Research On Decision-making Algorithm For Emergency Obstacle Avoidance Of Autonomous Vehicle Based On Safe Headway

In recent years,with the rapid development of intelligent driving vehicles,people have higher requirements for vehicle safety.At present,most of the obstacle avoidance assistance systems avoid obstacles through early warning and longitudinal braking,but seldom adopt steering mode,which has poor effect in the scene of fast speed,and will also cause certain loss to the traffic utilization rate.Therefore,this paper studies the emergency obstacle avoidance assisted decision algorithm of autonomous vehicle based on safe headway.Firstly,the functional framework of the emergency obstacle avoidance assistance system is designed,including environmental awareness,decision planning and collaborative control,and the information interaction between each link is briefly described.On the basis of the analysis of emergency obstacle avoidance scenarios,on the one hand,emergency scenarios are abstracted as typical experiments for the standardization of subsequent studies;On the other hand,because the speed will affect the limit distance of avoiding obstacles,this paper uses the safe headway to replace the distance,simplifies the complexity of the problem caused by the change of the speed,and designs the decision-making logic of the driving behavior of avoiding obstacles based on the safe time distance threshold.For the consideration of traffic laws and safety,the steering wheel torque based driver take-over steering strategy is designed,which realizes the application background of man-machine co-driving and obstacle avoidance assistance.Secondly,in order to achieve emergency steering obstacle avoidance safely and smoothly,this paper adopts polynomial as the trajectory planning method for emergency steering obstacle avoidance after analyzing various trajectory planning methods and considering that polynomial method has the characteristics of curvature continuity and flexibility.By polynomial trajectory model is established,in emergency auxiliary system of obstacle avoidance are limited in the time domain,the future around the obstacle and transport modeling.The safe headway of the front and rear vehicle are taken as the longitudinal constraint conditions,and the road width and lane line boundary are taken as the lateral constraint conditions.The optimal polynomial parameters are obtained by quadratic programming method.Therefore,the trajectory planning algorithm designed in this paper not only ensures the safety of obstacle avoidance,but also improves the driver’s comfort.Then,according to the characteristics of trajectory planning algorithm,the control algorithm is decoupled and the lateral displacement control scheme and the longitudinal velocity control scheme are designed.The trajectory following problem is transformed into an optimization problem with constraints by the lateral displacement control based on model prediction,and the control variable sequence is determined and output to EPS to realize the steering behavior.The feedforward control is obtained from the vehicle model and the desired trajectory,and the feedback control is obtained from the current lateral displacement deviation and course Angle deviation.Finally,the lateral displacement tracking controller based on FBC is established.The longitudinal speed control makes use of the difference between the expected speed and the actual speed,and the expected acceleration is determined by PI control.At the same time,the polynomial trajectory proposed in this paper is verified by simulation,and two transverse control algorithms are compared.Finally,the SCANER /Simulink co-simulation model is built,and the simulation test of the emergency obstacle avoidance decision algorithm is carried out for the two typical scenarios designed.The experimental results show that the emergency obstacle avoidance system can avoid the obstacle by steering when the braking can not avoid the collision.At the same time,the validity of the emergency obstacle avoidance system in general scenarios is verified by using the built real vehicle test platform in the form of lane change triggered,and the adaptability of the trajectory planning algorithm based on polynomial and the lateral displacement following control algorithm based on FBC is also verified.