A Cooperative Trajectory Planning System Based on the Passengers' Individual Preferences of Aggressiveness

• Published in: IEEE transactions on vehicular technology Vol. 72; no. 1; pp. 395 - 406
• Main Authors: Yan, Yongjun, Wang, Jinxiang, Wang, Yan, Hu, Chuan, Huang, Hanwen, Yin, Guodong
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.01.2023; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Aggressiveness; Automobiles; Cooperative speed adjustment; Customization; Games; Lane changing; Passenger comfort; personalized driving; Planning; Potential fields; Safety; Trajectory; Trajectory optimization; Trajectory planning; Vehicle safety; Vehicles
• ISSN: 0018-9545; 1939-9359
• DOI: 10.1109/TVT.2022.3203083

Diverse passengers have various driving habits and driving preferences, thus the function of personalized driving is one of the most important features for autonomous vehicles (AVs). In this paper, the variable driving style factor is proposed to describe the preferences of aggressiveness for different passengers. Based on this, a cooperative trajectory planning framework that includes longitudinal speed adjustment and lateral trajectory planning for lane changing is proposed. In the longitudinal speed adjustment, a cooperative game approach that considers driving safety, ride comfort, travel efficiency, and the preferences of aggressiveness is proposed to extend the longitudinal distance between the ego car and cars moving on the adjacent lane before lane changing. In the lateral trajectory planning, an alternate trajectory set (ATS) method is proposed to obtain a set of trajectories, and the artificial potential field (APF) method is deployed to select the optimal safe trajectories subject to the environmental constraints. Then, the B-spline method is utilized to generate safe and smooth lane-changing trajectories considering the driving style factors. Finally, a modified pure pursuit tracker is designed to accurately track the planned trajectories. Experimental results show that the real-time system can generate safe and personalized trajectories for AVs with different driving preferences, demonstrating the feasibility and effectiveness of the proposed framework.