Image-based Robot Rubust Control With The Uncertain Parameters

The robot technology is one of the greatest inventions in the20th century. It has changed the traditional industrial mode of production and made a huge impact on human’s life and work. With the development of science and technology, people are not satisfied with traditional robots gradually and in favor of intelligent robot with feelings. Governments also began to list it in the high-tech plan. Among all the sensors which can be used in intelligent robot, vision sensor stands out by its great amount of informations, non-contact measurement methods and so on. Its appearance increased the adaptability of the robot and broaden the fields of application of the robot. However, accurate calibration of the system is not easy in practical applications, so the visual servo method was subjected to many restrictions. Therefore, uncalibrated visual servo system came into being, and become a research focus. It considers the case of no pre-calibration of the robot and camera parameters, and directly design a controller based on the obtained image errors to drive the robot, and make it convergence. This article is a study for this.The main work of this paper includes the following aspects:(1) Summarized the visual servo control research’s background, development status and significance, as well as the background and main content of the uncalibrated visual servo tracking control.(2) Described the overall structure and properties of the robotic model and the camera model. Derived the robot’s Jacobian matrix, image’s Jacobian matrix and the extended Jacobian matrix in detail.(3) A neural network-sliding mode controller was developed for the case of uncalibrated robot model and the existence of dead-zone input. It enhanced the robustness of the system, achieved the purpose of tracking the target object. Finally, the result was simulated by computer.(4) An adaptive sliding mode controller was developed for the case of uncalibrated robot model and camera model. It reduced the restrictions for asymptotic stability, solved the object tracking problem with the uncalibrated mode in three-dimensional space. At last, the tracking results were simulated by computer.