Based On Visual Navigation And Positioning Of Ground Mobile Robot Docking Technology Research

This study mainly researches on the docking technology of mobile robot which is based on vision, including the design of docking system, color image segmentation, object recognition, camera calibration, distance measurement based on monocular and robotics control strategy and so on. By choosing appropriate algorithms and their combination, the docking system has achieved good results. In this paper, the research in the followings:According to the problem and task, designing a reasonable docking system, it is divided into two parts:docking station and docking robot. The docking station is composed by a circular color-coded and a60-degree cone interface; docking robot is four wheels robot using visual navigation, connected to the computer with the motor drive by CAN bus, to complete the docking mission through the identification and location of the color-coded.In this paper, the docking station marked segmentation based on a new threshold segmentation method based on HIS color model is brought forward which improves former segment result. When marking image is a single color, distinguish well with background, the segmentation method is very effective. Then edge extraction used the simplified Canny operator, and compared with other edge extraction operators; the center of gravity of mark is calculated used mark contours, instead of using the tradition region pixel.In object recognition, using the object contour to rough judgment, and then using the template matching method to further matching, this method greatly reduces the time of recognition; bring forward monocular distance measuring using ranging principle of geometric constraints combined with the characteristics of the docking system, that is fitting ellipse according to the extracted contour, obtaining the parameters such as angle and distance by the elliptical. The algorithm has higher accuracy and feasibility.Docking robot control strategy, divided into the two stages:the docking station search and docking robot moves. In the docking station search stage, using object recognition algorithms to search the docking station mark in robot vision; in the docking robot moves stage, using the control algorithm brined forward in this paper, that the robot vision divided into three zones, to adjusted the position of the robot according to the location information of the mark, determine whether successful docking according to the angle and distance of the docking robot.