Obstacle Avoidance Control Strategy Of Autonomous Vehicle Based On Model Predictive Control

With the development of computer technology and semiconductor technology,the improvement of people's living standards,automatic driving is more and more concerned by the industry.In the process of intelligent vehicle driving,the road environment changes at all times,such as other vehicles driving in different lanes,traffic signs,etc.,the controller needs to take effective control strategies to control the intelligent vehicle to avoid obstacles and drive smoothly in the lane.In order to design the path planning strategy for intelligent vehicles to avoid obstacles,we should not only consider the road safety,meet the road driving specifications,but also consider the dynamic and kinematic characteristics of the vehicle itself,ensure the driving stability,and how to make a balance between the two,which is the focus of this paper.To solve this problem,this paper proposes a path planning based on model predictive control,and completes the following work:First of all,analyze the kinematic characteristics of the obstacle vehicle,predict the movement state of the obstacle vehicle in a short time according to its kinematic characteristics,and determine the collision safety conditions of the main vehicle and the obstacle vehicle based on this.The mathematical model of driving risk field is established,which includes the kinetic energy field of obstacle vehicle based on two-dimensional Gaussian distribution,the potential field of lane keeping based on the crossing time and the potential field of target trend.Secondly,a path planner based on model predictive control is designed.The dynamic and kinematic model of the main vehicle is established.In order to prevent the main vehicle from colliding with the barrier vehicle,ensure the main vehicle to drive in the lane area,and make the main vehicle drive to the target position,the optimization objective is to minimize the kinetic energy field of the main vehicle,the lane keeping potential field of the main vehicle,and the target trend potential field.In order to ensure that the main vehicle can maintain its dimension without interference Driving at the current expected speed,taking the minimum tracking speed error as the optimization objective 2,taking the main vehicle as the guarantee that the control change of the vehicle in the driving process will not be too large,taking the minimum change of the front wheel angle and the longitudinal acceleration of the main vehicle as the optimization objective 3.The final optimization objective is determined by selecting the appropriate target weight factor,and considering the factors of the main vehicle steering system saturation and driving stability,the dynamic constraints on the front wheel angle and the side slip angle of the mass center are added respectively.Finally,a path planner based on BP neural network is established.Considering the two factors of driving risk and vehicle stability,the driving risk factor and vehicle instability factor are determined.The two factors are input into BP neural network as parameters for training,and the driving risk weight is determined,and the weight is input into the obstacle avoidance path planner based on model predictive control.This paper uses the idea of artificial potential field method to establish the driving risk field,and combines the driving risk field with the model predictive control algorithm by taking the driving risk field function as the optimization objective.It not only completes the modeling of the dynamic driving road environment through the driving risk field,but also solves the multi-objective optimization problem in the obstacle avoidance path planning through the model predictive control algorithm.