Research On Control Method Of Robot Arm Based On Visual Servo

On modern production lines,workers' jobs are gradually replaced by robotic arms.Because the robotic arm has the characteristics of high precision,high efficiency,and high reliability.The visual servo technology of machine vision and robot arm control provides a feasible solution for the robot arm to track and grab the target workpiece.However,the current development of visual servo control is immature,which mainly represents the accuracy and real-time performance of workpiece tracking and grabbing.In order to solve the problems of unclear target edge segmentation and insufficient tracking speed of planar moving workpieces,the simulation modeling,kinematics analysis,target contour segmentation and tracking capture of visual servo manipulators have been studied.Because the motion state of the end effector of the visual servo manipulator is difficult to determine,the Baxter simulation model is established in the Robot Operating System(ROS).The mathematical model of the manipulator is established using the Denavit-Hartenberg(D-H)method,which obtains the position and posture coordinates of the end effector and the kinematics positive solution of the manipulator.The joint angle of the manipulator is calculated using the pose separation method,which obtains the inverse kinematics solution.For the problem of unclear target edge segmentation in visual servo manipulator,this paper proposes an adaptive threshold edge segmentation algorithm.This algorithm studies the problem of sensitivity to noise and the need for manual threshold calibration in edge segmentation algorithms.When segmenting the contour edge,the average weighted template is used for iterative convolution,and the filter kernel is optimized and the problem of noise sensitivity is solved.The optimal threshold is obtained by traversing the variance between groups,which realizes the automatic selection of the threshold and overcomes the error introduced by manual selection of high and low thresholds.Experimental results show that the improved algorithm has significantly sharpened edges compared to the original algorithm.The target of the tracking algorithm is investigated,and a tracking algorithm based on extending Kalman particle filtering is proposed,which improves the real-time and stability of the target tracking.The weighted particles and variance obtained in the initial stage of the particle filtering algorithm and then passing to the extended Kalman filter equation.Within the equation,the probability density function is calculated in the state space according to Monte Carlo method for the fitted state vector of the weighted particles.With the observations value,the result is updated and corrected.The weighed addition is used as a new estimate and the variance continues to be passed into the particles filter algorithm to estimate the position and velocity of the target.Through repeated experiments,the results show that the extended Kalman particle filter algorithm can increase the target tracking speed by 48.8% on average.