Research On The Application Of Object Detection And Manipulator Grasping Technology Based On Depth Camera

Ultrasonic honing technology is gradually applied in various fields,and the output of ultrasonic honing device is increasing year by year.Aiming at a series of problems such as low assembly efficiency,high labor intensity,hidden safety risks and the parts’ bumps and bruises in the existing production and assembly process of its components,a mechanical arm grasping system based on depth camera was designed.Based on depth camera and target detection algorithm,this paper carried out the application research on the grasping technology of sixdegree-of-freedom mechanical arm.The research content of this paper includes the following:(1)D-H coordinate system of SNR3-C30 manipulator was established by D-H method,and forward kinematics and inverse kinematics models were established based on it.Matlab Robotics was used to simulate and verify the derived forward and inverse kinematics models,and Monte Carlo method was used to analyze the manipulator workspace.(2)Build the robot arm simulation platform based on ROS.The 3D model of the robot arm was established with Solid Works,converted into URDF format and imported into Rviz.The motion planning of the robot arm was carried out with Move It in ROS,and the grasping program of the robot arm was written to complete the obstacy-avoiding grasping of the robot arm in Gazebo simulation environment.(3)Design target detection and positioning system based on depth camera.The network structure of YOLOv5 target detection is introduced.A total of 4800 images of eight kinds of ultrasonic honing device parts are collected independently to make data sets.YOLOv5 was used to train the data set,and the analysis of the training results showed that the m AP value of YOLOv5 was 95.38% and the detection speed was 26 frames /s.Finally,the depth value of the object to be grasped(the ultrasonic honing device component)was obtained by the depth camera,and the error of the obtained depth value was tested,and the final error was 2.12%.(4)The internal parameters of the depth camera and radial tangential distortion coefficient were obtained by Zhang Zhengyou calibration method,and the transformation matrix of object coordinates from the camera coordinate system to the base coordinate system of the manipulator was obtained by eye-to-hand hand-to-hand calibration.The coordinate position of the target object was transmitted to the manipulator control cabinet according to Socket communication.On the basis of the above work,a mechanical arm grasping experiment platform based on depth camera was built,and the grasping experiment of mechanical arm was carried out.The grasping success rate was 87.5%.