Design Of Robot Tracking System Based On Real-time Visual Guidance

With the development of robot technology,the unmanned factory has become the trend of the future factory.The higher the factory automation means the more complex the industrial production environment.The robot control system with vision can adapt to the complex industrial environment.To reduce the cost of hardware and the complexity of the system,this paper develops a real-time visual/motion control platform based on the Windows real-time Expansion Kit KRTS(Kithara Real time Suite).The platform can run on the PC and transmit data between the servo and industrial cameras through Ethernet.The platform can shorten image processing time because of PC's powerful computing performance,and visual system and motion control system can transmit data without special communication protocol.The platform achieves the versatility and standardization of control platform.The paper developed robot tracking system based on a real-time vision guidance,and the system software used the modular programming ideas.In this system vision processing system and motion control system are independent but are in the same Visual Studio Project.It is convenient to maintenance and subsequent development.The paper established mathematical model of a SCARA robot tracking.It mainly includes:SCARA manipulator kinematics model,robot trajectory planning model and visual tracking model.According to the three models,the paper proposed mathematical algorithm correspondingly.Finally,the algorithms were simulated to verifying correctness of algorithms.In this paper,we studied the related algorithm of visual tracking and then extracted the feature of the image,then use GMM algorithm to track the nib.According to information above we calculated the location,speed and acceleration of the nib in every image processing cycle real time.In order to make the robot motion smoothly,an algorithm based on position,speed,acceleration and time is proposed.Finally,the acceleration and deceleration algorithm are simulated and verified.Finally,experiments are designed to verify the accuracy and stability of the robot writing system.In addition,a real-time visual system testing platform is built to test the processing jitter of real-time vision.The testing platform verified the real-time performance of the vision system.