Abstract: In an intelligent vehicle of human-machine cooperative driving, the driver and the automatic controller share the control authority of the vehicle. Among them, the human driver can better adapt to the unknown environment, and the automatic controller has higher control accuracy in the known environment. Thus, the vehicle control authority can be transferred between them to achieve the control effect of “1+1＞2”. An authority transfer strategy is proposed, which takes into account the driver’s individual handling preferences. Firstly, a model predictive control（MPC） method is used to design the vehicle controller, and the designed controller is anthropomorphically improved according to the driver’s individual handling characteristics in visual preview, feedback control,proportional gain, brain response delay and neuromuscular system. Then the conflict of human-machine in the process of authority transfer can be reduced. Secondly, a flexible authority transfer strategy based on the spline curve method is proposed, which is constrained by the driver’s personalized preview time and reaction time. Then, the proposed strategy can be more in line with the driver’s handling preferences. Finally, the proposed strategy of authority transfer is compared with two commonly used authority transfer methods, step and gradual. The results of the driver-in-the-loop experiments show that, compared with the two commonly used methods, the path tracking accuracy of the unexperienced driver is improved by 33.8% and 32.4% through using the proposed authority transfer strategy, and the driving comfort during the process of authority transfer is improved by 50.6% and 45.8%,respectively. The path tracking accuracy of the experienced driver is improved by 42% and 33.3%, and the driving stability is improved by 57.8% and 48%, respectively.

Key words: human-machine cooperative driving, control authority transfer, human-like controller, model predictive control