Research On Path Planning Method For Unmanned Vehicles Based On Traffic Sign Detectio

Driverless technology is a research hotspot in the automotive field,and path planning is one of its core technologies.As an important traffic guidance facility,traffic signs are an indispensable part of the traffic system.In order to improve the existing traffic sign guidance path planning methods,the traffic sign detection technology is studied,the detected traffic signs are taken as the constraints of path planning,and the path planning methods that comply with the traffic sign instructions are deeply studied.The main research contents of this paper are as follows:(1)A yolov3 tiny traffic sign detection network integrating Depthwise Separable Convolution,Channel Attention and Spatial Pyramid Pooling module is proposed,and the EIo U intersection union ratio loss function and K-means++ clustering algorithm are introduced to optimize.Using lightweight network and making full use of multi-scale feature information,the accuracy and speed of traffic sign detection in edge devices are improved.Ablation experiments are carried out on the self-made traffic sign data set to prove the effectiveness of each improvement point.Compared with several other commonly used algorithms,it is proved that the improved yolov3 tiny traffic sign detection algorithm achieves the balance of accuracy and real-time in edge devices.(2)An ant colony algorithm based on adaptive pheromone concentration is improved.The adaptive pheromone concentration and penalty coefficient are proposed to change the pheromone update mode,accelerate the convergence speed of the algorithm and avoid the local optimization problem.Through barrier free pruning and path fitting,the path distance is further shortened and the path is smoothed.When traffic signs are detected in the environment,a new global path search strategy is designed to improve the quality of path planning results when there are traffic sign restrictions.Through comparative experiments,it is proved that the proposed path planning algorithm can meet the traffic sign constraints and obtain the optimal path.(3)Two experimental scenarios are built.The improved yolov3 tiny algorithm and the improved ant colony algorithm integrating traffic sign information are deployed on unmanned vehicles for experiments.Through the combination of traffic sign information and obstacle information scanned by lidar,the construction of map is improved.Plan the path in line with the instructions of traffic signs and complete the driving from the starting point to the target point.The effectiveness of the method studied in this paper is verified by practical application.