Study On Binocular Stereo Vision System Of Underwater Welding Robot

The spent fuel pool is the key auxiliary equipment to nuclear power plant, mainly used for cooling the spent fuel, maintain the irradiating radiation shielding and absorption. When subjected to external impact load or stresses, pool may lead to partial rupture of the pool bottom, so it requires underwater welding robot underwater relief. Underwater robot equipped with binocular vision system as the "eyes" is mainly used for detection and accurate three-dimensional measurement of the bottom of cracks, holes and other defects, and take the measurement results as position input of visual control system, thus controlling underwater robots to automatically detect, sealing and welding. Harsh environment in the nuclear fuel tank not only made more stringent requirements for underwater equipment, but also led to more and complex noise and difficult and inaccurate positioning for underwater imaging. Therefore, the development of the binocular vision system for nuclear power disaster underwater welding robot to achieve high-precision positioning radioactive water environment has practical significance and strategic significance.In this paper, the binocular vision system of nuclear disaster underwater welding robot was developed. the study mainly focused on two aspects: the theory and algorithms, building experimental platform and doing some measurement experiments. The specific contents were as follows:1.In order to improve the quality of the underwater images polluted by the mixed noise, an adaptive filtering method for mixed noise underwater images was proposed, of which the effectiveness was demonstrated by experiments.2.The sub-pixel corner detection of underwater calibration board image was realized, and the influence of image noise on corner detection and the influence of corner detection on accuracy of the underwater camera calibration were explored. The results have theoretical significance for improving the accuracy of feature extraction and underwater camera calibration.3.In order to improve calibration accuracy of the underwater binocular vision system. Then, an improved Zhang calibration method for underwater binocular vision system was proposed. And through polar contrast correction tests and calibration methods experiments, the correctness, accuracy and robustness of the method were verified.4.A binocular vision system of the underwater welding robot was designed. Laboratory experiments of doing the three-dimensional measurement in the underwater environment was conducted to verify the suitability and accuracy of the system. Experiment results show that, in the world coordinate system, when the true length of the measurement target was within 300 mm, the maximum relative error was less than 1.5%, and the experimental results were well.