Research On Robot Visual Servoing Based On High-speed Visual Feedback

There have been more and more cases of integrating vision technology into robot systems.Based on the background of the robot system,this dissertation studies the field of vision processing and industrial robot.Through the establishment of the robot visual servo model,the hardware platform of the high-speed vision system,and the design of the robot visual servo control algorithm,a complete set of robot visual servo system based on high-speed image feedback has been developed.Firstly,Analyzing and deriving a system model.This system uses the laboratory's six-axis Efort industrial robot as a research platform,uses the kinematic modeling method based on the exponential product formula to establish the robot's positive kinematics model,and then uses the relationship between the twist and the pose matrix to derive the robot's Jacobian matrix and the relationship between the robot end-effector and the camera velocity.Combined with the mathematical model of the camera,appropriate image features are selected to establish a model of the robot visual servo system based on image feedback.Thus,the relationship between the feature speed in the image and the camera speed of the robot end can be established.Through the relationship between differential kinematics and speed conversion,the relationship between each coordinate system in the visual servo system is obtained,so that the speed in the camera coordinate system at the end of the robot arm is converted into the motion speed of each joint of the robot.Secondly,Building the robot visual servo high-speed image system.An Eo Sens MC1362 high-speed industrial camera and Xilinx Kinte X Ultra Scale+ series of highperformance FPGA development kits are used to build an image processing system,and image data transmission is performed using the Full mode of the Camera Link protocol.Based on the hardware acceleration of image processing based on SDSo C,the tasks are pipelined through Pipeline,Unroll and Dataflow instructions provided by HLS,and the line cache and window cache strategies are used to accelerate the convolution operation in image processing.Use the Ar Uco marker as the target image,find the outer contour of the Ar Uco marker through the 4\8-area contour tracking method;find the four corner points of the outer contour through the change of the contour angle;use the four corners of the outer contour At this point,a pose estimation algorithm that is easy to implementin FPGA is given,and the phase pose with theoretical error of 0 is obtained.In addition,an outer contour detection method based on historical information is proposed to speed up image processing.Finally,Analyzing the visual servo system and designing the visual servo controller.By separately analyzing the position-based visual servo system and the image-based visual servo system,the visual servo controller under high-speed image feedback is designed,and the stability of the classic control algorithm is analyzed using the Lyapunov function.The shortcomings of the classic algorithm are verified by simulation.The speed of the camera motion is compensated for feedforward combined with the characteristics of high-speed image feedback,and the superiority of the method is verified by simulation.Analyzing the feedback delay of the image feature information of the visual servo system,give the calculation method of the delay,study the delay problem in the target positioning and target tracking control,and give an estimation and compensation method for the visual characteristics.The simulation verifies that the method can Improve the performance of visual servo system under delay The visual servo positioning experiment of C60 robot verified the correctness of the method.