Non-Cooperative Game Theory Based Driver-Automation Shared Steering Control Considering Driver Steering Behavior Characteristics

The challenging issue of driver-automation copilot opens up a new frontier to enhancing driving safety and comfort. Due to the complex and changeable external environment and driver-automation interaction, shared control which can adaptively coordinate the control authority between human driver and intelligent automation according to different driving conditions still remains open and challenging. The novelty of this paper is to introduce the concept of "desired lateral driving zone"", act-and-wait" intermittent characteristics and several statistical quantitative indexes for driving ability, aiming to better describe lane-keeping behaviors of different driver groups, especially for the steering maneuver scenarios. In order to enhance safety and reduce conflict issues, this research goes one step further in investigating the active coordination of control authority, and an adaptive driver-automation control authority model is designed, effectively providing smooth transition of control authority as the game evolves. Then a non-cooperative game based model predictive control algorithm taking into consideration driver steering behavior characteristics is proposed to improve friendliness and reliability of driver-automation interaction. Finally, the comparative experimental results demonstrate that the proposed shared control algorithm can achieve a favorable improvement in driver's acceptance, riding comfort and safety.