Research On Robot Mapping And Navigation And Positioning Technology Based On Multi-sensor Fusion

With the rapid development of robot technology,the accuracy of state estimation for robot mapping,navigation and positioning is more and more demanding.However,the information provided by a single sensor can’t meet the requirements of high precision mapping and navigation of the robot.Therefore,it has important research significance and practical value to improve the mapping and navigation positioning accuracy based on the multi-sensor fusion method.In Shaanxi Province Key Chain "inspection robot based on VR tunnel is the key technology research"(2019ZDLGY03-01)project as the backing,the robot based on multisensor fusion precision built figure with navigation and positioning technology research,the main research work includes: in order to ensure the accuracy of the sensor output data and more consistency,the sensor based on the analysis of the mathematical model,completed the cameras and radar calibration and registration,implement the space and time synchronization between the multi-sensor synchronization.Aiming at the limitation of robot pose estimation due to GNSS failure in tunnels and other environments,a visual inertial navigation fusion algorithm based on nonlinear optimization was designed to improve the accuracy of mobile robot pose estimation when GNSS failure occurred.Aiming at the failure of visual scale information in the visual inertial navigation fusion system when the robot is running at a constant speed,a fusion algorithm of visual inertial navigation and lidar is proposed to realize real-time correction of visual scale information by lidar observation data.In order to verify the effectiveness of the proposed multi-sensor fusion algorithm in engineering applications,a SLAM system based on point cloud map was designed.Finally,the KITTI data set was used to test and verify in two different environments,and the positioning accuracy of the high-precision mapping and navigation and positioning system designed in this paper was compared with the current mainstream LEGO-LOAM system.The experimental results show that the designed mapping and navigation and positioning system based on the fusion of visual inertial navigation and lidar has smaller relative pose errors and absolute pose errors,and the constructed point cloud images are clearer and the positioning accuracy is higher,which is more suitable for robot navigation.The designed SLAM system has good real-time performance and robustness.