Method Research On Path Planning And Trajectory Tracking Control Of Driverless Clean Vehicle

As an important part of the intelligent transportation system,driverless driving technology promotes the construction of smart cities and plays an important role in improving the quality of urban services and the efficiency of transportation.As an important part of urban service vehicles,driverless clean vehicles have important research value and practical significance to study their path planning and trajectory tracking control technologies to improve the efficiency of the daily operation of vehicles and ensure driving safety.This paper mainly takes driverless clean vehicles as the research object,and focuses on the method of full coverage path planning and trajectory tracking control in a structured park environment.The research content of this article mainly revolves around the following aspects:(1)Build a hardware platform that can realize the environment perception and navigation planning of driverless clean vehicles,and design a remote monitoring and dispatching system to develop and control the vehicles,improve the efficiency of algorithm development,and strengthen the safety management of the vehicles.(2)Based on the built driverless clean vehicle platform,research on the construction of environmental maps and vehicle positioning systems is carried out.Research has found that there is a problem that the vehicle cannot be automatically positioned in a closed or heavily occluded scene.In this regard,this paper proposes a method based on the DBow3bag-of-words model algorithm to assist the vehicle-mounted camera to automatically match and locate.First,study the SLAM positioning and mapping algorithm,select the cartographer algorithm with excellent performance and effect to construct the environment map.Then use the DBow3 word bag model algorithm to assist the vehicle camera to perform the loop detection of the map,and calculate the probability positioning pose of the vehicle in a closed environment,and finally output to the particle filter to track the vehicle pose,achieving a good automatic matching positioning effect.(3)After the realization of environment map construction and vehicle automatic matching positioning,this paper has carried out in-depth research on the path planning technology of driverless clean vehicles.It is found that when vehicles perform full coverage cleaning tasks in the environment,there are common problems with high repetition rate and low coverage.And when encountering the area with more obstacles,it is easy to produce the problem that the line planning is messy or even can not be covered.Therefore,this paper proposes a full coverage path planning algorithm based on divide and conquer idea,which generates segmented path sequence according to the environment map,then uses divide and conquer algorithm to match path points,and finally realizes the design of full coverage path planning algorithm by combining A* and TEB algorithm.Through the experiment,compared with the boustrophedon cellular decomposition combined with the biologically inspired neural network coverage algorithm,the results show that the algorithm designed in this paper can still better plan the neat and feasible path in the area with more obstacles,the coverage rate is as high as 91.3%,and the repetition rate is reduced to 9.4%.(4)It is the key to realize the safe driving and efficient service of the vehicle to track the path of the driverless clean vehicle accurately.In this paper,firstly,through the analysis and modeling of the two degree of freedom kinematics characteristics of the vehicle,the MPC and pure tracking control algorithms are respectively adopted to track the clean route of the vehicle planning.The test results show that the MPC controller optimization solution time is unstable,and it is difficult to obtain a reliable solution in a short time in the case of large curvature and large lateral offset.The pure tracking controller solution speed is fast,and it can adjust the working condition parameters dynamically to correct the tracking deviation,which is suitable for low-speed driverless vehicle trajectory tracking.Therefore,this paper selects the pure tracking algorithm to track the planned route in real time,and proposes the use of road curvature for feedforward control,which converts the forward-looking distance into a linear function associated with curvature and speed changes,so as to improve the pure tracking algorithm and solve the problem of poor path tracking effect of vehicles at the corners with large road curvature.Finally,the improved algorithm is used to track the planned full coverage path,the test results show that the algorithm in this paper can still accurately track the trajectory of the driverless clean vehicle in a curve scene with large curvature and large deviation.Starting from the path planning and trajectory tracking problems of driverless clean vehicles,this paper proposes a method based on the DBow3 bag-of-words model algorithm to assist cameras in automatic matching and positioning,which solves the problem that vehicles cannot be automatically positioned in closed or blocked scenes.On the basis of realizing accurate vehicle positioning,a full coverage path planning algorithm based on the divide and conquer idea is proposed to solve the problem that the route is prone to clutter when the vehicle is planning the path in the area with many obstacles.Finally based on the planned path,an improved trajectory tracking control algorithm is proposed to realize the accurate tracking of the vehicle’s path,which solves the problem of poor trajectory tracking effect of vehicle in the road scene with large curvature and large deviation.