Manipulator Motion Planning Algorithm Based On Monocular Vision Estimation

In recent years,China has issued a number of policies on industrial intelligent manufacturing.Intelligent manufacturing is intended to specifically integrate informationization and industrialization,which is an important direction for the development and transformation of China's manufacturing industry in the future.Machine vision and mechanical arm grabbing technology are widely used in the field of industrial automation production,providing many conveniences for manufacturing production.However,there are many uncertainties in the placement of objects in different automated production environments.At present,in the process of industrial production,most of the industrial mechanical arms grasp and place the workpiece according to some fixed instructions,which have certain limitations.This kind of working mode needs to make the workpiece in a fixed position and grasp an object in a fixed posture.Once the environment and the object posture change,the mechanical arm will not be able to grasp and lead to task interruption or failure.Machine vision has some unique advantages,such as large information volumn,large measurement range,high precision and many other advantages.It is widely introduced into the process of industrial automation,and it is used as the brain and eyes of the robot arm to realize the automatic decision-making and grasping task.Therefore,this paper will design a robot arm grasping system dealing with the object posture estimation of monocular vision,the path planning of the robot arm based on particle swarm optimization algorithm,and the grasping strategy of any position and posture.The structure of this paper is as follows:First of all,in view of the complex industrial production scene,it is difficult to separate the target object and the complex background information.This paper proposes a method of extracting the object feature information by using the image preprocessing algorithm,analyzes and studies the method of eliminating the complex background.After that,the target object area is obtained.,the background and the object are separated,and then the acquired image is analyzed in detail.It lays a foundation for the following object position and posture acquisition.After image preprocessing,the processed image with the target object is analyzed based on the algorithm of machine vision.Using the known information of the target object and the camera model,amodel-based method of object posture estimation is proposed.This method overcomes some shortcomings of the environmental factors that have a great influence on the acquisition of the object's pose,and improves the measurement accuracy of the object's pose in the static complex background.In addition,the whole scheme is simple and effective by using the geometric characteristics of the target set.Acquiring the target position and posture makes a good preparation for the subsequent manipulator grasping.Aiming at the problem of manipulator motion planning under complex constraints,a novel multi-population particle swarm optimization based on elite group(PSOEL)is proposed to solve the problem of five degree of freedom(DOF)manipulator motion path planning.PSOEL is composed of several sub populations and an elite population.Using the pre-selection and cross mechanism,PSOEL effectively jumps out of the local optimum,and approach or reach the global optimum infinitely.This multi-swarm PSO algorithm has fast convergence speed and overcomes the shortcomings of the traditional PSO algorithm,such as easily falling into the local optimum,slow convergence speed and low efficiency.Compared with the traditional PSO algorithm,the results show that PSOEL is the best in path planning.In order to realize that the end effector of the manipulator grasps the object with any pose,an algorithm based on azimuth is proposed.According to the orientation information of the target object in the basic coordinate system of the manipulator,the pre-grasping posture can be achieved in advance,which can effectively prevent the manipulator from entering the deadlock state,adjust the posture of the actuator at the end of the manipulator according to the posture of the object,complete the object grabbing task in a gradual way,and solve the problem of arbitrary posture grabbing of the object.This paper uses MATLAB to simulate the MDH model of the 5-DOF manipulator to verify the correctness of the model.MATLAB simulation experiments were carried out on the angles at which each joint axis evolved by various group evolution algorithms should be rotated.At the same time,the actual manipulator gripping experiments were carried out,and the experimental comparisons were carried out.The results proved that the robotic arms could be grasped autonomously.