Research On Visual Servo System Of 3-degree Shuttlecock Robot

With the rapid development of machine vision and hardware,the visual servo system is the combination between visual information and robot control,which made robot have the intelligent swithing capacity with the external environment.This is one of the main direction of developing of robot.Visual servoing system through the detection of the object image get visual information in order to control the movement of robot,work spare can be achived and objectives in Cartesian coordinates in spare relative to the location of cameras. Visual tracking which based on computer image processing and analysis or comprehend of dynamic image become concern topic.Image processing confront the complexity of the challenges.Parallel processing technology play a important role in this.thereby,robot visual servo control is a fundamental research subject with very important theoretical research sence and industry application prospect.Visual servo system of robot is a complicated domain that is concerned with computer vision, kinematics, image processing, dynamics, control science and real-time calculation. This paper studies the visual servo of robot and and builds a 3-degree robotic arm experiment system.Firstly, the paper analysed the advantages and disadvantages of various robotic visual servo systems.And according to the characteristics of the experiment system,a kind of two-closed-loop structure that includes a visua controller as the outside loop and a joint servo controller as the inside loop is chosen.Accordingly,we adopt image-based visual control policy in this paper.The visual subsystem consists of a 1394 camera and a personal computer.The software programs are designed for image collection, image emphasizing, three-dimensional image restruction and characteristics extracting on OPENCV platform. And the characteristics vectors are transmitted from image processing computer to control computer through serial port. On the other hand, I/O board PC connects sensors, joint controller of robot and the control computer. A real-time windows target emulator is built on OpenCV in order to drive the robotic arm to he desired position.The core of Tray Loader control System in servo part is Mitsubishi Q02HCPU, which achieved the logic sequence control of the whole system; Mitsubishi Q172CPU used as a movement control CPU of the system to achieve the targeted control; SSCNET as control network of servo system, it achieved the controller and servo amplifier large amounts of data in the real-time communication. With a high degree of synchronization of serial communications makes a simultaneous multi-axis control, interpolation, and other senior targeting to improve the system's response and reliability. In this thesis, GT software as a development platform, GOT985 as a hardware device, it was designed the HMI monitoring program of Tray Loader control System to achieve the real-time monitoring and remote debugging of system.At present, all the design and debugging work of both the joint servo control and visual subsystem are completed.The experiment resulats is running high accuracy, reliability, fast response.These results for position control experiments indicate that the system performance is good. It is practicable in the package assembly and transit domain.