Research On Autonomous Object Recognition And Grasping Technology Of Robotic Arm

With the development of science and technology,a variety of intelligent devices have entered into the lives of ordinary families,which bring a lot of convenience and fun to our daily life.With the beginning of the aging of China's population,a major problem in today's society is how to solve the pension problem of many old people in the future;At the same time,in the medical process,the daily care of inpatients also brings a lot of pressure to the work of medical staff,and many simple repetitive work often takes up a lot of people's time and energy.With the continuous development of robotics,it has become possible to use robots to provide daily care and care for the disabled and semi-disabled elderly and patients.In the daily service provided by the robot,it is the most frequently used function to identify and grab the target objects needed by the user.Therefore,it is of great academic value and practical significance to study the effective recognition and grasping technology of the target objects by the robotic arm.In this paper,the autonomous object recognition and grasping technology of the robot arm is studied in sections.The research results are as follows:According to the specific working environment of the grasping platform,an effective and feasible solution is designed to realize the function of autonomous object recognition and grasping.In view of the research progress at home and abroad,this paper analyzes the advantages of the methods of object recognition technology based on vision sensor and the excellent research results and efficient scheme of object grasping motion planning technology of manipulator under special environment.The object recognition and grasping system can be divided into the object recognition and coordinate calculation and positioning module and the object grasping module to realize the autonomous recognition and grasping of the object by the manipulator in the working environment.In order to realize autonomous object recognition of robots,this paper designs and implements an object recognition algorithm based on BOW+SVM framework.The unsupervised learning method is used to construct the recognition model of the target object,and then the feature points of the recognition model are matched with the environmental image.Finally,the target object is determined and selected in the box.Firstly,SURF algorithm is used to extract the feature points of the target object,and then FLANN algorithm,feature point secondary screening algorithm and RANSAC algorithm are combined to identify and select the target object.Then,the calibrated depth camera is used to locate the coordinates of the target object,and the color-depth coordinate algorithm is designed.Aiming at the problems existing in the color depth coordinate mapping and coordinate acquisition of RGB-D sensor,the cross mean filter algorithm is used to improve the success rate and accuracy of information extraction,provide effective data support for the following fetching process.In order to realize the robot's flexible grasp of the target object,the ROS software environment on the Linux system was used to build the grab platform with the software's built-in Moveit framework,and the smooth grasp of the target object was designed and simulated.The URDF framework is used to construct the kinematics and dynamics models of the mechanical arm and manipulator used for grasping.The framework can better reflect the movement state of the actual work of the mechanical arm,which is helpful for better motion planning of the grasping system.With the help of the information subscription and publishing links of ROS system,real-time force feedback information is provided to the control system in the flexible grasping work.According to the feedback information,the wrong grasping action can be corrected in time to ensure the smooth progress of grasping activities of the grasping platform and target objects.