AUV Local Path Planning Based On Monocular Vision

Autonomous Underwater Vehicle(AUV)is an important tool for human exploration and development of marine resources.AUV has the ability to perform its own exploration and sampling tasks in an unknown ocean environment.With the rise of the concept of ocean ranch in recent years,AUV plays an increasingly important role in the development of marine fisheries.In this paper,based on the tiny AUV researched and made by our lab,designed the overall structure of the AUV,AUV mathematical modeling according to the structural characteristics of AUV.Studied the target localization based on monocular vision,tracking technology and local path planning technology and so on.Firstly,AUV developed by the task group conducted an overall structural analysis.A detailed analysis of the composition of the AUV control system;by mathematically modeling the AUV,establish AUV coordinate system and earth coordinate system,and get the transformation matrix of the motion coordinate system;given the kinematics equations for AUV,analysis the static force suffered by AUV,propeller thrust and water power,given the 6 DOF spatial motion equation;the effects of ocean currents on AUV motion and path planning are analyzed which laid the foundation for the follow-up AUV's local path planning.Secondly,for underwater imaging fuzzy,noise,poor light and other factors lead to the image quality problems,the underwater image preprocessing based on adaptive median filtering and standard deviation based enhancement;analysis the monocular camera imaging model,establishing linear and nonlinear models of monocular camera,using MATLAB camera calibration kit for monocular camera used in this article underwater calibration,obtain the monocular camera's internal parameters and distortion parameters;came up with a underwater target tracking based on contour feature and particle filter is proposed to track the underwater target.And the corresponding experiments are carried out for the proposed image processing methods respectively,verify the validity of the method.Thirdly,given the definition of AUV stability according to the structural characteristics of the AUV,and analysis the stability of AUV in horizontal and vertical planes respectively;came up with a AFP local path planner based on fuzzy logic,for the different fuzzy input in current environment or no current environment,and design the fuzzy path planner with and without current respectively;analyzed disadvantages of fuzzy about local path planning,proposed a local path planning method based on Q-learning.At last,the corresponding simulation and experiment are carried out respectively for the fuzzy local path planning method and the Q-learning local path planning.For different environmental scenarios,verified the validity of the algorithm through the simulation and experiment respectively.