Irregular Object Grasping Technology Based On Multi-claw Manipulator

With the development of economy and the progress of science and technology,robots are playing an increasingly important role in industrial production and technological research and development.Manipulator is a very common robot,similar to the shape of the human hand,has a strong flexibility,is widely used.This paper mainly studies how to use the multi-claw manipulator to grasp the irregularly shaped target objects in the indoor environment.The research content includes the construction and navigation of the robot moving platform,the detection and grasp of irregular objects and other technologies.Based on the research of manipulator grasping and robot map building navigation,this paper analyzes the experimental environment and the basic requirements of the experiment,and chooses Turtlebot2.0 Lidar and xtion Pro live camera are built on the robot mobile platform,and this set of device is used to complete the task of sensing environment and detecting irregular objects.use uHand2.0 The multi claw manipulator is connected with PC through VNC,which can control the multi claw manipulator through PC,and then control the multi claw manipulator to grasp irregular objects.This paper studies and analyzes the working mechanism of the robot operating system ROS,the cooperation between the ROS system and the turnlebot2.0 robot mobile platform,and the use of ROS to control the robot mobile platform to complete the task of map building and navigation.In the ROS system,the gmapping SLAM algorithm package and navigation navigation package are implemented,and the indoor environment of the robot mobile platform is mapped.Through AMCL positioning and A* algorithm,the autonomous positioning and navigation of the robot mobile platform is realized,and the experimental error is within 12 cm.The deep learning,convolution neural network and deep learning based target detection algorithm are studied.The fast r-cnn target detection algorithm with the best performance andthe fastest speed is selected.Through the communication between tensorflow and ROS,and using fast r-cnn algorithm to complete the detection of irregular objects.Through the research on the position of irregular objects,the detection algorithm is used to find the points of irregular objects.Finally,rassberry Pi is used to control the multi claw manipulator to grasp the irregular object successfully.Many experiments can successfully grasp irregular objects,and the success rate of the experiment reaches 100%.The task of grasping irregular objects by multi claw manipulator has been completed.