Research on Visual Servo System of Mechanical ARM
|School||Harbin Institute of Technology|
|Keywords||Visual servo Mechanical arm Digital image processing BP neural network|
It is an important developing direction that the robot is accompanied by vision so that it can interact with its working environment. It was researched early in foreign countries, but stayed in junior phase in our country wholes many studies just forcus on simulation concepts, reasoning method and the trial of basic components and key technologies, so it is meaningful to develop these details independently.The visual servo belongs to the complex misalignment control, which involves to many disciplines, including image processing, dynamics, kinematics, control theory, real-time computing, etc. This paper proposed scheme which based on the image visual servo. This plan has avoided to system’s precise demarcation process, reducing system hardware’s request, and has carried on the related experiment.The mechanical arm system adopts PWM mode to actuate DC servo motor of each joint, constructing driver board by unifying the spacing protection circuit and detecting & processing circuits. In this paper proposed implementation scheme of the visual information processing system. Processing image by using related algorithm based on space domain, extracting image characteristic information, forming the visual servo circulation. We design vision servo controller by choosing the neural network. The method avoided difficult operations when calculating Jacobian matrix and the problem of matrix singularity. The gathering and training method of sample in neural networks is represented. The paper builds real-time simulation system based on RTW through the PCI-1711 data acquisition card, and puts forward a testing scheme for motor model. The servo controller of the joints, having two closed loops which including a position closed loop and a velocity closed loop, is designed in this paper. The double closed loop real-time simulation system based on motor model is established, and the debugging work of the joint servo controller is completed.The master control computer and the imagery processing computer correspondence is realized through the serial port. Finally, the corresponding experiment has been carried on.