Research On Some Issues Of Path Plan And Tracking Control For Autonomous Ground Vehicle

Autonomous Ground Vehicle (AGV) is an intelligent mobile robot, which can runautonomously, and continuously on road or cross-country. The development of AGV hasimposing on the defense, society, economy and academy, and becomes the tactic researchobject of high technology of all countries. Autonomous navigation is a fundamentalproblem for Autonomous Ground Vehicle. To resolve the practical problems in the projectof Autonomous Ground Vehicle (AGV), in the thesis, investigations to some keytechnology for autonomous navigation are presented. The focus of this dissertation can bedivided into two parts: path plan for and tracking control for mobile robots.Firstly, this dissertation fully summarized the state of the art of path plan and trackingcontrol, and tries to point out the advantages and disadvantages of them. Then a deepinvestigation has been made, and the achievements are as follows:To solve the navigation problem of indoor environments, a path planner using radardirection weights is proposed. Environment and target information are described usingtri-tuples and the direction weights are generated based on the tri-tuples. Combined withthe VFH+ method, the direction weights are filtered to make compensate to the width ofthe robot.Navigation in off road environment has always been a hard problem in robotautonomous navigation technologies. Based on the kinematics of car like vehicle, an arcregion decomposition based motion plan strategy is adopted to solve some of the problemslike safety and uncertainty of sensory. Fuzzy logic is used to evaluate the safety of the arcregion. To ensure safe traverses, terrain roughness is studied to abstract characters theaffects the vehicle. At last, a cost function was designed considering the above factors.A new method of neural network and particle swarm algorithm based mobile robotpath planning is presented. With combination of the advantages of wavelet network andRBF network, a four layers neural network is designed. In conventional method, manyhidden cells should design for every obstacle according to information of blocks, and thescale of network is very large with many obstacles. So PSO is used to train the parametersof neural network with its character of quick optimization to make the robot respondquickly to the dynamic environment.A novel controller with fuzzy non-linear parameter is proposed. The Controller isproved to be global stable with Lyapunov theory. In order to solve the problems of time consuming and hard to satisfy the control demands when seeking the parameters. Anon-liner fuzzy method is proposed to generate the parameters. Both particle swarmoptimization (PSO) algorithm and Ant Colony Optimization are used to generate thenon-linear fuzzy rule base automatically without prior knowledge.Based on the work above, pat planning and tracking control systems have beendesigned, and have been tested in indoor and outdoor environments. The system is provedto useful and efficient. As the function model of the project of "Autonomous Land Vehicle(ALV)" and fundamental research project "Micro Mobile Robots", it was successfullychecked and accepted by experts.