Research On Path Planning Of AUV Autonomous Docking And Semi-physical Simulation System

This paper aims to discuss several questions in the AUV autonomous docking.The short-distance and accurate docking of AUV relates many subjects and by far, oneof the most promising and difficult field among researchers.Based on the process of docking of AUV, this paper respectively does someresearch on the architecture system of autonomous docking of AUV, docking guidemethod based on sense of sight and path panning method of short-distance accuratedocking. Firstly, the paper gives some introduction on the application methods ofdocking around the world at present, discusses several influence factors affecting thedistance and accuracy of sensors applied in the AUV docking. Then, the papersummaries the architecture system of AUV which basically contains two kinds:Hierarchical Architecture System and Hybrid Architecture System. Combine theadvantages of these two methods, a new hybrid architecture system which can beapplied to the autonomous docking of AUV is introduced to establish the foundationfor further research. Next, after the introduction of basic concept of Binocular VisionTechnique and the process of graph acquirement of this method, this paper gives theway on how to acquire graphs in Vega simulation tool, how to set calibrate left andright vision graph in Matlab software and how to locate binocular vision in OpenCVlibrary. The accuracy requirements of distance on autonomous docking of AUV canbe obtained based on the vision simulation data and the results analysis of theBinocular Vision Locating Method. Finally, simulation experiment on AUVmovement characteristics shows the scale and ability that a AUV can move. This isimportant because the docking can only be adopted within the range of the movementof AUV. Improvements on the Artificial Potential Field is used as the path planningmethod in AUV docking after comparing the advantages and disadvantages of varieslocal path planning methods. Simulation experiment are carried out by VisionMethod(VM) and the method mentioned above, the Improved Artificial PotentialField Method(IAPFM), the location and posture information when AUV are dockingcan be received and compare the accuracy, we can say: VM is accurate only inlocation prediction and can cause errors in posture simulation which can not be ignored, while the IAPFM is good in both the location and posture prediction.Simple guide on the concept and significance of Semi-Physical SimulationSystem (SPSS) are introduced, followed by the introduction of system hardware,software and data interface. The experiment process on SPSS is also given here.Feasibility on the above methods is evaluated by the application of SPSS.