| As a new field based on various interdisciplinary, theory researchand practical application of the computer vision make a great progress.Especially, the development of image processing technology and highperformance computer provide in recent years favorable conditions forcomputer vision research and application. The robotic technologybecomes a top science integrated with many subjects now. It has muchtheoretic meaning and practical value to apply computer vision to therobotic field.In this paper, a servo system based on binocular stereo vision was builtto control a 6-DOFs redundant actuation parallel mechanism with 10inputs. It’s a closed-loop system. Real-time pose and inverse kinematicresult can be calculated by the system. Main steps list as following.First, a fast calibration method based on perspective imaging model isused to calibrate the binocular stereo vision system, and then result isoptimized by Partical Swarm Optimization(PSO) to improve the precisionensuring the high speed of the process.Second, an acquisition and display system of vision information basedon Labview was built. It can observe the object state dynamically.Next, a processing system of vision information was built. It contains two parts, image pre-processing and searching of feature point. Themethod of searching feature point considers the center of three featurecircles of different size as feature points, then these points are searchedbased on template. In this way, the process of stereo correspondence canbe omitted.Finally, the parameter of pose estimation can be calculated by matrixesof translation and rotation of 3D, and the inverse kinematic model wasbuilt by geometric property of redundant actuation parallel manipulator.The inverse solution is feedback to the control unit of the parallelmanipulator, the closed visual control system is then formed for theredundant actuation parallel manipulator.The simulation experiments prove the rightness and efficientness ofall algorithms in this paper. |
PDF Full Text Request