Research Of Robotic Arm Hand-eye Coordination System Based On Image

Robot has been experience the development from simple to complex, from single to diversity and from low to high level since its appearance. Especially along with the development of the electronic technology, the processing capability of computer and sensor technology has been highly increased, people are no longer satisfied with the robots which operate in a specific mode in a certain circumstances. Various sensors are added to improve its application and performance. Among the sensors, visual feedback is a leap in robot technology because it can provide robot with rich information in a non-contact way. The coordinated control stategy between robot vision sensor and actuator has a broad prospect of application and thus become a hotspot among the many research of machine vision.The rest equipments except the main body of the RM-501 mechanical arm in laboratory are damaged. The main aim of this paper is to design a control system and add in visual inspection system for the robot to provide robot with hand-eye coordination function, as to turn it into an open platform for teaching demonstration.This paper firstly introduces the background and development process of the robot hand-eye coordination in detail. Through analyzing its characteristics, advantages, disadvantages and applicable conditions, and combined with the characteristics and requirements of the project, this paper adopts monocular, eye in hand and image-based visual control method. The general eye-hand coordination system of the robotic arm which consists of joint drive control system, visual inspection system and visual controller, is then given, before the detailed design is carried out. The PWM method is used in joint drive control system which adopts the outer position loop and inner rate loop for the double closed-loop structure. The background subtraction method is used in the visual inspection system to achieve target detection. Simulation under MATLAB is carried out to verify the algorithm before transform it to the ADSP Blackfin 533 platform. Image feature vector is exported to host computer through D/A. The design of visual controller is based on image Jacobian matrix model. Its main task is to build up the relationship between image feature vector and robotic task space.Finally, the system integrating and test is carried out. The controller is built in the Simulink/RTW in the host computer, other subsystems are real-time connected to the host computer through PCI1711 board. The robotic arm's hand- eye coordination is finally realized.The designs and debuggings of subsystems are done and the system integrating and test is carried out. The results are ideal. However, the positioning and tracking precision need further studies.