Motion Control System Design Based On Linux Integrated Machine Vision

In the design of complex visual robot systems,the most vital thing is to tackle the issue of hard real-time and soft real-time compatibility and the scarcity of open source motion control library.In response to the mentioned bottleneck,laboratory robotic arm is applied by this thesis as the motion control model,the workpiece grabbing is exploited as the background,on the visual servo control platform,a variety of motion control systems based on the Linux integrated machine vision are designed.The main research contents of this thesis are as follows:(1)The functional requirements and performance requirements of the system during the sorting of workpiece were summarized,the matter of the X86 controller on the market could not meet the industrial real-time demands was probed.Firstly,the task switching and interrupt trigger mechanism in the traditional Linux system was analyzed,the foremost reasons for restricting the real-time of the system were summarized.Afterwards,the hard and soft real-time dispatching problems was refined by analyzing the mainstream real-time improvement solutions based on Xenomai.Finally,the real-time system was tested from the following three perspectives,that is,Ether CAT communication,external interrupt trigger and visual processing procedure.The test results confirmed that the real-time of the above system has increased by 30% compared with the traditional visual robot systems,and its communication speed reached a microsecond level,which provided real-time platform for the next robotic motion control and visual algorithm.(2)A motion control model was established by the adoption of ABB manufacturers’ IRB1410 type robotic arm.Firstly,based on the D-H parameter method,the physical parameters of the robotic arm were exploited to establish a mathematical control model.Secondly,the solution of robotic arm positive motion equation was derived through the mathematical model,and verified by the imported robot tool box in the MATLAB.Ultimately,the arm sports space of the model was seeked by Monte Carlo method and the trajectory planning of the robotic arm was prepared.As for the target detection and positioning,a visual detection system was built by this thesis based on the visual characteristics of the workpiece.Among them,in the process of identification and positioning of the workpiece,for the sake of handling the malpractice of target recognition based on Blob analysisit,target recognition algorithm based on edge treatment was exploited.The experimental results analyzed that the edge processing rate of Blob operator was higher and the positioning was more accurate.(3)The “3-5-3” hybrid polynomial interpolation algorithm was chosen to plan trajectory,after comparing the advantages and disadvantages of three polynomials,five polynomials,and “3-5-3” hybrid polynomials based on the performance requirements of the sorting robotic arm system.Simultaneously,aiming at meeting the simplest and shortest performance indicators,the improved particle group algorithm was utilized to calculate the optimal time solution of the robotic arm in the trajectory planning.For shortening the workload of the developers,an open source polynomial and particle group search for excellent dynamic libraries were created on the Codesys platform,eliminating the tedious steps of using software such as MATLAB to solve the robotic arm parameters.(4)Implement systematic construction and testing based on the above work.Firstly,the hardware platform of the experiment was introduced,and then the key functional module of the system was designed.Finally,the system was tested comprehensively according to the performance requirements of the system.The test results demonstrated that the system of this thesis met the real-time standards of the industrial and conformed to the expected design goals.