Research On Visual Navigation Technology Of Product Oil Tank Cleaning Robot

With the continuous expansion of the national oil reserve base,China's oil reserves have also gradually increased.As a special application for robotic cleaning,refined oil storage tanks are rapidly developing.As countries and companies have increasingly improved the safety requirements for storage tanks and the awareness of green environmental protection,companies have gradually paid attention to the improvement of the safety,economy and efficiency of tank cleaning.The traditional cleaning methods are gradually being replaced,pushing the cleaning robots towards a highly intelligent,autonomous,and personnel-free tank.At present,some achievements have been made in the design and application research of tank cleaning.However,there are still problems such as poor cleaning autonomy,low cleaning efficiency,poor cleaning effect,and long traversal time.On this account,this paper introduces visual navigation technology into tank cleaning robot.Machine vision is a key research topic in today's robotics and artificial intelligence.In recent years,robot navigation and positioning have become more and more widely used.Vision sensors have many advantages such as wide detection range,comprehensive detection information,and fast information acquisition.In this paper,the following work has been done specifically for the research of visual navigation technology for refined oil tank cleaning robots:1.Study the overall plan of the visual navigation system for tank cleaning robots.The advantages and disadvantages of robot navigation technology are summarized and analyzed.The composition of robot vision navigation system is discussed.Based on this,the overall scheme of vision navigation system and the selection of software and hardware and the construction of the entire visual navigation platform are completed.2.A global image target extraction method based on geometric matching is studied.After basic research such as camera calibration and image acquisition,image preprocessing methods such as color plane extraction,contrast enhancement,and filtering are used to reduce the influence of external interference on the image.Secondly,through the image feature extraction algorithm,based on the image geometry matching method,combined with LabVIEW software to achieve the target extraction in the global image.The rapid identification and positioning of obstacles and robot position information in the environment enables the detection of the oil tank environment.3.The path planning method based on the optimized and improved BUG algorithm is studied.The coordinate modeling method is used for environment modeling,and the cleaning environment is fully covered based on the inward-connected regular hexagonal topology.Based on the recognition result of environmental detection obstacles,the path planning method based on Bug algorithm is optimized and verified by simulation analysis.The planned path is combined with the robot's own camera ranging system for obstacles.The distance between the robots is measured to achieve robotic walking and safe,fully covered tank cleaning tasks.4.Experimental Research.Build a robot vision navigation test system.The verification test of tank size,obstacle position,and robot walking path identification proved the accuracy and rationality of the system.