Based On Visual Servo System Of Six Degrees Of Freedom Manipulator

Information obtained by the human brain from the visual perception of the outside world. As the trend of the fast growing artificial intelligence, the strengthen vision of the robotics is very important. Robot visual servo technology is in the fast development nowadays, unlike conventional robot control technology, by introducing visual information and feedback to the robot control system for controlling the movement of the robot to improve environmental adaptability. Without visual servo calibration, attracts more people's attention and research because of its good adaptability. In this paper, I use arm PUMA560 as the body of the robot servo motion systems using the TMS320DM6437 processor.It is the TI's Da Vinci platform which is dedicated to visual processing, constructed the uncalibrated visual servo control system through the effective handling of visual information collected, combined the robot motion with visual perception perfectly.In the manipulator joint movement control section, first ly I established PUMA560 arm joint model, and established a dual-loop control of speed and position, verified by simulation. Then establish ed manipulator forward and inverse kinematics model based on the coordinate system and improved linkage DH method.In the visual processing portion processing, platform used in this paper is Leonardo da Vinci DM6437, for the acquisition of image acquisition, pre-processing operations such as Gray, on this basis, after a lot of simulation, and finally use the Hough transform moving target feature extractio n and centrally located. At the same time, through the study of the program, the source code, mainly in C language, optimize, and by optimizing compiler settings, greatly reducing processing time of the algorithm for each frame of the image and improve the system in real time.Finally, in the design of the uncalibrated visual servo control system, first ly I do the trajectory planning in image space, and then by learning and training of the variable step size BP neural network, and finally the next time the target speed estimated value is mapped to machinery arm end of the robot arm through simulation, can achieve the purpose of the drive arm to achieve trajectory tracking tasks in accordance with the target object.