Research on Driving Takeover Methods for Human-Machine Co-Driving Intelligent Vehicles in Dangerous Traffic Situations

To achieve rapid and reliable takeover of driving control in human-machine co-driving intelligent vehicles and to improve driving safety, methods for driving control takeover in dangerous traffic situations are investigated in this paper. Three takeover methods for human-machine co-driving intelligent vehicles are proposed: button-triggered takeover, brake pedal takeover, and dynamic accelerator pedal takeover. Evaluation indicators are analyzed. An intelligent vehicle simulation driving system is developed based on the open-source autonomous driving simulation software CARLA, DAQ-USB3213A data acquisition card, and an existing real vehicle. Simulated dangerous scenarios involving a child unexpectedly crossing an intersection are constructed within the system, and experiments on the takeover methods are conducted. Based on the experimental data and the proposed evaluation metrics, the characteristics of driver takeover behavior are studied, and the effectiveness of different driving control takeover methods is compared. The results show that dynamic accelerator pedal takeover exhibits significant advantages.