Research On Trajectory Planning Of Manipulator Based On Machine Learning

The research on the trajectory planning of manipulator based on machine learning is based on the project of intelligent plastering robot,and purpose of the research is to construct a manipulator working system of the intelligent plastering robot under the working conditions of the external angle,door and window.In this paper,the related algorithms in binocular vision,machine learning and multi-DOF manipulator are studied and improved to complete the development of the working system.The main contents of this paper are as follows:1.In this paper,the key technologies of binocular vision 3D reconstruction are studied,and a layered matching binocular vision system is designed.In this paper,through the comparison test of multiple feature point detection algorithms,the SURF algorithm is selected to extract and describe the image feature points.According to the matching environment characteristics of binocular vision,the partitioning strategy is adopted to quickly approximate the nearest neighbor matching algorithm FLANN,the matching efficiency have been improved.Finally,stereoscopic display is performed using triangulation and texture mapping.2.In this paper,the related theory of neural network is studied,and a network model for pose estimation of manipulator is designed in combination with binocular vision.According to the corresponding relationship between the camera pose of the acquisition image and the point cloud coordinates after 3D reconstruction,the acquisition and processing methods of the neural network training samples is planned.A network model for estimating camera pose is trained using the calibrated sample data.3.In this paper,the kinematics control technology of multi-DOF manipulator is studied,and a six-DOF manipulator kinematics control model is designed.In this paper,the Danish six-DOF manipulator UR10 is used as the simulation target,and its D-H parameter table is constructed.The forward and inverse kinematics control equations of the manipulator model are derived by analytical method,and the underlying control of the model is completed.In the joint space and Cartesian space,the trajectory planning governing equation is derived by the fifth-order polynomial interpolation algorithm,and the trajectory planning simulation experiment is carried out on the manipulator model.4.In order to verify the research results of this paper,it is integrated on the intelligent plastering robot platform,the hardware and software components and design of the robot are planned,the relevant workflow and software development are improved,and the test is carried out under the actual working conditions.It proves the meaning and value of the previous research.