| As essential medical equipment,most domestic packaging for surgical needle cartridges and sutures is still done by hand today.They have been sorted and placed at the suture winder machine with great danger and low effectiveness.In order to make a precise,effective and constant workpieces sorting,it will likely become more popular to assemble the packaging product with an automated identification and self-sorting system.With a robotic arm control system and visual recognition,the study researched recognition algorithms,robotic arm control,and software design.The main research content and achievements include the following aspects:Firstly,the overall scheme of the recognition and sorting system is designed based on the assembly requirements and recognition features of the packaging card.This scheme combines multiple template matching methods to complete forward and reverse recognition and positioning,converts the end pose of the robotic arm through hand eye calibration,and outputs control information to the configuration file after trajectory planning.The motion control card reads the file information and sends motion control signals to complete the packaging card sorting.Based on the collection requirements and the sorting range of the robotic arm,hardware selection and structural design will be carried out.Secondly,the pre-processing methods of image graying and Bilateral filter are used to remove noise and sharpen edges.The exterior contours of the packaging,the skeleton contours,and the edge of the positioning holes were ensured and fitted using the Canny algorithm based on eXtended Line Descriptions(XLD)and a graph-based method for the least squares fitting algorithm.Strong and weak edges of the graphs were detected through the double thresholding,and the outliers were removed when fitting sub-pixel points based on clipping factors,completing the extraction and fitting of all edges.In accordance with the general plan,the multiple fitting plan based on XLD was produced as template images which were recognized and positioned for the packing.The images of the packaging in various positions were acquired to examine the great precision and efficacy of visual recognition through the placement experiment.The results show that the visual recognition accuracy and efficiency are 1.259 pixels and 224.3ms,respectively.Thirdly,the D-H parameter method is used to analyze and establish the kinematics model of the packaging card sorting manipulator;The forward and inverse kinematics derivation and simulation verification of the manipulator are completed based on algebraic method;Analyze the workspace of the robotic arm using Monte Carlo method;Using the method of 3-5-3 polynomial interpolation to plan the trajectory of the joint space of the robotic arm;Based on particle swarm optimization(PSO),the 3-5-3 polynomial interpolation algorithm is optimized with time as the optimization objective,and the inertia weight and learning factor of the PSO algorithm are optimized using dynamic adjustment method.Through simulation,compared with the joint space trajectory planning of the robotic arm achieved using the 3-5 3 polynomial interpolation method,the planning time was reduced by 1.575 seconds,improving the efficiency of the robotic arm sorting.Finally,analyze the principles of camera calibration and hand eye calibration,and complete the calibration experiment based on Zhang Zhengyou’s camera calibration and nine point calibration methods;According to the functional requirements of the recognition and sorting system,visual and control system software were designed,as well as system software to automate the recognition and sorting process.An experimental platform for the recognition and sorting assembly system was built,and packaging card assembly experiments were completed.The experimental results showed that the overall accuracy and efficiency of packaging card recognition and sorting were 0.16998mm and 3.355s,respectively.The overall assembly system scheme designed in this article has strong practicality. |
PDF Full Text Request