Design Of The Visual Servo System Of 5-dof Manipulator Rm-501

With the development of science and technology, many kinds of robots are widely used in various fields to complete different complex tasks. Robots have become an important part of the automation production. Adding visual feedback into servo system is considered a leap in the progress of robot. Manipulator's visual feedback which is considered the most important sensor of robot can be compared to human's eyes. It is a non-touch measure having wide signal range and integrated information. Furthermore, manipulator's visual feedback brings more interaction between robots and environment. Hence, forming a position closed loop control system of robot by introducing visual feedback into robot position and movement control systems to provide control signal as soon possible after fast image processing of the acquired images is one of the favors for most researchers.Visual servo which is a complex non-linear control involves digital image processing, kinematics, dynamics and control theory. This paper studies a control system of robot RM-501, a 5 degree-of-freedom manipulator. Several broken circuits are fixed and visual feedback is added in to design an open-ended and interactive robot visual servo system.This paper first presents comprehensive information of the development of robot and visual servo. After analyzing the advantages and shortcomings of various robot visual servo systems, this paper adopt the visual servo of"eye-in-hand, based on image"concept according to the project's features and performance requirements. The overall system is a dual closed loop structure which the outer loop is visual servo and while the inner loop is joint servo. This paper designs a visual controller using BP neural network and maps the target information of image space to the robot coordinates. Robot joint servo system is a dual closed loop structure with position loop and velocity loop. Each DC motor of each joint is driven by PWM rectifier. Visual subsystem completes acquisition, pre-process and feature extraction of images under Visual DSP++ using BF533 experimental platform, and then transmit characteristic vectors of images to master computer through serial port. On the other hand, the A/D input of sensor signal in robot and the D/A output control signal of joint controller are implemented using PCI1711 data acquisition board. RTW simulation system is established under Simulink to enable real-time communication between controllers and sensors, and finally complete the real-time control of the manipulators.The design and debugging of the joint controller and visual subsystem are done by far. Overall debugging of the joint servo system of the manipulator is implemented and the experimental results meet the design requirements. However, further research is need for the control accuracy and visual servo.