Vision-based Self-correction Control Of Robotic Arms

With the development of science and technology, robots being the important partshave been widely applied to various industries to replace humans performing variouscomplex tasks. Robots’ various applying is significant in improving the productiveenvironment, reducing costs, increasing productivity, improving product quality, allocatinglabor resources rationally, shifting from the labor-intensive enterprise to science andtechnology industries.Most of the traditional welding robot uses "Teach–by-show" control mode, but thiscontrol mode lacks of flexibility and has strict requirements of workpieces and installationpattern. It is very essential to improve the control mode of welding robots for the improvedproduction requirements of versatility and intelligence. Visual servo robots obtain externalinformation of work environment from the vision sensor and send to the controller,according to the information the robots adjusts Pose in real time achieving accurate trackingand locating of the target. Visual servo robots compared with conventional robot controlmethod have higher accuracy and flexibility. It is a trend that robots are more intelligentand more adaptable to the external environment.This thesis throws detailed study on a six degrees of freedom control system namedPUMA560and comprehensive analysis on the development of visual servo robot controlsystem. According to the characteristics and performance requirements of this issue, thethesis designs eye-in-hand system of monocular vision、image-based visual servo controlmode which has double-loop control structure taking visual servo system as the outer ringand joint servo control as the inner ring. Also it proposes a servo prediction controlalgorithm based in image model, taking the image coordinates of the target feature pointsas the visual feature and considering joint constraints、speed restrictions and field ofview constraints in the system, performing predictive control method in limited scope,using dynamic online optimization of linear matrix inequality resulting optimal、 stronganti-disturbance capacity visual servo control signals.The thesis does some simulation on the proposed algorithm and some comparison ofthe effects of different constraints on the control system, proving the feasibility of visualpredictive control algorithm. Using the two degrees of freedom platform of the laboratory,we test the image-based visual servo control algorithm, successly achieving error correctingand self-tracking control of the weld, proving that the image-based visual servo controlsystem practical and effective.