Research On Local Collision-free Path Planning And Path Tracking Control Technologies In The Autonomous Navigation Of Intelligent Vehicle

Developing the intelligent vehicle technologies has attracted wide attention worldwide which wouldnot only improve the vehicles’ performance in safety, environmental protection and energy-saving aswell as reduce traffic accidents and increase traffic system efficiency, but also has widestdevelopment space and application prospects in the field of industry, military affairs and science.The developmental history of intelligent vehicles was reviewed, the state-of-art of the intelligentvehicle system was summarized, and the current research directions and the key technologies ofintelligent vehicle system were introduced in this dissertation firstly.The control architecture of intelligent vehicle which works as an overall structure of theinformation processing and system controlling of the vehicle is a base of the entire intelligent vehiclesystem. By analyzing the advantages and disadvantages of different architectures existed, anintegrating both distributed and central decision-making based new control system architecture,which is used for the automatic navigation of the intelligent vehicle in the road environment waspresented. This new architecture, in which the system decision-making is classified into twocategories: low level distributed decision-making and high level central decision-making, takes inaccount both real time ability and intelligence of the vehicle system well, due to the dispersingproperty and parallelism of the decision-making process in itself.Environmental information sensing is a prerequisite for the autonomous navigation and assistivedriving of the intelligent vehicle system. To solve the uncertainty of sensor information and thedrawbacks of single level sensor information fusion in the autonomous navigation system ofintelligent vehicle, a2-level sensor information fusion method based on both BP neural network andfuzzy neural network was proposed and has been applied to path following and obstacle avoidancecontrol. A BP neural network is used to fuse information from multi-ultrasonic sensors so that theuncertainty of the sensors’ information can be decreased and high accuracy of obstacle distance canbe obtained. A fuzzy neural network controller is used for tracking and obstacle avoidance in orderto realize preferable decision control of navigation.Path planning of intelligent vehicle, which further processes the sensed environmentalinformation and provides the intelligent vehicle reference for path following control guarantees theintelligent vehicle to complete the navigation tasks safely. Considering the possibility of obstaclesexisting on the road, a new local collision-free path planning method of the intelligent vehicle wasproposed in this work, which utilizes both CCD camera and laser ranging radar. The traversableroads boundaries are detected by using a monocular vision sensor, and the possible obstacles existedin the road are detected by using laser ranging radar in the detected traversable road area at first, then locating the accurate positions of the obstacles in the image area which contains obstacles. The localcollision-free path is planned by using improved VFH method. The proposed path planning methodwhich has both speed and steering angle control can save more time and strengthen the real timeperformance in contrast to the traditional steering angle control based path planning method.The motion tracking control of intelligent vehicle is the way to complete the navigation tasksand the embodiment of the ultimate effect of path planning. Aimed at the low control accuracy of theexisted path tracking method as well as its poor real time performance in the path tracking system, anovel path tracking control method on basis of road artificial potential was proposed. Imitating thedriving behavior of human beings, this method simply generates an artificial potential on the roadthe vehicle runs on according to the error information between the vehicle and the centerline of theroad in front of the vehicle, and the parameters of the controller can be adjusted online.In order to further verify the theories and methods presented in this dissertation, a test bed ofintelligent vehicle is designed and set up, which consists of road image processing system, ultrasonicsensor and laser ranging radar based distance measuring system, vehicle speed measuring system,automatic steering and driving control system as well as the onboard sensors and equipments.Experiments of path tracking and autonomous obstacle avoiding were carried out based on theestablished test bed, and the experimental results indicate the effectiveness of the proposed methods.