Research On Path Planning And Tracking Of Driverless Vehicles

The emergence of cars has brought great convenience to people's travel,but the number of cars has increased year by year,leading to traffic accidents,traffic congestion,environmental pollution and other issues.At the same time,the state of the driver determines the driving state of the car,and the technically mature driverless system is more reliable than the uneven driver technology.As an important development direction of the future automotive industry,unmanned technology has attracted the attention of universities and enterprises all over the world,and has become a hot research topic.Path planning and tracking control play a pivotal role in the unmanned technology system,and its performance will directly determine the success of driverless car technology.Therefore,this paper has carried out in-depth research on path planning and tracking,and established a relatively perfect driverless vehicles simulation platform to comprehensively examine the effects of the two.The main research contents are as follows:1.This paper briefly introduces the research background and significance of this topic,and analyzes the research status of driverless vehicle technology,path planning and tracking at home and abroad.In view of the fact that the current commonly used maps are not accurate and cannot be directly used for unmanned car navigation,this paper proposes to convert the commonly used WGS84 coordinate system into the driverless vehicle navigation coordinate system by coordinate transformation,and establish a map suitable for driverless vehicle navigation.2.The road topological relationship abstraction is carried out on the high-precision map,the map is abstracted into an digital map with edges and vertices,and the weight of the edge is calibrated by the lane length as the road resistance.The map is stored by the MATLAB program in the form of adjacency matrix.Then choose Dijkstra algorithm for global path planning,and introduce the idea of the algorithm and the execution steps of the algorithm in detail.Finally,the global path planning is carried out on the global high-precision map,and the shortest from East Gate to West Gate is obtained.3.A brief overview of local path planning,considering the real-time nature of unmanned vehicle path planning,this paper chooses artificial potential field method(APF)for local path planning.Firstly,the shortcomings of the current local path planning are summarized,and the ideas and mathematical models of the traditional artificial potential field are briefly introduced.Then,the driverless artificial potential field model is improved from the three aspects of road structure,cubic polynomial road boundary fitting and real-time virtual local target points,and an artificial potential field model more suitable for driverless vehicles is established.Finally,considering the dynamic and changing external environment and the dynamics and kinematic constraints during vehicle driving,a model predictive path re-planner is designed between the artificial potential field model and the path tracker,and the local path of real-time planning is obtained.The path information is output to the tracking controller module in real time.4.Established a path tracking controller and a comprehensive joint simulation platform for real-time path planning and path tracking,and the path planning and tracking effects are verified on this basis.The model predictive control algorithm is used to design the path tracker,then the influence of design parameters on the path tracking controller is studied,and the model predictive control parameters suitable for the model are obtained.The analysis results show that when the sampling period is 0.1s,the prediction step size is 20,and the control step size is 10,the path tracking controller can take both calculation speed and control precision into consideration.The path planning and path tracking programming is completed by S-Function,the vehicle model is parametrically modeled in CarSim,and several typical road environments are established.Joint CarSim and MATLAB have established a comprehensive simulation platform integrating path planning and tracking.Set different road environments in CarSim through a comprehensive simulation platform.The simulation results show that the maximum distance of the lateral position of the path from the road boundary is close to 1.3m,which is much larger than half of the actual vehicle body.It shows that the improved APF algorithm can safely plan and track the shape of the road,and verify the effectiveness of the path planning.In different road environments,the control quantity of the vehicle during the tracking process is less than 10°,which satisfies the constraints of the steering system of the vehicle.At the same time,the incremental increment of the steering wheel is less than 0.8°,which satisfies the constraints of path tracking.The validity of the path tracking controller was verified.