Vibration Suppression Of Deformable Linear Object Based On Vision Feedback

Deformable Linear Object, such as hose and cable, is widely used in daily life andindustrial applications, for instance, assembling hoses and cables in the productionlines in factories, suturing wounds of the medical robots. In these applications, itremains a problem for robot manipulation and precise control, which also limits thefurther development of automation of electronics automotive and relative industry.National Natural Science Foundation Project “Research on the manipulation ofthe Deformable Linear Object” is generated based on the above, aiming at exploringthe theory of manipulation of DLO, provide guidance of practical applications andpush relative area forward.In this thesis, we firstly introduce the research profile of DLO and the foundationof the experiment platform, including the passive monocular vision system based on aCCD camera of Gigabit Ethernet, the software system is consist of OpenCV andMFC, we focus on the moving target tracking and image processing algorithm,dynamic background descending, edge detection based on Canny operator andthreshold method of moving target tracking, compared the three image processing withthe merits of the algorithm according to the performance of time complexity,illumination stability and motion blur stability, we found the third method is the fittestone under the environment of the experiment. More in-depth, we improve thethreshold into Adaptive-threshold, which is chosen as the algorithm of our robot visionsystem.We also introduce the camera model, Linear Pinhole model and Nonlinear modelwhich take distortion into consideration, the theory of calibration techniquesespecially MATLAB calibration toolbox, and then we use it calibrate the camera.Finally, we propose an approach to suppress the vibration of DLO based onrobot vision. By means of real-time image processing of high-speed visual feedback tofigure out the geometry coordinates and posture of DLO, we use PID control andverify the validity of the theory combined with the expected posture of the DLO.