Research On Lane Change Process Of Self-Driving Vehicles In The Highway Sence

Autonomous driving technology is one of the important development directions of automobiles at present.The lane changing and overtaking process in the highway scene is an important part of self-driving and driving assist system.Vehicles driving at high speeds are likely to be exposed to emergency situations.Reasonable online change lane decisions,safe and feasible lane changing trajectory planning and accurate trajectory tracking have important implications for the self-driving vehicle on the highway.The steer-by-wire(SBW)system eliminates the traditional mechanical connection and achieves more flexible steering control,which contributes to achieve higher levels of autonomous driving.It can also help track the planned lane change trajectory more safely and effectively.Firstly,based on the study of relevant lane change model,a lane change decision mechanism in the highway scene is designed,which can comprehensively consider discretionary and mandatory lane change.The lane change intentions model based on BP neural network is established in the decision mechanism to simulate the intention of drivers.The lane change model for safety is designed based on the minimum safe clearance.The Nagatani model is optimized and served as the lane change revenue model,which constitutes the lane change decision mechanism.The designed lane change decision mechanism can satisfy the lane change expectations of traditional drivers and obtain lane change revenue on the premise of safety,overcoming the traditional misjudgment of drivers caused by inaccurate perception of road and traffic conditions.Once a self-driving vehicle makes a change lane decision in the highway scene,the corresponding lane changing trajectory can be planned.Considering the kinematics and dynamic constraints of the vehicle at high speed,the fourth-order Bezier curve is adopted as the trajectory planning method.By analyzing the driving conditions of the main vehicle and the vehicle ahead in the current lane and optimizing the control points required for trajectory fitting,reasonable and feasible Bezier curves can be planned for the self-driving vehicle as lane changing trajectories in different conditions.Finally,a steer-by-wire test bench with a front axle structure is built,with two sets of electric power steering motors and controllers used as prototypes.A corresponding internal model controller(IMC)is designed,considering mechanical delay of SBW system.Then,the nominal model of the steering actuator in the experimental bench is obtained by the system identification,which as a part of the designed 2-DOF IMC controller.The experimental results show the hardware-in-the-loop experimental platform of SBW system can well track the desired front wheel input and the driver’s steering wheel angle input,with shorter delay and smaller steady-state error compared with the parameter-optimized PID controller.It can also well track the double lane change condition and the planned lane change trajectory.The effectiveness of designed SBW system and the traceability of the lane change trajectory have been verified.