Study Of Manipulator Motion Control Based On Binocular Vision

Robot vision technique has got much attention and great development since it came out, and now is widely used in many fields such as industry, agriculture, medical, services, and aerospace. But with the development and application, a higher demand has been put forward for the accuracy, speed and stability of visual servo robot systems, which also leave much room for improvement to real-time robot vision control technique.According to the state and trend, the research takes binocular visual servo 6-DOF (degrees of freedom) manipulator control system as the object, and carries on deep study and analysis on kinematic analysis theory, trajectory planning theory and binocular vision theory. It focuses on stereo matching, the most difficult part of binocular vision, and make the research and improvement mainly on the theory.The paper completed the following five researches and designs. Firstly a D-H model is built for the 6-DOF manipulator based on the analysis and description, in addition, the kinematics and inverse kinematics solution are also worked out. Secondly the research makes a study of trajectory planning both in joint coordinates and in Cartesian coordinates, and also realizes a simulation for the planning algorithms on MATLAB. Thirdly a thorough analysis of binocular vision is carried out, especially for dense stereo matching. In this analysis an improved region growing algorithm combined with pre-screened adaptive Shi-Tomasi algorithm is proposed and verified. Fourthly a binocular visual servo manipulator real-time control system is designed based on LabVIEW and FPGA. At last, practicality experiments and result analysis are conducted to test the system performance of the system and verify the research of the theories.The test and analysis indicates that the system can accurately sense the change of working environment using binocular vision technique, and can plan and control the trajectory of the manipulator based on the change. Besides, the results verify the improvement of the dense stereo matching algorithm in this paper, which makes up for the deficiencies of traditional algorithms to some extent, both in accuracy and speed.