Mobile Robot Control Based On Visual Feedback

During the process of exploitation of the sea, underground resources and astrospace, Tasks are often carried out in the limit working environment which is difficult or hazardous for mankind to approach such as high temperature, high pressure, strong radiation, suffocation etc. In order to complete the complex operations in such harsh circumstances, the teleoperation construction robot technology based on the telepresence has been used widely. Due to the interference of the cosmic rays and the limitation of the transfer condition, the signals being sent back are always with noises and remarkable length of time-delay, the system will be unable to work normally in severe situation. In recent years, the teleoperation construction robot system has autonomous function has been asked for. When the transmission lines are disrupted and teleoperation can not be carried out, the autonomous task system still can complete the key task which must be completed in the part-time independent nodes automatically. Such research and development have great theoretical significance practical value. This dissertation combines investigation on domestic and international research on robot stereo vision technique, relies on the experiment platform of sic degree of freedom teleoperation hydraulic-servo manipulator in laboratory, sets up a complete" Autonomous Task Control System of Mobile Robot Based on Vision Feedback", and researches and develops the autonomous task control system of the construction robot. In order to realize the Autonomous Task of construction robot, the dissertation firstly studied the recognition of target object and space location technology. The calculation method of the three-dimensional space position is studied by installing Stereo camera and collecting two-dimensional images of target object. Which include camera calibration, image preprocessing, stereo matching etc, for eradicate the camera distortion. After gaining the position information in three-dimensional space of the target object, three-dimensional reconstruction of the target object technology is researched. During the process of the camera calibration, neural network technology is applied, compared to traditional camera calibration technology, nonlinear modeling of the cameras is avoided, camera calibration accuracy is enhanced and more accurate depth information of the target object is obtained. In order to carry out the autonomous control, the manipulator kinematics modeling, manipulator trajectory planning and control are researched. The main research works have been accomplished as follow:(1)The teleoperation construction robot was taken as the study object, stereo vision is applied into the teleoperation construction robot field for the first time, and the teleoperation construction robot autonomous task experimental system based on stereo vision technology are structured. And by successful trajectory planning and target grasp detection and control, the autonomous task of construction robot is preliminarily realized.(2)Image matching algorithm based on the edge is presented, the geometric model on the three eyes camera depth calculation is built for the first time by using space analytic geometry and image superposition, and the 3D reconstruction of the target object is proceeded.Research works show that the autonomous task control system of the construction robot based on the stereo vision technology have preliminarily been able to complete their task, and the task process is smooth. The research work done in this article provided valuable experience and success examples for the deeper autonomous task technology of the teleoperation robot in the future.