Research On Route Planning Algorithm Of Multiple Intelligent Vehicles Based On Traffic Sign Detection In Unknown Environment

Driverless technology is a research hotspot in the field of automobile,and path planning is one of the key technologies,is an integral part of the transportation system.Therefore,it is of great practical value to study and realize the real-time and accurate traffic sign detection of intelligent car and the path planning and collision avoidance of multi-car in complex environment.This paper focuses on the detection of intelligent vehicle traffic signs,and studies the path planning and space-time conflict in multi-intelligent vehicle system under the constraint of traffic signs information as follows:(1)Aiming at the traffic sign detection problem of smart cars,this paper analyzes the advantages of choosing YOLOv5 for traffic sign detection,and proposes a GACB_YOLOv5algorithm to further improve the detection performance of YOLOv5 algorithm in traffic sign detection.The improvements include: adding Gamma transform to improve image quality and accuracy in data processing stage;using a loss to optimize the border regression loss function to optimize the model training convergence rate and border positioning accuracy;Adding the attention mechanism of CBAM makes the model pay more attention to the important features to improve the detection performance.A large number of experiments were carried out with the self-made traffic sign data set,the results show that the improved algorithm improves the accuracy of traffic sign detection.(2)Aiming at the problem of path planning and real-time obstacle avoidance for a single car in a dynamic unknown environment,this paper further improves and integrates three algorithms,A*,FLOYD and DWA,and realizes real-time dynamic obstacle avoidance of a single car on the basis of global path planning.Specific improvements include the introduction of parent node influence and obstacle ratio optimization A* evaluation function;Design corner optimization algorithm;The optimization strategy based on improved FLOYD bidirectional smoothness is proposed.The DWA algorithm is optimized based on the idea of dynamic velocity weight to avoid known obstacle interference and obstacle density,and finally algorithm fusion is carried out.Experimental results show that the improved fusion algorithm improves global optimization and local dynamic obstacle avoidance ability.(3)To solve the path planning problem of distributed multi-intelligent vehicle system,a three-layer planning strategy is adopted: the top layer is global path planning;The middle layer is local path planning;The bottom layer is collision avoidance algorithm: collision detection is carried out by rolling window method and position and steering information is exchanged to determine whether collision occurs.Then,a multi-intelligent vehicle collision avoidance coordination algorithm is designed by using vehicle adaptive priority.The simulation results verify the feasibility and superiority of the proposed algorithm.(4)In the actual scene,the route planning algorithm integrated with traffic sign detection technology is deployed on the intelligent car to carry out single-vehicle and multi-vehicle route planning experiments.The map was constructed and improved,the real-time communication of three cars was established and the experimental scene was designed,and then the dynamic obstacle avoidance and optimal path selection of cars were realized and the spatiotemporal conflicts of multiple cars were solved under the premise of obeying the traffic sign indication information.The effectiveness of the improved and designed methods is verified by experiments.