Weld-line Tracking And Real-time Intersection Recognition Based On Three-line Laser Structured Light For Derusting Wall Climbing Robot

Nowadays,the removal of rust on large ship hulls still relies on manual operation during which workers hold heavy devices and stand on high elevation.However,the working environments of manual derusting are usually harsh and the work is of high tension,which does great harm to the workers' health and is not people oriented at all.Meanwhile,a lot of sands are released into the air and sea water and therefore pollute the surrounding environment seriously.Today,a lot of work have been replaced by machines and intelligent manufacturing is booming.It is of profound and great significance to develop a wall climbing robot system which is able to remove rust on ship hulls automatically.To achieve this goal,it is necessary for us to design a system that is able to track a weld-line while the climbing robot is moving and recognize the type of the weld intersection.Weld-line tracking and weld intersection recognition is key to this.In this paper,a weld-line tracking and real-time intersection recognition method based on three-line structured light is proposed.The system applies series of digital image processing algorithms,such as image enhancement,image binaryzation,noising spots removal,image districting,broken skeleton connecting and deburring,to the images captured by the camera while the wall climbing robot is removing rust on the ship hulls.A sectional fitting method is proposed to obtain the feature points of the weldline.The noising points on the skeletons are removed by a algorithm which combines median filtering and chain code characteristic distinguishing.After that,a set of points are selected into the fitting groups to do sectional fittings and obtain the feature points of the weld-line.Based on the high fault tolerance of the three-line structured light,the fitting principle of the outlines of the weld-line is proposed and the center line of the weld-line and the tracking position are determined.According to the geometrical characteristics of the three-line structured light,5 variants which are related to shape of the laser stripes and the distances between two neighboring laser stripes are defined.The influence of the intersection types to the 5 variants are discussed and a function to recognize the intersection is proposed.The experimental results show that the system is able to recognize the center of the weld-line without interference and the deviation of the position of the recognized weld-line center is between ±0.72 mm,and the deviation of the angle of the weld-line center line is between ±1.5°,which proves that the system meets the need of weld-line tracking of the derusting wall climbing robot.Cross-intersection and T-intersection can be recognized by the vision system,laying foundation for path planning of the wall climbing robot.