论文部分内容阅读
In the context of the continuous development of welding technology toward automation and intelligence,the production conditions of traditional heavy industries such as shipbuilding are still lagging,and about 80% of the weldments still rely on manual welding.The multivariety,single-piece and small-batch production mode of shipbuilding has brought great difficulties to the application of welding automation production lines.Due to the large size of the ship’s welded parts and the variable working position,the robot is required to be lightweight,portable,can work in a complex unstructured environment and can quickly change the working place.Traditional industrial robots obviously cannot meet this production demand.The emergence of fully reconfigurable modular robots can make up for this deficiency,but it has problems such as low absolute positioning accuracy and other factors that cause poor calibration of the weld tracking system.Therefore,based on the requirements of marine welding automation application,this paper conducts a preliminary study on the key technologies involved in the uncalibrated weld seam tracking system..Firstly,based on the comparative analysis of the calibration-based weld seam tracking system and the uncalibrated robotic weld seam tracking system,the basic control scheme of the uncalibrated weld seam tracking system is preliminarily determined,and the modeling process of the uncalebrated weld seam tracking system is clarified.And the camera model and image Jacobian matrix model of the uncalibrated weld seam tracking system are established.Then,combined with the actual welding task visual environment,the gray feature of the weld seam image is directly used as the image features of the uncalibrated weld tracking system for visual servoing control.In order to solve the core problem of the uncalibrated welding seam positioning algorithm,a gray Jacobian matrix corresponding to the gray features of the weld seam image is established.Based on different optimization methods,a real-time update strategy of camera motion speed in the uncalibrated weld seam tracking system is proposed.Then the experimental tests of different update strategies are carried out,which effectively validates the rationality of the proposed LM-based camera motion speed update strategy.Finally,the control strategy of weld seam tracking is theoretically analyzed,and the PI control model is introduced to eliminate the steady-state error of weld seam tracking.In order to verify the feasibility of the uncalibrated robotic weld seam tracking system,the overall scheme was designed according to the actual welding environment,the hardware of laser vision subsystem and task execution subsystem is designed,and the software design of control modules such as weld image processing,weld positioning tracking,and robot communication.Through the experimental test,the rationality of the component selection of each hardware module and the design of the control software module are verified.Finally,the weld seam positioning and the weld seam tracking experiments of different groove types are designed.The experimental results of weld seam tracking are compared and analyzed from the aspects of algorithm implementation,the norm-error between the desired and current weld seam image,and the feasibility and robustness of the uncalibrated robotic weld seam tracking system are verified.