Object Recognition And Manipulator Grasping Based On Stereo Vision

The 21 st century is an era of intelligent development.With the birth of the first industrial robot in the early 1960 s,the dream of people chasing robots to replace heavy labor has not stopped.At present,robots are widely used in pharmaceutical,aerospace,mining,medical,family services and other fields.However,most of these robots can only accomplish some simple tasks,so how to improve the intelligent degree of robot becomes a hot research direction.Grasping objects is one of the important abilities of intelligent robots.To a certain extent,their ability determines the degree of intelligence of robots.Based on this,this paper will take the vision-based object attitude estimation and the grasping of the manipulator as the breakthrough point to improve the autonomous grasping ability of the manipulator and study the grasping problem of the intelligent robot.Firstly,based on machine vision algorithm,a pose recognition method is proposed.This method based on the geometry of the target object and the imaging principle of the camera,combined with the model-based attitude estimation method,to achieve the attitude of the target object.This method can improve the shortcomings of the surrounding environment change on the recognition result under visual conditions,and the visual sensor used is low in cost.By setting the target as the geometric feature of the target object,the whole method has strong robustness and is simple and effective.In addition,the posture transformation of the target object in the robotic arm space is completed by rotation transformation,which lays the foundation for the next robotic arm grabbing.Secondly,based on recognizing the attitude of the target object,a robot arm grabbing strategy is proposed based on how to complete the grabbing of the target object according to the method of latitude and longitude positioning.The strategy can complete the pre-pickup action according to the azimuth angle of the target object in the coordinate system of the robot arm,thereby reducing the chance of being caught in a deadlock.At the same time,it can complete the adjustment of the posture of the manipulator according to the posture of the object,and complete the grasping approaching the object in a progressive manner.Finally,the establishment of ROS experimental platform combined with ZED binocular camera and Kinova robotic arm in the experimental environment to verify the design method is correct and effective.