Research Of Robot Automatic Returning For Charging Based On Monocular Vision

With the quick development of technology, robotic is playing a more and moreimportant role in human society. Power technique is a key technique to robotic and isthe assurance for achieving a long-time autonomous working. Because of the capacitylimitation, the mobile power is not the best solution to robot power problem. So far, themain technological means is automatic returning for charging guided by the infrared ray.But the short points of infrared ray are its transmission distance is very short and cannot break through obstacles while can be disturbed by electromagnetic interferencewhich will affect the success ratio and scope of application of robot automatic returningfor charging. Thus it has been a hotspot robot technology to solve the key technicalproblems in robot automatic returning for charging.Aiming at the power technique problems that appear in resent mobile robotics, thispaper studied the key techniques of mobile robot automatic returning for charging. Thecontents of research are as follows:①Calibrating camera with Zhang Zhengyou calibration method. Designing thecharging-dock based on the color and shape characters for experiment.②Aiming at the remote-docking problems, a remote-docking algorithm based onmonocular vision and speedometer is put forward. A door beam model is created for theuse of robot self-localization based on monocular vision. And a room map model isdesigned for robot’s localization using speedometer. A motion control strategy used forimproving the efficiency and accuracy of robotic remote-docking.③An algorithm for recognizing charging-dock based on color character in HSVspace and shape character is proposed. CamShift algorithm is used for tracking thecharging-dock. The charging-dock’s hue color histogram is created in the way of offlinetraining which is used for color objective segmentation. Then the charging-dock’s shapemodel is created using body gesture proportion which is used for charging-dockrecognition.④To meet the needs in the process of tracking charging-dock, a monocular visiondistance measuring algorithm based on aperture imaging principle. Experiments arecarried out on the Pioneer3-DX mobile robot platform based on OpenCV. Theexperiment results show that the algorithm proposed by this paper is adapt to lightchanging to a certain degree while can clear the similar color and shape objects’ interfering. This paper’s algorithm also can track the charging-dock steadily and has ahigh success ratio of automatic returning for charging.