Research On The Visual Positioning Method Of Autonomous Underwater Vehicle In Shallow Water

The development of Marine resources is a challenge and a major demand for China’s development.In response to China’s strategic needs of "ocean power" and "Specialized,Fined,Peculiar and Innovative",the new technology of underwater calibration of a new generation of autonomous robot cameras is developed,and the underwater calibration method of anchor cameras is studied to achieve accurate underwater visual detection of autonomous robots.It is an important measure to promote the intelligent process of Marine industry and drive industrial upgrading.It has important practical and strategic significance to strengthen the sea by science and technology under the complex situation at home and abroad.Autonomous Underwater Vehicle(AUV),as a powerful equipment to explore the ocean environment and resources,can carry out a task independently with less human intervention,greatly improving the ability of human exploration of the ocean.At present,underwater robot has become one of the important tools for ocean development,and vision-based underwater robot is widely used in shallow waters.However,on the one hand,due to the complex underwater environment,information acquisition and water quality are greatly affected;on the other hand,due to the high requirements on real-time and safety of the work task,the AUV is vulnerable to the influence of underwater moving obstacles,resulting in the low autonomy of the AUV and unable to perform complex work.Therefore,it is of great practical significance to study the system performance of underwater robot.Specifically,the following achievements have been made:(1)In view of the difference between underwater camera imaging model and air environment,the influence of underwater environment on camera imaging is analyzed according to the principle of keyhole imaging,and an improved Zhang Zhengyou calibration algorithm is proposed.Secondly,aiming at the problem of local convergence of existing camera calibration algorithms,a new method of camera internal parameter optimization is proposed.This method is based on gull algorithm and adaptive parameter differential evolution algorithm.Sa DE algorithm is not easy to fall into local convergence to increases.population diversity.Experiments under shallow water conditions show that this algorithm has good accuracy and feasibility for camera internal parameter optimization,and can better solve multidimensional nonlinear optimization problems.(2)For underwater images due to absorption and scattering prone to color distortion and the problem of low contrast,puts forward an algorithm based on dark channel prior image restoration technology of underwater robot,the dark channel prior,on the basis of the simplified model of underwater of single image is used to estimate the background light,and on the basis of the recovery of underwater image.At the same time,the grey world hypothesis theory is used to calibrate each channel of the image.Aiming at the situation that the restored image is too dark,the HSV spatial multi-scale fusion algorithm is constructed to correct the image,which can effectively improve the underwater image contrast and improve the color distortion.(3)Aiming at the problems of complex structure,high cost and difficult recovery of current underwater positioning system,a combination algorithm of VIN-Mono camera and inertial measurement element is applied to the positioning of underwater robot.By integrating high frequency IMU data with camera data containing rich information,the algorithm improves the underwater positioning accuracy of the robot compared with a single algorithm,and provides an optimized solution for the high-precision positioning of the underwater robot.(4)In view of the characteristics of complex working environment and many obstacles of underwater robot,a path planning algorithm based on complex function is proposed by studying common path planning algorithms and selecting local path planning methods.After the complex function is used to transform the working environment from three-dimensional space to twodimensional plane,the minimum circle method is used to establish the environmental map to reduce the calculation of environmental modeling,and then the artificial potential field method is used to carry out the path planning to ensure that the planned path will not be affected by water flow disturbance in the process of reaching the working position.Finally,the feasibility of the proposed algorithm is verified by an underwater task.(5)On the basis of completing the key technology research,a shallow water underwater robot visualization system test platform is built,the dynamic model of underwater robot manipulator is established,the system hardware and software joint commissioning is completed,and the system is used in the actual field for visual control test,simulating the underwater rescue task.The test results show that the system can realize the underwater visualization operation of the robot.