Research On Local Path Planning For Intelligent Vehicle Driving Based On Improved Artificial Potential Field

Active obstacle avoidance is one of the key components of the core technology of intelligent driving vehicles.It uses advanced sensing technology including millimeter wave radar,laser radar,machine vision and so on to perceive and feedback the road environmental information,and then processes and analyzes whether there are obstacles through the system,and plans the path to avoid obstacles.However,due to the highly dynamic characteristics of the road environment,it is necessary to identify the unpredictable obstacles quickly,accurately and in real time and adjust the local path planning.Therefore,this paper takes the local path planning method as the research object,takes the security and timeliness as the evaluation indexes,improves the traditional artificial potential field,constructs the local path planning algorithm based on the improved artificial potential field,and uses the Edge Computing to optimize it.The main research contents are as follows:(1)Through the study of the traditional artificial potential field algorithm,the mechanical generation mechanism of its local minimum and target unreachable problem is analyzed.Based on the function of millimeter wave radar to measure the azimuth θ0 of obstacles,In the static scene,introducing the distance between the experimental vehicle and the obstacle as the repulsive force regulating factor S(M,Mg)to ensure that the repulsive force near the target point is not too large to solve the problem of unreachable target.On this basis,establishing the mathematical model of the additional force k·S(M,Mg)cosθ by introducing the direction Angle θ(>θ0)and the gain coefficient k of the additional force,which ensures that the resultant force of the experimental car is not zero at the minimum point of the traditional algorithm,which solves the minimum problem of the traditional artificial potential field algorithm.This method is proposed for the first time in this paper.Its advantages are simple model and easy to obtain related control parameters.According to the detection of the velocity and position of the object by radar,the position of the dynamic obstacle can be predicted at any moment,and the obstacle avoidance algorithm based on position prediction is established.(2)The algorithm flow and the time-effectiveness and safety evaluation model based on local path planning of improved artificial potential field algorithm were designed.The effective value ranges of additional force gain coefficient k,gravitational gain coefficient a and repulsive force gain coefficient β were determined by orthogonal experiment method.Using MATLAB simulation software,the peak value of repulsive force and resultant force is analyzed numerically.In the static scene,the repulsion regulator can obviously reduce the repulsion effect near the target point.The peak values of repulsive force and resultant force of the improved algorithm both decrease exponentially with the increase of the distance between the obstacle and the starting point of local path planning.The repulsive force can be attenuated to zero near the target point,which ensures that the experimental vehicle can achieve the targe;The introduced additional force can change the magnitude and direction of the resultant force at the local minimum point,which solving the local minimum problem of the traditional algorithm.The simulation time of the two algorithms before and after the improvement is 0.26s,the timeliness is basically the same,and the safety index is also increased from 0.0188 of the traditional artificial potential field method to 0.305.In dynamic scene,the position prediction obstacle avoidance algorithm can avoid dynamic obstacles reliably.(3)On the basis of local path planning and taking timeliness as the optimization objective,The edge computing V2I(Vehicle to Infrastructure)unloading model and the local device,V2I and V2V(Vehicle to Vehicle)combined unloading strategy are applied to the improved artificial potential field algorithm local path planning.For the V2I model,the task is divided into task units,and the optimal unloading ratio(0.93)and its energy consumption(27.2J)based on the minimum time delay are obtained under different conditions such as experimental vehicle speed and communication status.The optimization timeliness of the joint unloading model in different unloading sequences is analyzed.V2I unload first(6.72s)is better than V2V unload first(7.08s).Finally,the method of changing the step size was adopted to carry out the path planning calculation independently by using the local equipment and the edge server.The results show that the two paths basically coincide,which proves the feasibility and stability of the edge calculation applied to the local path planning data calculation.