The Development Of Underwater Binocular Vision Positioning System And The Application Research

In this thesis, an underwater binocular vision system was studied, which was used for underwater welding robots, providing three-dimensional coordinates of the defects or ruptures for robots. The binocular vision system likes eyes of the robots, guiding the movement direction of the robots. An underwater welding robot with vision system can finish welding work by remote control, as a result, humans don’t need to carry out welding work in the harsh underwater environment or in hazardous environments, and reduced humans’ labor intensity.Compared with the binocular vision system on land, the underwater binocular vision system is more complex, and the positioning accuracy can be affected by the water environment, therefore, the impact of the water environment needs to be studied in order to eliminating or reducing the impact, improving positioning accuracy. The hardware and software aspects of the underwater binocular vision system were designed and studied in this thesis, and the specific work as follows:(1) The configuration between the vision system and the robots was designed, the hardware working block diagram was designed also. The hardware equipment was selected, and the underwater equipment was specially designed ensuring sealing well. Based on the above, a complete set of underwater binocular vision experiment platform was established.(2) Different calibration methods were researched and the underwater camera imaging was modeled geometrically to analyze the characteristics of the underwater imaging. At last, a new camera calibration method based on radial alignment constraintand vanishing points model was proposed to weaken underwater image distortion.(3) The basic theory of stereo matching algorithm was researched and the underwater stereo imaging was analyzed both in geometric and mathematic ways. The results show that the refraction of light might cause the increase of disparities on x-axis while the projected points on y-axis shifted and the offsets were equal with the same direction. Therefore, the searching range on x-axis was increased and the semi-global matching algorithm based on BT algorithm was used to ensure the accuracy and the rapidity.(4) Tested the binocular vision system by experiments, using the control variables method to inspect the effects on the positioning accuracy with different calibration distance, different measure distance and different degrees of turbidity.(5) Completed the hand-eye calibration work between the vision system and the manipulator, and experiment was conducted to test the correctness of this work. The result shows that the manipulator could be moved to the destination position upon on the coordinates provided by vision system.