition, the proposed LCI can meet the differentiated LC demands, as the LCI strategies focusing on economy, comfort, and efficiency can reduce the SOC consumption by 28.6%, the wheel angular velocity by 94.4%, and the LC duration by 70.0%, respectively. Besides, the robustness of the strategy is verified by the relatively stable performance under SOCs and environment temperatures. Thus, this paper has the potential to clarify the LC optimization requirements for autonomous EVs and assist in the electrification and intelligent development of transportation systems.