Synchronous Modeling And Task Planning Of Platform Trimming Robot

With the rapid development of science and technology,the application and popularization of robot automation technology in the field of sports is increasing gradually.In this thesis,a kind of mobile robot is studied,which can realize local autonomous navigation and adapt to high precision and high efficiency trimming unstructured platform surface in ski training field.The purpose of this thesis is to explore a new method of quick trimming ski jump in ski training ground,solve the problems of low modeling accuracy and poor trimming accuracy of ski jump,and realize the autonomous navigation of ski jump dressing robot.The specific research content of this thesis is as follows:(1)Three-dimensional modeling under unknown environment is studied,and an improved three-dimensional point cloud segmentation method is proposed for ski jump surface modeling.It includes filtering,downsampling,point cloud registration,point cloud segmentation,normal vector estimation,point cloud smoothing and three-dimensional modeling for the original point cloud data collected.In this process,the improved 3D point cloud segmentation method is used to carry out point cloud segmentation,and the superiority of the new method compared with the traditional method is verified by simulation and practical experiment.(2)The SLAM method based on Lidar is studied to solve the local autonomous navigation problem of the jumping trimming robot on the ski training ground.A multi-sensor fusion positioning method was proposed.The attitude calculation information of IMU of inertial navigation system was introduced into ESKF algorithm as state prediction,and the attitude matching between Lidar and constructed point cloud map was used as observation data to update the robot position and attitude information.In the process of the experiment,the A-LOAM algorithm,Le GO-LOAM algorithm and Le GO-LOAM algorithm introduced IMU three algorithms to simulate,construct the three-dimensional point cloud map with KITTI data set,and compare and analyze the generated trajectory and the truth value.Secondly,the ALOAM algorithm,Le GO-LOAM algorithm and Le GO-LOAM IMU algorithm are tested and compared,and the reliability of multi-sensor fusion localization is confirmed by experiments.(3)The platform of the skip trimming robot was built,the hardware system and software system of the robot were designed,and the actual field experiments were carried out.The overall design scheme of software and hardware system is proposed,and the hardware control equipment such as sensor,controller,motor and driver is selected.The software control system with Ubuntu system as the core is adopted to complete the control and communication between the upper computer user layer,the perception layer and the robot ontology execution layer.The robot platform was used to perform skip trimming experiments,which verified the effectiveness of the improved 3D point cloud segmentation method proposed in this thesis.Good results have been obtained in surface modeling,surface trimming and robot autonomous navigation.