The Research And Application Of Industrial Robot Trajectory Planning Based On Image Recognition

In recent years,the industrial robot technology and industry has been rapid developed.Besides,the industrial robot technology has been widely used in production,and the industrial robot has become the most important automatic equipment in modern manufacturing.However,with the continuous expansion of market demand,as well as the continuous development of science and technology,higher and higher requirements have been put forward to the practical application of industrial robots.In this paper,we studied image recognition and trajectory planning technology of the industrial robot which mainly included the method of feature extraction of wheel image contour based on the symmetry,the method of corner detection of wheel hub image contour based on the curvature scale space theory and the method of trajectory planning and generation based on the image feature.On the basis of the proposed theory,we discussed the advantages and disadvantages of various solutions.And then,we developed an automation manufacturing system and software of industrial robot based on image recognition technology which has good performance and can be used in practical production.Besides,this paper also deals with a simulated experiment and analysis on its performance for an experimental system.In the first chapter,we expounds the background and significance for the application of industrial robots,summarizes the current situation and difficulties of the image recognition technology,trajectory planning technology and wheel hub recognition system in the field of industrial robot.Finally,we put forward the research goal of this paper,and give the main research contents and chapters arrangement based on the research goal.In the second chapter,the basic principle of image preprocessing method for image recognition system of industrial robot is discussed in detail.And then,for the wheel hub image,we studied the theory of image de-nosing and image enhancement,extract The regions of interest(ROI)based on the mutation detection of image grey value and process image binary with threshold method.Finally,we finish the image preprocessing for the wheel hub image.In the third chapter,on the basis of image preprocessing,we propose a new method for extracting the contour feature of the wheel hub image based on the symmetrical Hough transform.In this new method,we reduce the dimension of the parameter from 3 dimensions into 2 dimensions based on the graphic symmetry of the wheel hub image.Compared with the traditional Hough transform and random Hough transform,our method has a more accurate recognition results as well as huge computational advantages.Besides,we propose a new image contour corner detection method for the wheel hub images based on the random sample consensus(RANSAC)algorithm theory.In this new method,we solve the error problem of corner detection through the wheel rotational symmetry,and successfully get the correct contour corners of the wheel hub image.Finally,we construct a feature vector of the wheel hub image for image recognition,and successfully finish the testing experiment on the experiment platform.In the fourth chapter,we research the trajectory planning theory and method of industrial robot.And we establish a kinematics model for our industrial robot based on the D-H Model theory.Besides,we also solve the solution of the robot kinematics' forward solution and inverse solution through the Robotics Toolbox in Matlab software.Finally,according to the key features of the wheel hub image extracted in the first two chapters,we complete the trajectory planning and generation of the wheel hub manufacturing.In the fifth chapter,we verify the practical feasibility of image processing system for industrial robot image recognition.On the basis of the IRB1600 industrial robot system,we integrate a series of equipment into the industrial robot image recognition automatic processing system,including industrial cameras,light sources,computers and so on.Then,we develop software which integrated all the method we proposed in the previous chapter.Finally,we process actual test experiment on the system,and the experimental results verify the performance of the proposed algorithms.In the sixth chapter,we summarize the whole article,analyze the deficiencies,and prospect for the future research work.