The Research On Visual Servo System Of Autonomous Robot Based On Image

With the rapid development of robot kinematics, control theory, the technology of image process, computer hardware and camera, visual servo-control system gradually becomes the hot topic in the field of the robot research. Especially, the robot vision is made combined with the neural network, intelligence control and information fusion. There is a great progress in the theory and the practice.The research mainly has two parts: the extracting target's system based on multi-feature fusion and the servo-control method based on fuzzy control.In the extracting target's system based on multi-feature fusion, firstly, this paper adapts the dividing method based on hue and saturation, and this method can show its great advantage in the great illumination condition. This method also makes the foundation of the exactly recognizing the target. Through the way, it can gain the range of the target. If there is interference in the range, it can apply the Sobel edge-detection method to pick up the outline of the target in the shorting researching range. Secondly, through the comparison of the outline, it can get rid of the other objects having the same color with our target, and it can confine our following target. Finally, the run-length coding is used in the image fusion process to ensure the real-time of the system. In the process of image fusion, it also acquires the target's pix area information and the gravity coordinate information. So it provides the original data for the autonomous robot following.In the control field of the autonomous robot, it can apply the fuzzy control which should be similar to the human thought to complete the autonomous robot following programming of the target. The double input signals which contain the different value information of pix area and coordinate and the double output signals which contain the left wheel and right wheel form negative feedback circle control system to effectively control the robot's following of the target.Above all, in the platform of the MT-R robot, a lot of related experiments are made to prove the practicability of the theory which is the research on visual servo system of autonomous robot based on image. The results of the experiments show that the target can be identified in the complex environment, and the autonomous robot can track the target in real time.