| Robotic manipulator's visual servo system has an important role in the field of robotic. It breaks the traditional work style of teaching or work off-line program-ming and it doesn't need pre-programming for robotic manipulator's motion, which greatly raised the robotic manipulator's flexibility. Robotic manipulator's visual servo system can be widely used in industrial manipulator systems, home servicing robot systems and so on.Binocular camera is similar to people's eyes and can get the depth information of the scene. Robot manipulator's visual servo system based on binocular camera can take the advantages of binocular camera and expand the scope of application of robotic manipulator. Based on detailed discussion of all aspects of binocular cam-era's working principle, a fixed-camera visual servo system is built in this article and a block of color-based high-precision real-time feature extraction method is pro-vided.This paper carried out the following aspects:(1) Through the process of design and production of binocular camera, syn-chronization problem in acquisition process is discussed in detail in this paper, the principle of binocular camera's calibration and epipolar geometry are also discussed. Based on the camera's intrinsic and extrinsic parameters obtained by calibration, a rectifying method of smaller distortion is used to rectify stereo pairs and has achieved very good result.(2) Stereo matching is the emphasis and nodus in research of binocular camera, on the basis of a variety of matching algorithm and related methods, a stereo match-ing algorithm applied to robotic manipulator's visual servo system is provided in this article, the speed of the algorithm is ensured at the expense of precision, so the practicality of the algorithm is improved.(3) Aiming at the high precision of calibration and stereo matching required by fixed-camera visual servo system, a method for real-time feature extraction based on block of color is presented in this paper. The localization of feature includes coarse location and fine location. The coarse location includes detection and tracking and it initializes the tracking of the object by detecting the original position of the object and calculating the size of the object and the position of central point. The fine loca-tion is on the base of coarse location. It includes cutting a small image from the ori- gin image, extracting the contour of the object from the small image, classifying the contour by K-L transform and fitting the edge points by Lease Square Error (LSE) method. The sub-pixel coordinates of the feature points are calculated by the equa-tions of lines, which provides foundation for the measurement of object's pose and position.The subject is supported by National Science Foundation (NO.60774077), Na-tional 863 Project (NO. 2007AA04Z226) and Beijing Municipal Commission of Education's Program of Technology (NO.KM200810005016). Research results to improve the quality and efficiency of robotic manipulator and expand robot's scope of application are very important. |
PDF Full Text Request